CN114202772B

CN114202772B - Reference information generation system and method based on artificial intelligence and intelligent medical treatment

Info

Publication number: CN114202772B
Application number: CN202111484175.0A
Authority: CN
Inventors: 许炳生
Original assignee: HUNAN CHANGXIN CHANGZHONG TECHNOLOGY CO LTD
Current assignee: HUNAN CHANGXIN CHANGZHONG TECHNOLOGY CO LTD
Priority date: 2021-12-07
Filing date: 2021-12-07
Publication date: 2022-08-09
Anticipated expiration: 2041-12-07
Also published as: CN114202772A

Abstract

The disclosed embodiment provides a reference information generation method and system based on artificial intelligence and intelligent medical treatment, which determines gait feature information of a medical user according to posture measurement data of each key part, establishes gait image corresponding data between the gait feature information of the medical user and corresponding posture image data, inputs the gait image corresponding data into an artificial intelligence network model which is trained in advance, obtains a user obstacle classification attribute set corresponding to the gait image corresponding data, generates corresponding reference prompt information according to the user obstacle classification attribute set corresponding to the gait image corresponding data, thereby carrying out abnormal state early warning on the medical user, further considers the posture image features while considering the gait features, improves the prediction accuracy of the user obstacle classification attribute set, and further improves the prediction accuracy and timeliness of the abnormal state early warning, the extra measurement workload and the equipment detection cost of the subsequent related medical personnel are reduced.

Description

Reference information generation system and method based on artificial intelligence and intelligent medical treatment

Technical Field

The present disclosure relates to the field of artificial intelligence technologies, and in particular, to a reference information generation method and system based on artificial intelligence and intelligent medical care.

Background

Most of the intelligent hardware on the market at present is not very intelligent, for example, a competitive intelligent sphygmomanometer is not a traditional electronic sphygmomanometer with a function of connecting a mobile terminal, and perhaps a Bluetooth electronic sphygmomanometer is more suitable. The intelligent blood glucose meter is not different from the intelligent blood pressure meter, the medical equipment only added with the data analysis function is far from being intelligent, and the health advice simply generated logically is far from reaching the medical height. As for the remote strategy blood pressure, the blood pressure data is transmitted to the cloud platform. However, many intelligent hardware manufacturers strive to make their products more intelligent, for example, by monitoring the heart rate, pulse, blood glucose and blood pressure of a user and combining the historical data of the user, the monitoring, prediction and early warning of the user state are obtained.

Artificial intelligence based on artificial intelligence has been able to meet these needs from some perspectives, but it is only how to let these functions go to the ground and be practical that the intelligent medical service provider needs to pay attention. For example, for some users with cognitive impairment, dyskinesia is mainly used as a main manifestation, so the conventional scheme usually analyzes the gait abnormal condition of the user by using the gait characteristics of the user as a basic analysis characteristic, and further provides an early warning basis for the state of the user disorder. However, a lot of noise errors still exist in the analysis mode, so that the prediction result is inaccurate, the early warning prompt is not timely, and the additional measurement workload and the equipment detection cost of subsequent related medical care personnel are increased.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, the present disclosure is directed to a method and system for generating reference information based on artificial intelligence and smart medical treatment.

In a first aspect, the present disclosure provides a reference information generating method based on artificial intelligence and smart medical treatment, which is applied to a cloud computing service system, where the cloud computing service system is in communication connection with a plurality of medical monitoring terminal devices, and the method includes:

acquiring attitude measurement data of each key part of a medical user and attitude image data corresponding to the attitude measurement data, which are recorded by the medical monitoring terminal equipment in a preset recording period;

determining gait feature information of the medical user according to the posture measurement data of each key part, and establishing gait image corresponding data between the gait feature information of the medical user and the corresponding posture image data, wherein the gait image corresponding data is used for establishing corresponding data between each gait feature information unit in the gait feature information and a posture image frame corresponding to the corresponding posture image data;

inputting the gait image corresponding data into an artificial intelligence network model which is trained in advance to obtain a user obstacle classification attribute set corresponding to the gait image corresponding data;

and generating corresponding reference prompt information according to the user obstacle classification attribute set corresponding to the gait image corresponding data, wherein the reference prompt information is used for carrying out abnormal state early warning on the medical user.

In a second aspect, an embodiment of the present disclosure further provides a reference information generating system based on artificial intelligence and smart medical treatment, where the reference information generating system based on artificial intelligence and smart medical treatment includes a cloud computing service system and a plurality of medical monitoring terminal devices in communication connection with the cloud computing service system;

the cloud computing service system is used for:

According to any one of the above aspects, in the embodiment provided by the present disclosure, by acquiring the posture measurement data of each key part of the medical user and the posture image data corresponding to the posture measurement data, which are recorded by the medical monitoring terminal device in the preset recording period, determining the gait feature information of the medical user according to the posture measurement data of each key part, establishing the gait image corresponding data between the gait feature information of the medical user and the corresponding posture image data, inputting the gait image corresponding data into the artificial intelligence network model trained in advance, obtaining the user obstacle classification attribute set corresponding to the gait image corresponding data, generating the corresponding reference prompt information according to the user obstacle classification attribute set corresponding to the gait image corresponding data, thereby performing the abnormal state early warning on the medical user, and further considering the posture image characteristics while considering the gait features, the prediction accuracy of the user obstacle classification attribute set is improved, the prediction accuracy and timeliness of abnormal state early warning are further improved, and extra measurement workload and equipment detection cost of subsequent related medical personnel are reduced.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings that need to be called in the embodiments are briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present disclosure, and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

Fig. 1 is a schematic view of an application scenario of a reference information generation system based on artificial intelligence and smart medical care according to an embodiment of the present disclosure;

fig. 2 is a schematic flowchart of a reference information generation method based on artificial intelligence and smart medical care according to an embodiment of the disclosure.

Detailed Description

Fig. 1 is a schematic view of a reference information generating system 10 based on artificial intelligence and smart medical treatment according to an embodiment of the present disclosure. The reference information generating system 10 based on artificial intelligence and smart medical treatment may include a cloud computing service system 100 and a medical monitoring terminal device 200 communicatively connected to the cloud computing service system 100. The artificial intelligence and smart medical based reference information generating system 10 shown in fig. 1 is only one possible example, and in other possible embodiments, the artificial intelligence and smart medical based reference information generating system 10 may also include only at least some of the components shown in fig. 1 or may also include other components.

In this embodiment, the cloud computing service system 100 and the medical monitoring terminal device 200 in the reference information generating system 10 based on artificial intelligence and smart medical treatment may cooperatively perform the reference information generating method based on artificial intelligence and smart medical treatment described in the following method embodiments, and the following detailed description of the method embodiments may be referred to for the specific steps performed by the cloud computing service system 100 and the medical monitoring terminal device 200.

Referring to the technical problems of the aforementioned background art, dementia, which is a syndrome characterized by a decline in cognitive functions and can affect two or more cognitive fields such as learning and memory, language, executive functions, complex attention, sensorimotor functions and social cognitive functions, is taken as an example of a subject. Elderly people with combined cognitive dysfunction (including dementia and mild cognitive impairment, especially dementia) seriously affect their quality of life and can cause a series of adverse events, creating a heavy burden on the home and society. With the rapid increase of the aging trend of the population structure, the nursing burden of the mentally disabled elderly is particularly heavy in the predicament of not getting rich and old.

The most common type of dementia in the elderly is Alzheimer's Disease (AD). At present, it is generally accepted by the academia that AD is not an inevitable consequence of nervous system degeneration, although previous studies have demonstrated a significant correlation between cognitive dysfunction and β -amyloid (a β) deposition and tau protein degeneration; however, the pathogenesis of AD is still not well understood. At present, the benefits of anti-dementia drug intervention in late stage AD are limited, and the best therapeutic strategy is still multi-dimensional intervention in patients at the early stage of the disease, i.e. the Mild Cognitive Impairment (MCI) stage.

Tools currently used clinically to screen for Cognitive function include the Clinical Dementia rating Scale (CDR), the standardized mental state examination (MMSE), Montreal Cognitive Assessment (MoCA), Alzheimer's Disease Assessment Scale-Cognitive part (ADAS-cog), and the like. The staff who has received professional training needs to evaluate the scale, about 10 minutes are needed at the fastest speed, about 45 minutes are needed at the slowest speed (9, 10), the whole evaluation process takes longer time, and the evaluation result of the scale is influenced by non-cognition ability factors such as vision, hearing and culture degree of a patient. In the current vast basic community medical treatment and chronic disease management practices, the cognitive function assessment coverage is still narrow and low in accuracy due to the reasons that many old people are poor in activity ability, limited in economic capability, poor in compliance, poor in health concept, not strong in hospitalization willingness and the like although the demand for cognitive function assessment is large. Therefore, a more convenient, effective, economic and objective biological measurement method is established to identify the high risk group with the senile combined cognitive function impairment and dementia, and the contradiction between the shortage of basic medical resources and the incapability of meeting the cognitive function evaluation requirement caused by the rapid increase of the population base of the elderly can be solved to a great extent.

The inventor of the application finds that walking is a complex movement process, and gait can intuitively reflect the movement function and health condition of a human body. Walking is considered in the related art to be a complex and delicate activity requiring integrated executive functions, attention, and judgment of internal and external environments. Walking speed is more closely related to the cardiac function, hospitalization rate, risk of death, and functional status of the elderly. Previous studies have shown that elderly people with neurodegenerative diseases often have a combination of cognitive dysfunction and gait abnormalities, which are strongly correlated (13, 14). A meta-analysis report: MCI patients had significant changes in walking speed (d = -0.74, p < 0.01), stride (d = -0.65, p < 0.01), stride time (mean: d = 0.56, p = 0.02, coefficient of variation: d = 0.50, p < 0.01) compared to healthy elderly (15). Another study also showed a correlation between the amount of a β deposition and walking speed (regression coefficients (β) = -0.086, p =0.005) (16). The users often use the 'teeter walking' to describe the gait of infants and the old, and people with certain life experience can often distinguish the identity and the age through the gait and the posture.

From the aspect of informatics theory, human beings have thinking activities of recognizing things, and machines can be abstracted and simulated by a certain program and realize similar functions. At present, most gait analysis instruments commonly used in the related art need to be operated by trained professionals in a professional three-dimensional gait laboratory, a large number of reflective balls and myoelectric patches need to be worn, the operation difficulty is high, the acceptance of patients is low, and the popularization difficulty in basic medical treatment is high. In recent years, with the rapid improvement of the data processing capability of a computer, the development of an artificial intelligence technology, the popularization of equipment such as visible light and infrared camera shooting, a dynamic capture technology, a pressure sensing device, a microwave radar, laser positioning, three-dimensional body motion sensing and the like, the methodology of the field is rapidly developed, wearable equipment is no longer a necessity for gait recognition, and the wearable equipment is no longer a barrier for recognition due to different visual angles, different clothes, different environments and the like, so that the application range is gradually widened. The application examples comprise identity recognition of suspects through gait posture, public security monitoring unattended alarm, falling alarm of solitary old people and the like, wherein the specificity and sensitivity of a falling alarm system without any wearing equipment reach 99%, the success rate of judgment of the debugged machine on human sex through gait recognition reaches 80-100%, and the gait parameter identity recognition rate under a real outdoor scene reaches more than 91% (18, 22). The state and the degree of certain specific diseases can be distinguished by combining gait analysis; in the field, antecedents such as Tan and the like use somatosensory equipment to dynamically evaluate gait data of patients with Parkinson's disease and paraplegia after stroke, and find 1 index with the sensitivity degree exceeding the pace speed.

With the field of image recognition, artificial intelligence and health condition judgment by applying gait parameters, the technology accumulation reaches a certain degree, and the application range is continuously widened. By utilizing the technical progress in the field, gait characteristics with stronger correlation with cognitive function damage are found, and the research in the field can be inspired; under the current situation that the quantity and the quality of medical resources are not uniformly distributed, but a communication network is developed, if a convenient, objective, cheap and accurate intelligent mobile terminal device is further constructed for identifying a person with cognitive dysfunction or even a potential high-risk patient, the intelligent mobile terminal device has popularization in primary medical treatment; furthermore, the early warning or preliminary screening diagnosis can be made for the cognitive dysfunction of the old, the prognosis of the old can be improved through a proper intervention technology, and the early warning or preliminary screening diagnosis method has positive significance for maintaining the body function of the old and reducing the family and social care pressure brought by aging.

In order to solve the technical problem in the foregoing background, fig. 2 is a flowchart illustrating a reference information generating method based on artificial intelligence and smart medical treatment according to an embodiment of the present disclosure, which may be executed by the cloud computing service system 100 shown in fig. 1, and the reference information generating method based on artificial intelligence and smart medical treatment is described in detail below.

Step S110, acquiring the attitude measurement data of each key part of the medical user and the attitude image data corresponding to the attitude measurement data, which are recorded by the medical monitoring terminal device in a preset recording period.

In this embodiment, the preset recording period may be flexibly set according to the actual pre-diagnosis condition of the medical user. The posture measurement data of each key part of the medical user and the posture image data corresponding to the posture measurement data may be collected in advance, or may be collected in real time, or a part of the posture measurement data may be collected in advance, and a part of the posture measurement data may be collected in real time. The critical areas may include a plurality of predetermined joint areas of the lower limb of the medical user, and may include, for example, a waist joint (pelvic area) and hip, knee, and ankle joints on the left and right leg sides. In addition, the attitude image data can be acquired in real time by an image acquisition device in the external environment, thereby further combining the attitude image data on the basis of the attitude measurement data.

Step S120, determining the gait feature information of the medical user according to the posture measurement data of each key part, and establishing gait image corresponding data between the gait feature information of the medical user and the corresponding posture image data.

In this embodiment, the gait image corresponding data may be used to establish corresponding data between each gait feature information unit in the gait feature information and a pose image frame corresponding to the corresponding pose image data. For example, the gait feature information may include a plurality of gait feature information elements with a unit time segment (e.g., 0.5S) as a division node, and correspondingly, the posture image data may include a plurality of posture image frames with a unit time segment (e.g., 0.5S) as a division node. In this way, corresponding data between each gait feature information unit in the gait feature information and the corresponding posture image frame corresponding to the corresponding posture image data can be established, so that when artificial intelligence is performed subsequently, the gait features are considered, and meanwhile, the posture image features are further considered, and the accuracy of subsequent classification is improved.

Step S130, inputting the gait image corresponding data into a pre-trained artificial intelligence network model to obtain a user obstacle classification attribute set corresponding to the gait image corresponding data.

In this embodiment, the artificial intelligence network model may be obtained by training in advance based on a training sample set, for example, the training sample set may include gait images corresponding data samples and corresponding user obstacle classification attribute set samples, and then the initial artificial intelligence network model performs feature learning based on the gait images corresponding data samples and the corresponding user obstacle classification attribute set samples, so as to obtain an artificial intelligence network model that is obtained by training in advance.

The user obstacle classification attribute set may include one or more user obstacle classification attributes, which may be used to characterize a classification attribute condition of cognitive impairment of the medical user, and the classification attribute also includes a probability value distribution of each specific user obstacle classification.

Step S140, generating corresponding reference prompt information according to the user obstacle classification attribute set corresponding to the gait image corresponding data, wherein the reference prompt information is used for carrying out abnormal state early warning on the medical user.

Therefore, the corresponding reference prompt information can be generated according to the user obstacle classification attribute set corresponding to the gait image corresponding data, for example, when the probability value distribution of a certain user obstacle classification is abnormal, an abnormal state early warning is sent out to prompt a related medical user to perform subsequent rehabilitation training in a targeted manner.

Based on the steps, the gait feature information of the medical user is determined according to the posture measurement data of each key part of the medical user and the posture image data corresponding to the posture measurement data, which are recorded by the medical monitoring terminal device in the preset recording period, the gait image corresponding data between the gait feature information of the medical user and the corresponding posture image data are established, the gait image corresponding data are input into the artificial intelligent network model which is trained in advance, the user obstacle classification attribute set corresponding to the gait image corresponding data is obtained, the corresponding reference prompt information is generated according to the user obstacle classification attribute set corresponding to the gait image corresponding data, and therefore the abnormal state early warning is carried out on the medical user, and the posture image feature is further considered while the gait feature is considered, the prediction accuracy of the user obstacle classification attribute set is improved, and the prediction accuracy of the abnormal state early warning is further improved.

In one embodiment, on the basis of the above steps, the following steps can be further included.

And S150, acquiring rehabilitation training motion data after the targeted rehabilitation training is performed on the prompted abnormal early warning information.

And step S160, analyzing the rehabilitation training motion data based on the rehabilitation training label learning network trained in advance to obtain a rehabilitation training label set corresponding to the rehabilitation training motion data.

And S170, performing further reference information prompt on the medical user corresponding to the data corresponding to the gait image according to the rehabilitation training label set corresponding to the rehabilitation training motion data.

Based on the steps, abnormal early warning information prompting is carried out on the medical monitoring terminal equipment according to the user obstacle classification attribute set of the gait image corresponding data of the medical monitoring terminal equipment, rehabilitation training motion data after the pointed abnormal early warning information is subjected to targeted rehabilitation training is obtained, the rehabilitation training motion data is analyzed based on a rehabilitation training label learning network trained in advance, a rehabilitation training label set corresponding to the rehabilitation training motion data is obtained, and further reference information prompting is carried out on the medical user corresponding to the gait image corresponding data according to the rehabilitation training label set corresponding to the rehabilitation training motion data. Therefore, by analyzing the rehabilitation training motion condition of the user, the further reference information prompt of the medical user corresponding to the matched rehabilitation training label set can be determined.

For example, in one embodiment, the above method may further comprise: acquiring reference positive rehabilitation training motion data and reference negative rehabilitation training motion data when a reference rehabilitation training motion condition is generated, and performing feature extraction on the reference positive rehabilitation training motion data to obtain reference positive rehabilitation training motion features of the reference positive rehabilitation training motion data, wherein the reference positive rehabilitation training motion features are used for indicating positive feedback motion features of the reference rehabilitation training motion condition; performing feature extraction on the reference negative rehabilitation training motion data to obtain reference negative rehabilitation training motion data of the reference negative rehabilitation training motion data, wherein the reference negative rehabilitation training motion data is used for indicating negative feedback motion features of the reference negative rehabilitation training motion situation; taking the reference negative rehabilitation training motion data as opponent data of the reference positive rehabilitation training motion characteristics to generate training basic data of the rehabilitation training label learning network; and training the rehabilitation training label learning network by adopting the training basic data, wherein the rehabilitation training label learning network is used for analyzing based on the rehabilitation training motion data to obtain a rehabilitation training label set corresponding to the rehabilitation training motion data.

In one embodiment, for step S170, the following steps may be included:

step A110, a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device is obtained, the rehabilitation training push network comprises a target rehabilitation training strategy corresponding to the target rehabilitation training push application, a plurality of relational rehabilitation training strategies and a plurality of training synchronous information, the training synchronous information corresponds to a rehabilitation activity rule of a rehabilitation subscription object of the target rehabilitation training push application, and each relational rehabilitation training strategy is synchronously configured with the target rehabilitation training strategy through the corresponding training synchronous information.

In one embodiment, a target rehabilitation training strategy corresponding to a target rehabilitation training push application is a rehabilitation training strategy to be decided with an unknown key probability, and a classification probability evaluation is performed on a target rehabilitation training strategy with an unknown key probability.

The target rehabilitation training strategy and the relationship rehabilitation training strategy may be rehabilitation training strategies such as a balance function training strategy, a muscle strength training strategy, a muscle tension treatment training strategy, a hemiplegic limb function training strategy and the like, and the relationship rehabilitation training strategy may be a rehabilitation training strategy having a certain relation with the target rehabilitation training strategy. The training synchronization information may be training flow direction information providing a certain association between the target rehabilitation training strategy and the relationship rehabilitation training strategy, and is used to associate the target rehabilitation training strategy and the relationship rehabilitation training strategy. It is worth to be noted that the rehabilitation training strategy may represent different training recommendation schemes, and the rehabilitation training push network may represent a relationship between the target rehabilitation training strategy and the training recommendation schemes of other relationship rehabilitation training strategies. The rehabilitation training strategy can be determined according to rehabilitation activity rules applied by target rehabilitation training push, the rehabilitation activity rules can represent the characteristics of the rehabilitation training strategy, such as training related parts, training related actions, training related equipment, used training flow behaviors and the like of the rehabilitation training strategy, and the relationship between the target rehabilitation training strategy and the relationship rehabilitation training strategy can be established by utilizing the rehabilitation activity rules, so that a rehabilitation training push network is established.

For example, assuming that the target rehabilitation training Strategy strand 0 is associated with the rehabilitation training Strategy strand 1, the rehabilitation training Strategy strand 2 and the rehabilitation training Strategy strand 9 by training the site QA, or the target rehabilitation training Strategy strand 0, the rehabilitation training Strategy strand 1, the rehabilitation training Strategy strand 2 and the rehabilitation training Strategy strand 9 are associated with other rehabilitation training strategies by training the site QA, the target rehabilitation training Strategy strand 0 may be used as the target rehabilitation training Strategy, the rehabilitation training Strategy strand 1, the rehabilitation training Strategy strand 2 and the rehabilitation training Strategy strand 9 may be used as the relational rehabilitation training Strategy of the target rehabilitation training Strategy strand 0, correspondingly, the site QA involved may be used as the rehabilitation training Strategy strand 1, the rehabilitation training Strategy2, the strand 9, the target rehabilitation training Strategy strand 46595 and the rehabilitation training Strategy may be associated with the target rehabilitation training Strategy strand 5842, the target rehabilitation training Strategy strand 46 9, the training synchronization information corresponding to the target rehabilitation training strategy can be determined based on the rehabilitation activity rule of the target rehabilitation training strategy, illustratively, the rehabilitation activity rule associated with the target rehabilitation training strategy is a training-related part, correspondingly, the training synchronization information between the target rehabilitation training strategy and the target rehabilitation training strategy can be the training-related part, and the training-related part QA can be one training-related part or a plurality of training-related parts, and is not particularly limited.

Step A120, based on the rehabilitation training push network, obtaining a key probability of a key rehabilitation subscription object in rehabilitation subscription objects corresponding to all relationship rehabilitation training strategies associated with each training synchronous data, wherein the key rehabilitation subscription object is a rehabilitation subscription object with a prescription matching relationship when matching the relationship rehabilitation training parameter scheme.

In one embodiment, the key probability may be a ratio of the rehabilitation subscription object having the prescription matching relationship in the matching relationship rehabilitation training parameter scheme to all rehabilitation subscription objects under a certain rehabilitation activity rule corresponding to the rehabilitation subscription object, and is not particularly limited.

Illustratively, the rehabilitation training Strategy stratgy 7, the rehabilitation training Strategy stratgy 6 and the rehabilitation training Strategy stratgy 5 are all prescription-matched to a certain prescription matching process through the training related equipment QB, if the rehabilitation training Strategy stratgy 7 sends matching information of prescription matching to a certain prescription matching process through the related rehabilitation training consultation interface EP within a preset time sequence, the rehabilitation training Strategy stratgy 7 is regarded as a key rehabilitation subscription object, and the rehabilitation training Strategy stratgy 5 and the rehabilitation training Strategy stratgy 6 do not send matching information of prescription matching to a certain prescription matching process through the related rehabilitation training consultation interface EP within a preset time sequence, the rehabilitation training Strategy stratgy 5 and the rehabilitation training Strategy stratgy 6 are regarded as non-key rehabilitation subscription objects. Correspondingly, the total rehabilitation training strategy evaluation index participating in the prescription matching is 3, wherein the key rehabilitation subscription object is 1, and the key probability of the key rehabilitation subscription object is 1/3.

Step a130, obtaining a historical balance evaluation index corresponding to each training synchronization information, and obtaining a reference information prompt index corresponding to the target rehabilitation training push application based on the key probability corresponding to each training synchronization information and the historical balance evaluation index corresponding to each training synchronization information, where the reference information prompt index is used to prompt the user corresponding to the target rehabilitation training push application with the relational rehabilitation training parameter scheme.

In an embodiment, because the relationship rehabilitation training strategy under each piece of training synchronization information may have a relationship rehabilitation training strategy corresponding to a relationship rehabilitation training strategy that has not participated in the relationship rehabilitation training parameter scheme, for example, has not participated in prescription matching, or may have participated in a relationship rehabilitation training strategy corresponding to a relationship rehabilitation training parameter scheme, but has not reached the requirement of the relationship rehabilitation training parameter scheme, for example, has participated in prescription matching, and has not reached a deadline, the above-determined key probabilities corresponding to each piece of training synchronization information may be greatly influenced by the relationship rehabilitation training strategies. Therefore, after obtaining the key probability of the key rehabilitation subscription object in the rehabilitation subscription objects corresponding to all the relationship rehabilitation training strategies associated with each training synchronization data, it is further required to obtain the historical balance evaluation index corresponding to each training synchronization information, where the size of the historical balance evaluation index may reflect the confidence of the training synchronization information in the key probability of the predicted target rehabilitation training strategy, that is, the larger the historical balance evaluation index is, the higher the confidence of the training synchronization information in the degree of the predicted target rehabilitation training strategy is, and the more accurate the key probability of the finally predicted target rehabilitation training strategy is.

In an embodiment, the historical balance evaluation index corresponding to each training synchronization information may be determined according to a training flow direction information evaluation index of a training synchronization information that completes a relational rehabilitation training parameter scheme in a relational rehabilitation training strategy corresponding to each training synchronization information, for example, according to a training flow direction information evaluation index of a training synchronization information that completes at least one prescription matching. It should be noted that the training flow direction information evaluation index of the completed related rehabilitation training parameter scheme indicates that whether the rehabilitation training strategy has an evaluation index of the rehabilitation training strategy of the flow direction relationship in the related rehabilitation training strategy corresponding to each piece of training synchronization information, so that the larger the evaluation index is, the higher the confidence of the training synchronization information is, and the historical balance evaluation index of each piece of training synchronization information can be calculated accordingly.

In one embodiment, since the reference information prompting index degree of the target rehabilitation training strategy is judged to be related to the key probability and the history balance evaluation index of each training synchronous information, the key probability and the history balance evaluation index of each training synchronous information can be obtained first, then the reference information prompting index corresponding to the target rehabilitation training push application is calculated based on the obtained key probability and the corresponding history balance evaluation index of each training synchronous information, and further, the corresponding user prompting relation rehabilitation training parameter scheme can be applied to the target rehabilitation training push application according to the reference information prompting index.

In one embodiment, a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device is obtained, a key probability of a key rehabilitation subscription object in rehabilitation subscription objects corresponding to all relation rehabilitation training strategies associated with each training synchronization data is obtained based on the rehabilitation training push network, a history balance evaluation index corresponding to each training synchronization information is obtained, a reference information prompt index corresponding to the target rehabilitation training push application is obtained based on the key probability corresponding to each training synchronization information and the history balance evaluation index corresponding to each training synchronization information, and the relation rehabilitation training parameter scheme is prompted for a user corresponding to the target rehabilitation training push application according to the reference information prompt index. Therefore, the classification probability assessment can be quantized, the basis is improved for determining the relational rehabilitation training parameter scheme, the classification probability assessment is carried out based on the key probabilities corresponding to various training synchronous information related to the target rehabilitation training pushing application, the accuracy of the classification probability assessment is improved, in addition, when the reference information prompt indexes are predicted, historical balance evaluation indexes corresponding to all the training synchronous information are obtained and used as references for the classification probability assessment of all the training synchronous information, the accuracy of the classification probability assessment is further improved, and whether the relational rehabilitation training parameter scheme is pushed for a user or not can be accurately determined.

The following describes a flow of a reference information generation method based on artificial intelligence and smart medicine, which may include the following steps:

step A210, obtaining a rehabilitation activity rule of a rehabilitation subscription object applied by the target rehabilitation training push and a rehabilitation activity rule of a corresponding relationship rehabilitation training strategy applied by the target rehabilitation training push, wherein the rehabilitation activity rule is used for representing rehabilitation training strategy characteristics of the rehabilitation training strategy.

In one embodiment, the rehabilitation activity rule may characterize the rehabilitation training strategy characteristics, for example, the training-related part, the training-related action, the training-related equipment, the used training flow behavior, and the like of the rehabilitation training strategy, and the relationship rehabilitation training strategy associated with the target rehabilitation training strategy may be determined by using the rehabilitation activity rule, and accordingly, the relationship between the target rehabilitation training strategy and the relationship rehabilitation training strategy may be established.

Step A220, generating a rehabilitation training push network corresponding to the target rehabilitation training push application based on the rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the rehabilitation activity rule of the corresponding relationship rehabilitation training strategy of the target rehabilitation training push application, taking the rehabilitation subscription object of the target rehabilitation training strategy as the target rehabilitation training strategy and taking the relationship rehabilitation training strategy as the relationship rehabilitation training strategy.

In one embodiment, after the rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the rehabilitation activity rule of the corresponding relational rehabilitation training strategy of the target rehabilitation training push application are obtained, the rehabilitation activity rule is used as an actual application activity rule in classification probability evaluation, and further, each actual application activity rule is used as training synchronization information, the rehabilitation subscription object of the target rehabilitation training strategy is used as the target rehabilitation training strategy, and the relational rehabilitation training strategy is used as the relational rehabilitation training strategy, so that a rehabilitation training push network corresponding to the target rehabilitation training push application is generated.

In practical applications, considering that some rehabilitation activity rules are not suitable for being used as practical application activity rules, for example, rehabilitation activity rules similar to part of experimental features, most rehabilitation training strategies may be similar to rehabilitation activity rules similar to part of experimental features, and therefore if the rehabilitation activity rules of this type are introduced as practical application activity rules, classification errors may be introduced, that is, inaccurate finally predicted reference information prompt index values may be caused.

For the above practical situation, before generating the rehabilitation training push network corresponding to the target rehabilitation training push application, rehabilitation activity rules similar to part of experimental features may be removed in some ways.

Firstly, the rehabilitation activity rule of the rehabilitation subscription object applied by the target rehabilitation training push is compared with the relevance of a preset rehabilitation activity rule, wherein the preset rehabilitation activity rule corresponds to the relational rehabilitation training parameter scheme.

And secondly, acquiring a rehabilitation activity rule matched with the preset rehabilitation activity rule from the rehabilitation activity rules of the rehabilitation subscription objects pushed and applied by the target rehabilitation training according to the correlation comparison result, and taking the rehabilitation activity rule as a target rehabilitation activity rule.

In this embodiment, preset rehabilitation activity rules corresponding to the relational rehabilitation training parameter scheme may be stored in advance, where the preset rehabilitation activity rules may be a training-related part 1, a training-related part 2, a training-related part 3, a relational rehabilitation training advisory interface 1, a relational rehabilitation training advisory interface 2, a training-related action 1, a training-related action 2, a training flow action 1, and a training flow action 2, the preset rehabilitation activity rules may be multiple, and the number of the preset rehabilitation activity rules is not limited here. Further, the rehabilitation activity rule of the rehabilitation subscription object applied by the target rehabilitation training push can be compared with the preset rehabilitation activity rule in the degree of correlation, and the rehabilitation activity rule matched with the preset rehabilitation activity rule is used as the target rehabilitation activity rule. Illustratively, if the rehabilitation activity rules of the rehabilitation subscription object of the target rehabilitation training push application include a training-related part 1, a training-related part 2, a training-related part 3, a training-related part 4, a relational rehabilitation training consultation interface 1, a relational rehabilitation training consultation interface 2, a training-related action 1, a training-related action 2, a training-flow action 1, and a training-flow action 2, the preset rehabilitation activity rules are the training-related part 1, the training-related part 2, the training-related part 3, the relational rehabilitation training consultation interface 1, the relational rehabilitation training consultation interface 2, the training-related action 1, the training-related action 2, the training-flow action 1, and the training-flow action 2, further, the target rehabilitation activity rules can be obtained as the training-related part 1, the training-related part 2, the training-related part 3, the relational rehabilitation training consultation interface 1, the training-related action 2, the training-flow action 2, and the training-flow action 2, The method comprises a relational rehabilitation training consultation interface 2, a training related action 1, a training related action 2, a training flow action 1 and a training flow action 2, wherein a training related part 4 cannot find a matched label in preset rehabilitation activity rules, the rehabilitation activity rule of the training related part 4 is removed, it can be understood that the rehabilitation activity rule of the training related part 4 is not suitable for being placed in an actual application activity rule, the training related part 4 may be a whole training related part, namely, an error rule is unlikely to be arranged behind the rehabilitation activity rule of the training related part 4, and if the rehabilitation activity rule of the training related part 4 is used as a target rehabilitation activity rule, an error may be introduced into a classification process to influence a final classification result. It can be understood that the main purpose of screening the target rehabilitation activity rule is to eliminate the rehabilitation activity rule which is not suitable for being used as the actual application activity rule, and improve the accuracy of classification probability evaluation.

And finally, generating a rehabilitation training push network corresponding to the target rehabilitation training push application based on the target rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the target rehabilitation activity rule of the corresponding relation rehabilitation training strategy of the target rehabilitation training push application, taking the rehabilitation subscription object of the target rehabilitation training strategy as the target rehabilitation training strategy and taking the relation rehabilitation training strategy as the relation rehabilitation training strategy.

In one embodiment, the rehabilitation activity rules which are not suitable for being used as the actual application activity rules are removed by comparing the rehabilitation activity rules of the rehabilitation subscription objects of the target rehabilitation training push application with preset rehabilitation activity rules in a correlation degree mode, the rehabilitation activity rules matched with the preset rehabilitation activity rules are used as the target rehabilitation activity rules, the target rehabilitation activity rules are used as the actual application activity rules in classification probability evaluation, the rehabilitation subscription objects of the target rehabilitation training strategies are used as the target rehabilitation training strategies based on each actual application activity rule after removal, and the rehabilitation training push networks corresponding to the target rehabilitation training push application are generated by using the relational rehabilitation training strategies as the relational rehabilitation training strategies.

Step a230, obtaining an evaluation index of a designated rehabilitation training strategy in all the relationship rehabilitation training strategies associated with each piece of training synchronization data, as a first evaluation index, where the designated rehabilitation training strategy is a relationship rehabilitation training strategy corresponding to a certified rehabilitation training strategy of the relationship rehabilitation training parameter scheme.

In an embodiment, obtaining the historical balance evaluation index corresponding to each training synchronization information may first obtain an evaluation index of a designated rehabilitation training strategy in all the relationship rehabilitation training strategies associated with each training synchronization data, where the designated rehabilitation training strategy is a relationship rehabilitation training strategy corresponding to a rehabilitation training strategy of the authenticated relationship rehabilitation training parameter scheme.

Step a240, obtaining an inverse number corresponding to a ratio of the preset index value to each first evaluation index, and obtaining a target evaluation index corresponding to each training synchronization information.

Step a250, taking the target evaluation index corresponding to each piece of training synchronization information as an independent variable, obtaining a function value of a preset target function, and obtaining a historical balance evaluation index corresponding to each piece of training synchronization information, where a base number value of the preset target function is greater than 1.

In one embodiment, based on the obtained first evaluation index and the preset index value of the designated rehabilitation training strategy, a target evaluation index corresponding to each training synchronization information may be obtained, where the target evaluation index may be a negative number, and for example, if the obtained first evaluation index is 4 and the preset index value is 2, the target evaluation index may be-1/2, and since there are multiple types of training synchronization information, the target evaluation index corresponding to each type of training synchronization information needs to be obtained.

Since the target evaluation indexes corresponding to each training synchronization information may be the same and may be different, the target evaluation index may be regarded as an argument, and a function value of a preset target function is obtained based on the argument, where a base value of the target function is greater than 1, and for example, e is taken as the base of the target function, so that the corresponding historical trade-off evaluation index in the above example may be obtained as e-1/2 b. If the preset index value x is expressed, the first evaluation index is expressed by confidence, correspondingly, a formula for calculating the historical balance evaluation index corresponding to the training synchronization information can be expressed as e-x/confidence, and the historical balance evaluation index corresponding to each training synchronization information can be calculated through the formula.

In another embodiment, a ratio of a preset index value to each of the first evaluation indexes may be further obtained, a target evaluation index corresponding to each of the training synchronization information is obtained, the target evaluation index corresponding to each of the training synchronization information is used as an argument, a function value of a preset objective function is obtained, and a historical balance evaluation index corresponding to each of the training synchronization information is obtained, a base number of the preset objective function is less than 1, in this case, the base number of the objective function may be 1/e, correspondingly, a formula for calculating a confidence level may be (1/e) x/confidence, and similarly, the historical balance evaluation index corresponding to each of the training synchronization information may also be calculated by the formula.

In both embodiments, the size of the historical trade-off evaluation index is between 0 and 1, and in one embodiment, the size of the historical trade-off evaluation index is not limited herein.

In another embodiment, the ratio of each obtained first evaluation index to a preset index value may be directly used as a historical trade-off evaluation index, in which case, the historical trade-off evaluation index may be greater than 1. Illustratively, if the obtained first evaluation index is 6, an index value 4 is preset, and correspondingly, the historical balance evaluation index is 1.5.

Step A260, obtaining a reference information prompt index corresponding to the target rehabilitation training push application based on the key probability corresponding to each training synchronous information and the historical balance evaluation index corresponding to each training synchronous information, wherein the reference information prompt index is used for prompting the relational rehabilitation training parameter scheme for the user corresponding to the target rehabilitation training push application.

In one embodiment, step a260 may comprise:

step a261, obtaining a reference information prompt index corresponding to each piece of training synchronization information based on the key probability corresponding to each piece of training synchronization information and the historical balance evaluation index corresponding to each piece of training synchronization information.

In one embodiment, there are multiple types of training synchronization information, for example, four types of training synchronization information, i.e., a training related part QA, a training flow behavior 32, a relational rehabilitation training consultation interface QF, and a training related action 34, where the same type of training synchronization information may include only 1 piece of training synchronization information or multiple different pieces of training synchronization information, for example, the training synchronization information relational rehabilitation training consultation interface QF may include only the relational rehabilitation training consultation interface 1 or multiple different pieces of training synchronization information relational rehabilitation training consultation interface 1, relational rehabilitation training consultation interface 2, relational rehabilitation training consultation interface 3, and relational rehabilitation training consultation interface 4, and therefore, for different situations, the reference information prompt index values corresponding to each type of training synchronization information are obtained differently.

In an embodiment, only one training synchronization information exists in each training synchronization information, the obtained key probability of each training synchronization information is the key probability of the corresponding type of training synchronization information, and then the reference information prompt index corresponding to each type of training synchronization information is further obtained based on the key probability of each type of training synchronization information. Illustratively, if there are two types of training synchronization information, there are training flow behavior 32 and the relational rehabilitation training advisory interface QF, respectively. The training flow behavior 32 only includes one training synchronization information training flow behavior 1, the relational rehabilitation training consultation interface QF only includes one training synchronization information relational rehabilitation training consultation interface 1, correspondingly, the obtained key probability of the training flow behavior 1 is the key probability corresponding to the type of the training flow behavior 32, and the obtained key probability of the relational rehabilitation training consultation interface 1 is the key probability of the type of the relational rehabilitation training consultation interface QF.

In another embodiment, if multiple different training synchronization information exists in each category of training synchronization information, the obtained key probabilities of the multiple different training synchronization information in each category need to be summed to obtain the key probability corresponding to each category of training synchronization information. Illustratively, if there are two types of training synchronization information, there are training flow behavior 32 and the relational rehabilitation training advisory interface QF, respectively. The training flow behaviors 32 respectively include a training flow behavior 1 and a training flow behavior 2, the relational rehabilitation training consultation interface QF respectively includes a training synchronization information relational rehabilitation training consultation interface 1 and a relational rehabilitation training consultation interface 2, correspondingly, the sum of the key probability of the training flow behavior 1 and the key probability of the training flow behavior 2 is the key probability of the training synchronization information of the type of the training flow behavior 32, and the sum of the key probability of the relational rehabilitation training consultation interface 1 and the key probability of the relational rehabilitation training consultation interface 2 is the key probability of the training synchronization information of the type of the relational rehabilitation training consultation interface QF.

Step a262, obtaining a reference information prompt index corresponding to the target rehabilitation training push application based on the influence index corresponding to each training synchronization information and the reference information prompt index corresponding to each training synchronization information.

In one embodiment, when a plurality of different training synchronization information exists in each type of training synchronization information, after a reference information prompt index value corresponding to each type of training synchronization information is obtained, a reference information prompt index corresponding to a target rehabilitation training push application is calculated and obtained by combining an influence index corresponding to each type of training synchronization information.

In one embodiment, when only one training synchronization information exists in each type of training synchronization information, after the reference information prompt index value corresponding to each type of training synchronization information is obtained, the reference information prompt index corresponding to the target rehabilitation training push application is calculated and obtained by combining the influence index corresponding to each training synchronization information.

Step A270, judging whether the reference information prompt index corresponding to the target rehabilitation training push application is larger than a preset reference information prompt index threshold value.

In an embodiment, after obtaining a reference information prompt index value corresponding to a target rehabilitation training push application, the reference information prompt index value may be further determined, and whether the reference information prompt index value is greater than a preset reference information prompt index threshold value or not is determined, where a user corresponding to the target rehabilitation training push application greater than the preset reference information prompt index threshold value is a matching user, and a user corresponding to the target rehabilitation training push application not greater than the preset reference information prompt index threshold value is a non-matching user.

Step A280, if yes, prompting the relation rehabilitation training parameter scheme for the user corresponding to the target rehabilitation training push application.

If the reference information prompt index value corresponding to the target rehabilitation training push application is larger than the preset reference information prompt index threshold value, determining that the user corresponding to the target rehabilitation training push application is a matching user, namely, the probability that the user can match the relationship rehabilitation training parameter scheme in the current training rehabilitation stage is higher, and further, prompting the relationship rehabilitation training parameter scheme for the user corresponding to the target rehabilitation training push application. Exemplarily, if the reference information prompt index corresponding to the obtained target rehabilitation training push application is 0.6 and the preset reference information prompt index threshold is 0.5, it may be determined that the user corresponding to the target rehabilitation training push application is a matching user, and a relational rehabilitation training parameter scheme may be pushed to the user.

In one embodiment, a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device is obtained, removing training synchronous information of a non-matching relation rehabilitation training parameter scheme, acquiring key probabilities of key rehabilitation subscription objects in rehabilitation subscription objects corresponding to all relation rehabilitation training strategies associated with each training synchronous data based on the rehabilitation training push network, acquiring historical balance evaluation indexes corresponding to each training synchronous information, and finally acquiring the key probabilities corresponding to each training synchronous information based on the key probabilities, and the historical balance evaluation index corresponding to each training synchronous information, and acquiring the reference information prompt index corresponding to the target rehabilitation training push application, and prompting the relation rehabilitation training parameter scheme for a user corresponding to the target rehabilitation training push application according to a reference information prompting index. Therefore, the classification probability assessment can be quantized, the basis is improved for determining the relation rehabilitation training parameter scheme, the classification probability assessment is carried out based on the key probability corresponding to various training synchronous information related to the target rehabilitation training pushing application, the accuracy of the classification probability assessment is improved, in addition, when the reference information prompt indexes are predicted, the historical balance evaluation indexes corresponding to all the training synchronous information are obtained and used as the references of the classification probability assessment of all the training synchronous information, the accuracy of the classification probability assessment is further improved, whether the relation rehabilitation training parameter scheme is pushed for the user or not can be accurately determined, and the influence brought by the reference information prompt indexes of the user is avoided.

The following describes a flow of a reference information generating method based on artificial intelligence and smart medical treatment according to still another embodiment of the present disclosure, where the reference information generating method based on artificial intelligence and smart medical treatment may include the following steps:

step A310, a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device is obtained, the rehabilitation training push network comprises a target rehabilitation training strategy corresponding to the target rehabilitation training push application, a plurality of relational rehabilitation training strategies and a plurality of training synchronous information, the training synchronous information corresponds to a rehabilitation activity rule of a rehabilitation subscription object of the target rehabilitation training push application, and each relational rehabilitation training strategy is synchronously configured with the target rehabilitation training strategy through the corresponding training synchronous information.

In the embodiment of the present disclosure, the content of step a310 in the foregoing embodiment can be referred to, and is not described herein again.

Step a320, obtaining an evaluation index of a designated rehabilitation training strategy in all the relationship rehabilitation training strategies associated with each piece of training synchronization data, as a second evaluation index, where the designated rehabilitation training strategy is a relationship rehabilitation training strategy corresponding to a rehabilitation training strategy of the authenticated relationship rehabilitation training parameter scheme.

Step A330, according to a second evaluation index corresponding to each piece of training synchronization information, removing target training synchronization information from all pieces of training synchronization information, wherein the second evaluation index corresponding to the target training synchronization information is smaller than a preset evaluation index.

Further, whether a second evaluation index corresponding to each piece of training synchronization information is smaller than a preset evaluation index is judged, if the judgment result is yes, the target training synchronization information with the second evaluation index smaller than the preset evaluation index is removed, for example, if the second evaluation index is 2 and the preset evaluation index is 5, correspondingly, the training synchronization information corresponding to the second evaluation index is removed as the target training synchronization information, and it can be understood that the smaller the second evaluation index is, the smaller the effect of the training synchronization information in the classification process is, and if the training synchronization information is introduced into the classification probability evaluation process, the result of the classification probability evaluation is inaccurate, so that the target training synchronization information needs to be removed.

Step A340, based on the rehabilitation training push network, obtaining a key probability of a key rehabilitation subscription object in rehabilitation subscription objects corresponding to all relationship rehabilitation training strategies associated with each training synchronous data, wherein the key rehabilitation subscription object is a rehabilitation subscription object with a prescription matching relationship when matching the relationship rehabilitation training parameter scheme.

Step A350, obtaining a historical balance evaluation index corresponding to each training synchronization information.

Step A360, obtaining a reference information prompt index corresponding to the target rehabilitation training push application based on the key probability corresponding to each training synchronous information and the historical balance evaluation index corresponding to each training synchronous information, wherein the reference information prompt index is used for prompting the relational rehabilitation training parameter scheme for a user corresponding to the target rehabilitation training push application.

Step a370, determining whether the reference information prompt indicator corresponding to the target rehabilitation training push application is greater than a preset reference information prompt indicator threshold.

And step A380, if yes, prompting the relation rehabilitation training parameter scheme for the user corresponding to the target rehabilitation training pushing application.

Step A390, if not, generating prompt information, and sending the prompt information to a rehabilitation subscription object of the target rehabilitation training push application, where the prompt information is used to prompt a user that the user cannot participate in the relational rehabilitation training parameter scheme.

In the embodiment of the present disclosure, the contents of step a340 to step a390 may refer to the contents of step a220 to step a290 in the foregoing embodiment, and are not described herein again.

In one embodiment, a rehabilitation training push network corresponding to the target rehabilitation training push application of the medical monitoring terminal device is obtained, removing training synchronous information of a non-matching relation rehabilitation training parameter scheme, acquiring key probabilities of key rehabilitation subscription objects in rehabilitation subscription objects corresponding to all relation rehabilitation training strategies associated with each training synchronous data based on the rehabilitation training push network, acquiring historical balance evaluation indexes corresponding to each training synchronous information, and finally acquiring the key probabilities corresponding to each training synchronous information based on the key probabilities, and the historical balance evaluation index corresponding to each training synchronous information, and acquiring the reference information prompt index corresponding to the target rehabilitation training push application, and prompting the relation rehabilitation training parameter scheme for a user corresponding to the target rehabilitation training push application according to a reference information prompting index. Therefore, the classification probability assessment can be quantized, the basis is improved for determining the relation rehabilitation training parameter scheme, the classification probability assessment is carried out based on the key probability corresponding to various training synchronous information related to the target rehabilitation training pushing application, the accuracy of the classification probability assessment is improved, in addition, when the reference information prompt indexes are predicted, the historical balance evaluation indexes corresponding to all the training synchronous information are obtained and used as the references of the classification probability assessment of all the training synchronous information, the accuracy of the classification probability assessment is further improved, whether the relation rehabilitation training parameter scheme is pushed for the user or not can be accurately determined, and the influence brought by the reference information prompt indexes of the user is avoided.

In one embodiment, the above method may comprise the steps of:

step a, obtaining a relational rehabilitation training parameter scheme of user prompt corresponding to target rehabilitation training push application based on a reference information prompt index, and obtaining rehabilitation plan execution information aiming at the relational rehabilitation training parameter scheme.

And b, analyzing the acquired rehabilitation plan execution information to obtain a current key execution action and limb part information related to the current key action, and obtaining missing plan execution information and missing rehabilitation plan flow information through the rehabilitation plan execution information, the current key execution action and the limb part information related to the current key action.

And c, processing the missing plan execution information and the missing rehabilitation plan flow information by adopting a preset supplementary training extension scheme, and generating training extension suggestion information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the preset supplementary training extension scheme to obtain supplementary training extension suggestion information corresponding to the rehabilitation plan execution information.

Therefore, the acquired rehabilitation plan execution information is analyzed to accurately analyze the current key execution action and the limb part information related to the current key action. And secondly, detecting missing plan execution information based on the rehabilitation plan execution information, the current key execution action and the limb part information related to the current key action, and further ensuring high matching of the obtained missing plan execution information and the missing rehabilitation plan flow information. On the basis, the missing plan execution information and the missing rehabilitation plan flow information are analyzed and processed by the aid of the preset supplementary training extension scheme, so that training extension suggestion information is generated by the missing rehabilitation plan flow information and the missing rehabilitation plan execution information when the preset supplementary training extension scheme is matched, and the supplementary training extension suggestion information can be accurately and quickly obtained. In addition, when the supplementary training extended suggestion information is screened by the method, the execution action and the action-related limb part can be taken into account, so that omission of part of the training extended suggestion information can be avoided, and the integrity of the obtained supplementary training extended suggestion information can be ensured.

In one embodiment, the method comprises the steps of:

the STP202 acquires a relational rehabilitation training parameter scheme based on a reference information prompt indicator for a user prompt corresponding to the target rehabilitation training push application, acquires rehabilitation plan execution information for the relational rehabilitation training parameter scheme, and determines basic execution action information of the rehabilitation plan execution information, where the basic execution action information includes a basic execution action and information of a limb part related to the basic action.

The rehabilitation plan execution information refers to data information recorded in the rehabilitation plan execution process, and the training extended suggestion information can be extended suggestions which have guiding significance in the training process. The basic execution action information is obtained by identifying execution action information of training extended suggestion information in the rehabilitation plan execution information, and is used as execution action information when data is associated, and for example, the basic execution action and basic action related limb part information may be included. The basic execution action refers to basic execution action record information in the identified training extension suggestion information. The basic motion-related limb part information refers to recorded information of basic motion-related limb parts in the training extension suggestion information obtained by recognition.

For example, the cloud computing service system 100 may obtain a relational rehabilitation training parameter scheme of user prompts corresponding to the target rehabilitation training push application based on the reference information prompt indicator, and obtain rehabilitation plan execution information for the relational rehabilitation training parameter scheme. The cloud computing service system 100 may determine basic execution action information of the rehabilitation plan execution information, where the basic execution action information includes a basic execution action and basic action-related limb part information. For example, the basic execution action information of the rehabilitation plan execution information may be determined by the training extended recommendation information retrieval strategy and the training extended recommendation information matching strategy. For example, the basic action related limb part information in the basic execution action information obtained after the action information is determined by the training extended recommendation information is the action related limb part information of the training extended recommendation information under the current missing plan execution information, and cannot be specified in advance on the search model corresponding to the reference training extended recommendation information. And the content except the basic action related limb part information can establish a matching relation in a search model corresponding to the reference training extended suggestion information. For example, the basic execution action is matched with the corresponding key execution action in the search model corresponding to the reference training extended suggestion information.

And the STP204 selects a current key execution action from the current reference training extended suggestion information corresponding to the rehabilitation plan execution information, and acquires the corresponding limb part information related to the current key action based on the rehabilitation plan execution information.

The current reference training extended suggestion information refers to current reference training extended suggestion information corresponding to rehabilitation plan execution information obtained according to a search model corresponding to the reference training extended suggestion information used in advance. The search model corresponding to the reference training extended suggestion information used in advance refers to a preset generation model of the training extended suggestion information.

For example, the cloud computing service system 100 acquires a search model corresponding to reference training extension suggestion information used in advance. And obtaining current reference training extended suggestion information corresponding to the rehabilitation plan execution information according to a search model corresponding to the reference training extended suggestion information used in advance, and then selecting current key execution actions according to the feature definition of each content in the current reference training extended suggestion information. For example, the cloud computing service system 100 may use a machine learning model (such as a deep neural network model) to obtain the current reference training extension recommendation information.

And the STP206 detects the missing plan execution information based on the current key execution action, the information of the limb part related to the current key action and the basic execution action information to obtain the missing plan execution information.

The missing plan execution information refers to corresponding missing plan execution information in the comparison process. The missing plan execution information detection means that the detection of the missing plan execution information result is performed based on the data-related parameter. The missing plan execution information is used to characterize the missing plan execution information of the training extension recommendation information.

For example, the cloud computing service system 100 performs data association analysis on the current key execution action and the limb part information related to the current key action, and then detects missing plan execution information in the local association information comparison process to obtain the missing plan execution information.

And the STP208 selects the limb part information related to the target action from the current reference training extended suggestion information according to the missing plan execution information, and determines the missing rehabilitation plan flow information corresponding to the current reference training extended suggestion information according to the limb part information related to the target action and the current key execution action.

The target action related limb part information is key action related limb part information obtained by converting missing plan execution information of the current reference training extended suggestion information according to the missing plan execution information. The missing rehabilitation plan flow information refers to action-related limb part information obtained based on a search model of a reference training extended recommendation information application service layer according to target action-related limb part information and current key execution actions, and the action-related limb part information is used for representing application distribution of action-related limbs of the training extended recommendation information.

For example, the cloud computing service system 100 performs missing plan execution information conversion on the current reference training extended suggestion information according to the missing plan execution information, selects key action related limb part information from the reference training extended suggestion information after the missing plan execution information conversion as target action related limb part information, and then determines missing rehabilitation plan flow information corresponding to the current reference training extended suggestion information according to the target action related limb part information and the current key execution action based on a search model of a reference training extended suggestion information application service layer.

STP210, based on the missing plan execution information, performing data association between the target action related limb part information and the current key execution action to obtain basic data associated information, and updating the current key execution action and the current key action related limb part information according to the basic data associated information and first comparison information of the basic execution action information.

Therein, the data association can be understood as a linear fusion. The first comparison information refers to content comparison information between the basic data association information and the basic execution action information.

For example, the cloud computing service system 100 performs missing plan execution information conversion on the target action related limb part information and the current key execution action through the missing plan execution information to obtain a converted undetermined basic action, performs basic execution action corresponding to the data associated content on the converted undetermined basic action to obtain basic data associated information, the basic data correlation information comprises each basic correlation execution action and the related limb part information of each basic correlation action, content comparison information between each basic correlation execution action and each corresponding basic execution action in the basic execution action information is determined, and determining content comparison information between each piece of basic associated action related limb part information and each piece of basic action related limb part information corresponding to the basic execution action information, and determining the sum of evaluation indexes of the content comparison information to obtain first comparison information.

STP212 determines whether the first supplemental training extension scheme is matched, executes STP202 when the first supplemental training extension scheme is matched, and returns to STP206 when the first supplemental training extension scheme is not matched.

And the STP214 generates training extended suggestion information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the first supplementary training extended scheme, so as to obtain supplementary training extended suggestion information corresponding to the rehabilitation plan execution information.

The first supplementary training extension scheme refers to conditions for generating training extension suggestion information, and comprises the steps that an evaluation index corresponding to the first comparison information is smaller than a preset threshold value, a preset cycle number is reached, or missing rehabilitation plan flow information and missing plan execution information are obtained and do not have obvious abnormal changes. The missing rehabilitation plan flow information and the missing plan execution information are not obviously abnormally changed, which means that the evaluation index between the missing rehabilitation plan flow information and the missing plan execution information obtained in the previous time and the missing rehabilitation plan flow information and the missing plan execution information obtained in the next time is smaller than a preset threshold. The supplementary training extended suggestion information is reference training extended suggestion information obtained by using the missing rehabilitation plan flow information and the missing plan execution information which are matched with the first supplementary training extended scheme to generate training extended suggestion information.

For example, when the cloud computing service system 100 determines whether the first supplementary training extended scenario is matched, when the first supplementary training extended scenario is matched, generating training extended recommendation information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the first supplementary training extended scenario is executed, and when the first supplementary training extended scenario is not matched, returning to the STP204, that is, detecting the missing plan execution information based on the current key execution action, the limb part information related to the current key action, and the basic execution action information, so as to obtain the missing plan execution information. And continuously circulating the wandering circulation until the first supplementary training extension scheme is matched.

In the method, during each wandering cycle, the missing plan execution information is used for selecting the limb part information related to the key action, so that more accurate limb part information related to the key action can be selected, then the missing rehabilitation plan flow information corresponding to the current reference training extended suggestion information is determined by using the limb part information related to the target action and the current key execution action, so that more accurate missing rehabilitation plan flow information can be obtained by using the limb part information related to the target action and the current key execution action to determine the missing rehabilitation plan flow information during each wandering cycle, then when the supplementary training extended scheme is matched, the missing rehabilitation plan flow information and the missing plan execution information are used for generating the training extended suggestion information, because the missing rehabilitation plan flow information and the missing plan execution information are used for generating the training extended suggestion information, the problem of omission in the analysis processing of the training extended suggestion information is avoided, and the supplementary training extended suggestion information can be accurately and quickly obtained.

In one embodiment, the step of detecting missing planned execution information is returned after updating the current critical execution action and the limb part information related to the current critical execution action according to the first comparison information of the basic data association information and the basic execution action information until matching the first supplementary training extension scheme, and the method comprises the following steps:

and the STP302 determines to obtain first comparison information based on the basic data correlation information and the basic execution action information, and updates the current reference training extended suggestion information based on the missing rehabilitation plan process information to obtain updated reference training extended suggestion information when the first comparison information does not match the first supplementary training extended scheme.

The updated reference training extended recommendation information refers to the reference training extended recommendation information obtained by updating the current reference training extended recommendation information by using the missing rehabilitation plan flow information.

And the STP304 selects an updated key execution action from the updated reference training extension suggestion information to obtain an updated current key execution action, uses the limb part information related to the target action as the updated limb part information related to the current key action, and returns to the step of detecting missing plan execution information based on the current key execution action, the limb part information related to the current key action and the basic execution action information to obtain the missing plan execution information until the first supplementary training extension scheme is matched.

And the updated current key execution action refers to a key execution action determined from the characteristic of the pending basic action in the updated reference training extension suggestion information.

For example, the cloud computing service system 100 selects an update key execution action from the update reference training extended suggestion information, obtains an updated current key execution action, uses the target action related limb part information as updated current key action related limb part information, and returns to perform missing plan execution information detection based on the current key execution action, the current key action related limb part information, and the basic execution action information, and performs a walking loop until the step of obtaining missing plan execution information matches the first supplementary training extended scheme.

In the above embodiment, by determining the first comparison information, when the first comparison information does not match the first supplementary training extended proposal, an updated key execution action is selected from the updated reference training extended proposal information to obtain an updated current key execution action, the target action related limb part information is used as the updated current key action related limb part information, and the missing plan execution information detection is performed based on the current key execution action, the current key action related limb part information and the basic execution action information to obtain the missing plan execution information until the first supplementary training extended proposal is matched, so that a loop wandering loop can be performed continuously to obtain more accurate missing plan execution information, and the accuracy and the reliability of the determined supplementary training extended proposal information are higher.

In one embodiment, the rehabilitation plan execution information is label distribution of floating training extended suggestion information, and the basic data association information comprises basic association execution actions and basic association action related limb part information.

STP302, determining to obtain first comparison information based on the basic data association information and the basic execution action information, including:

determining to obtain execution action comparison information based on the basic associated execution action and the basic execution action, and determining to obtain action-related limb part comparison information based on the basic associated action-related limb part information and the basic action-related limb part information; and obtaining first comparison information of basic data association information and basic execution action information based on the action related limb part comparison information and the execution action comparison information.

The label distribution of the floating training extended suggestion information refers to the fact that regular floating matters exist in the training extended suggestion information labels, and the execution action comparison information refers to comparison information of the execution action record information of the basic associated execution action and the basic execution action. The motion-related limb part comparison information is comparison information of the motion-related limb part information of the basic associated motion-related limb part information and the motion-related limb part record information of the basic motion-related limb part information.

For example, when the cloud computing service system 100 detects that the rehabilitation plan execution information is the label distribution of the floating training extended recommendation information, determining evaluation indexes between action characteristics of each basic associated execution action corresponding to label distribution of the floating training extended suggestion information and action characteristics of each corresponding basic execution action to obtain an evaluation index of each execution action, determining the sum of the evaluation indexes of each execution action to obtain execution action comparison information, and determining an evaluation index between the action characteristics of each piece of basic associated action related limb part information corresponding to the label distribution of the floating training extended suggestion information and the action characteristics of each piece of basic action related limb part information corresponding to the label distribution of the floating training extended suggestion information to obtain the evaluation index of each piece of action related limb part information, determining the sum of the evaluation indexes of each piece of action related limb part information to obtain action related limb part comparison information. And then determining the sum of the execution action comparison information and the action-related limb part comparison information to obtain first comparison information of the basic data association information and the basic execution action information.

In the above embodiment, when the rehabilitation plan execution information is the label distribution of the floating training extended suggestion information, the first comparison information of the basic data association information and the basic execution action information is obtained directly by determining the execution action comparison information and the action-related limb part comparison information, so that the efficiency of obtaining the first comparison information is improved.

In one embodiment, the rehabilitation plan execution information is label distribution of the floating training extended recommendation information;

the STP204, based on the training extended recommendation information category attribute, obtains the corresponding limb part information related to the current key action, including:

and the STP602 is used for acquiring target action associated parameters, associating the current key execution action with the basic execution action according to the target action associated parameters to obtain an associated execution action, and detecting missing plan execution information based on the associated execution action and the basic execution action to obtain key missing plan execution information.

The target action associated parameter refers to an action associated parameter determined by detecting the action associated parameter from each preset action associated parameter. The associated execution action refers to an updated basic action obtained by associating the current key execution action with data. The basic action to be updated is the basic action characteristic, and the basic action to be determined is the characteristic of the action to be determined. The key missing plan execution information is the missing plan execution information obtained when only key execution actions are used for detecting the missing plan execution information when the rehabilitation plan execution information is the label distribution of the floating training extended suggestion information.

For example, when the rehabilitation plan execution information is label distribution of the floating training extended suggestion information, the cloud computing service system 100 obtains the target action associated parameter, and associates the current key execution action to the basic execution action through data according to the target action associated parameter to obtain an associated execution action.

Then, when obtaining the associated execution action, the cloud computing service system 100 determines missing plan execution information with a small local fusion evaluation index of the associated execution action and the basic execution action, and obtains key missing plan execution information.

STP604 selects current key action related limb part information corresponding to label distribution of floating training extended suggestion information from the action database of the action related limb part of the reference training extended suggestion information of the current reference training extended suggestion information according to the key missing plan execution information.

The action database of the action-related limb part of the reference training extended suggestion information refers to preset calling information of an action-related limb part object of the action-related limb part information of the reference training extended suggestion information.

In the above embodiment, when the rehabilitation plan execution information is the label distribution of the floating training extended suggestion information, the current key action related limb part information corresponding to the label distribution of the floating training extended suggestion information is selected from the action database of the action related limb part of the reference training extended suggestion information of the current reference training extended suggestion information according to the key missing plan execution information, so that more accurate current key action related limb part information can be obtained, and subsequent use is facilitated.

In one embodiment, STP602, obtaining the target action associated parameter includes:

STP702 obtains each preset action associated parameter, and selects a current action associated parameter from each preset action associated parameter.

The preset action associated parameter refers to a preset action associated parameter value. The current motion-related parameter is a motion-related parameter used when determining the motion-related parameter.

STP704, associating the current key execution action to the basic execution action according to the current action association parameter to obtain an associated execution action corresponding to the action association parameter, detecting missing plan execution information based on the associated execution action corresponding to the action association parameter and the basic execution action to obtain missing plan execution information corresponding to the action association parameter, and selecting key action related limb part information corresponding to the action association parameter from the action database of the action related limb part of the reference training extended suggestion information of the current reference training extended suggestion information according to the missing plan execution information corresponding to the action association parameter.

The associated execution action corresponding to the action associated parameter refers to an associated execution action obtained when data association is performed by using the current action associated parameter. The missing plan execution information corresponding to the action correlation parameter is the missing plan execution information obtained by detecting the missing plan execution information according to the correlation execution action and the basic execution action corresponding to the action correlation parameter. The key action related limb part information corresponding to the action related parameters refers to the key action related limb part information selected by using the missing plan execution information corresponding to the action related parameters.

For example, the cloud computing service system 100 associates the current key execution action with the basic execution action according to the current action association parameter to obtain an associated execution action corresponding to the action association parameter, and then determines missing plan execution information of the associated execution action corresponding to the action association parameter and the basic execution action in the local association information comparison process to obtain missing plan execution information corresponding to the action association parameter. And then selecting candidate action data from an action database of the action related limb part of the reference training extended suggestion information of the current reference training extended suggestion information according to the missing plan execution information corresponding to the action related parameters, and determining key action related limb part information corresponding to the action related parameters from the candidate action data.

And the STP706 detects missing plan execution information corresponding to the action associated parameters based on the key action related limb part information corresponding to the action associated parameters, the current key execution action and the basic execution action information to obtain the missing plan execution information corresponding to the action associated parameters.

The missing plan execution information corresponding to the action related parameters refers to the missing plan execution information corresponding to the action related parameters, which is obtained by detecting the missing plan execution information by using the key action related limb part information corresponding to the action related parameters

For example, the cloud computing service system 100 associates the key action related limb part information corresponding to the action related parameter and the current key execution action progression data into the basic execution action, obtains basic data related information corresponding to the action related parameter, determines missing plan execution information when the evaluation index between the basic data related information corresponding to the action related parameter and the basic execution action information converges, and obtains missing plan execution information corresponding to the action related parameter.

STP708, selecting target motion related limb part information of the motion related parameters from the motion database of the motion related limb part of the reference training extended suggestion information according to the missing plan execution information corresponding to the motion related parameters, and determining missing rehabilitation plan flow information of the motion related parameters corresponding to the current reference training extended suggestion information according to the target motion related limb part information of the motion related parameters and the current key execution motion.

The target action related limb part information of the action related parameters refers to target action related limb part information selected according to the missing plan execution information corresponding to the action related parameters. The missing rehabilitation plan flow information of the action related parameters refers to the missing rehabilitation plan flow information determined according to the target action related limb part information of the action related parameters.

For example, the cloud computing service system 100 selects candidate motion data from a motion database of motion-related limb parts of the reference training extended suggestion information, and then selects target motion-related limb part information of the motion-related parameters from the candidate motion data according to missing plan execution information corresponding to the motion-related parameters. And then determining missing rehabilitation plan flow information of the action associated parameters corresponding to the current reference training extended suggestion information according to the target action related limb part information of the action associated parameters and the current key execution action.

STP710, performing data association on the target action related limb part information of the action related parameter and the current key execution action based on the missing plan execution information corresponding to the action related parameter to obtain basic data associated information corresponding to the action related parameter, and updating the key action related limb part information corresponding to the action related parameter and the current key execution action according to the basic data associated information corresponding to the action related parameter and the second comparison information of the basic execution action information.

The basic data associated information corresponding to the action associated parameter refers to basic data associated information obtained when data association is performed on target action related limb part information of the action associated parameter and a current key execution action according to missing plan execution information corresponding to the action associated parameter. The second comparison information is content comparison information between the basic data association information corresponding to the action association parameter and the basic execution action information.

For example, the cloud computing service system 100 analyzes the target motion-related limb part information of the motion-related parameter and the missing plan execution information corresponding to the current key execution motion using the motion-related parameter.

And then performing data association on the target action related limb part information of the action related parameters of the missing plan execution information analysis and the current key execution action to obtain basic data associated information corresponding to the action related parameters, and then determining the content information comparison between the basic data associated information corresponding to the action related parameters and the basic execution action information to obtain second comparison information. And updating the key action related limb part information corresponding to the action correlation parameter and the current key execution action according to the second comparison information.

And the STP712, determining whether the second supplementary training extension scheme is matched, executing STP714 when the second supplementary training extension scheme is matched, and returning to the STP706 for execution when the second supplementary training extension scheme is not matched.

STP714 obtains current second comparison information corresponding to the current action correlation parameter.

The second supplementary training extension scheme is a supplementary training extension scheme in which the second comparison information is local comparison information corresponding to the current motion correlation parameter. The missing rehabilitation plan flow information of the action associated parameters and the missing plan execution information corresponding to the action associated parameters do not have obvious abnormal changes when the preset cycle times are reached, namely the missing rehabilitation plan flow information of the action associated parameters and the missing plan execution information corresponding to the action associated parameters are consistent with values obtained in the last walking cycle and the current walking cycle.

For example, when the preset number of cycles is reached, that is, when the second supplementary training extension scheme is matched, the cloud computing service system 100 takes the second comparison information when the second supplementary training extension scheme is matched as the current second comparison information corresponding to the current action related parameter. When the preset loop times are not reached, that is, the second supplemental training extension scheme is not matched, the STP706 is returned to continue the execution of the wandering loop.

And the STP716, traversing each preset action associated parameter to obtain each current second comparison information corresponding to each preset action associated parameter, comparing each current second comparison information to obtain target second comparison information, and taking the preset action associated parameter corresponding to the target second comparison information as the target action associated parameter.

The target second comparison information refers to current second comparison information with the smallest evaluation index in each current second comparison information.

For example, the cloud computing service system 100 traverses each preset action associated parameter, that is, returns to the step of selecting the current action associated parameter from each preset action associated parameter, and the selected preset action associated parameter is not repeatedly selected. Until obtaining each current second comparison information corresponding to each preset action correlation parameter. And then comparing the current second comparison information to obtain target second comparison information, and taking a preset action associated parameter corresponding to the target second comparison information as a target action associated parameter. Namely, the convergence evaluation index corresponding to each preset action associated parameter is determined, then the minimum convergence evaluation index is further selected from the convergence evaluation indexes to serve as target second comparison information, and the preset action associated parameter corresponding to the target second comparison information serves as the target action associated parameter. Then, the cloud computing service system 100 specifies the target action related parameter, that is, the cloud computing service system 100 directly uses the target action related parameter when generating the training extended recommendation information for the subsequent training extended recommendation information category attribute.

In the above embodiment, the current second comparison information of each preset action associated parameter is determined, then the target second comparison information is determined from the current second comparison information, and the preset action associated parameter corresponding to the target second comparison information is used as the target action associated parameter, so that the obtained target action associated parameter is more accurate.

The functions of the respective functional blocks of the artificial intelligence and smart medical-based reference information generating apparatus 300 will be described in detail below.

The acquiring module 310 is configured to acquire gesture measurement data of each key part of the medical user and gesture image data corresponding to the gesture measurement data, which are recorded by the medical monitoring terminal device in a preset recording period;

the establishing module 320 is configured to determine gait feature information of the medical user according to the posture measurement data of each key part, and establish gait image corresponding data between the gait feature information of the medical user and corresponding posture image data, where the gait image corresponding data is used to establish corresponding data between each gait feature information unit in the gait feature information and a posture image frame corresponding to the corresponding posture image data;

the input module 30 is used for inputting the gait image corresponding data into a pre-trained artificial intelligence network model to obtain a user obstacle classification attribute set corresponding to the gait image corresponding data;

the generating module 340 is configured to generate corresponding reference prompt information according to the user obstacle classification attribute set corresponding to the gait image corresponding data, where the reference prompt information is used to perform an abnormal state early warning on the medical user.

Referring to the hardware structure diagram of the cloud computing service system 100 for implementing the artificial intelligence and smart medical reference information generation method, the cloud computing service system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

In a specific implementation process, at least one processor 110 executes computer-executable instructions stored in the machine-readable storage medium 120, so that the processor 110 may execute the reference information generation method based on artificial intelligence and smart medicine according to the above method embodiment, the processor 110, the machine-readable storage medium 120, and the communication unit 140 are connected through the bus 130, and the processor 110 may be configured to control the transceiving action of the communication unit 140, so as to perform data transceiving with the medical monitoring terminal device 200.

For a specific implementation process of the processor 110, reference may be made to the above-mentioned method embodiments executed by the cloud computing service system 100, which implement principles and technical effects are similar, and details of this embodiment are not described herein again.

In addition, the embodiment of the disclosure also provides a readable storage medium, wherein the readable storage medium is preset with a computer execution instruction, and when a processor executes the computer execution instruction, the reference information generation method based on artificial intelligence and intelligent medical treatment is realized.

Finally, it should be understood that the examples in this specification are only intended to illustrate the principles of the examples in this specification. Other variations are also possible within the scope of this description. Accordingly, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be seen as matching the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. The reference information generation system based on artificial intelligence and smart medical treatment is characterized by comprising a cloud computing service system and a plurality of medical monitoring terminal devices in communication connection with the cloud computing service system;

the cloud computing service system is used for:

generating corresponding reference prompt information according to a user obstacle classification attribute set corresponding to the gait image corresponding data, wherein the reference prompt information is used for carrying out abnormal state early warning on the medical user;

the cloud computing service system is further configured to:

acquiring rehabilitation training motion data of the medical user after the targeted rehabilitation training is performed on the prompted abnormal early warning information;

analyzing the rehabilitation training motion data based on a pre-trained rehabilitation training label learning network to obtain a rehabilitation training label set corresponding to the rehabilitation training motion data;

according to a rehabilitation training label set corresponding to the rehabilitation training motion data, further reference information prompting is carried out on the medical user corresponding to the gait image data;

the method for further referring to information prompt for the medical user corresponding to the gait image corresponding data according to the rehabilitation training label set corresponding to the rehabilitation training motion data comprises the following steps:

acquiring a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device according to a rehabilitation training label set corresponding to the rehabilitation training motion data, wherein the rehabilitation training push network comprises a target rehabilitation training strategy, a plurality of relational rehabilitation training strategies and a plurality of training synchronous information, the target rehabilitation training push application corresponds to the rehabilitation training label set, the training synchronous information corresponds to rehabilitation activity rules of a rehabilitation subscription object of the target rehabilitation training push application, and each relational rehabilitation training strategy is synchronously configured with the target rehabilitation training strategy through the corresponding training synchronous information;

based on the rehabilitation training push network, obtaining a key probability of a key rehabilitation subscription object in rehabilitation subscription objects corresponding to all relationship rehabilitation training strategies related to each training synchronous data, wherein the key rehabilitation subscription object is a rehabilitation subscription object with a prescription matching relationship when a relationship rehabilitation training parameter scheme is matched, and the key probability is a ratio of the rehabilitation subscription object with the prescription matching relationship when the relationship rehabilitation training parameter scheme is matched to all rehabilitation subscription objects under a certain rehabilitation activity rule corresponding to the rehabilitation subscription object;

acquiring a historical balance evaluation index corresponding to each training synchronous information, and acquiring a reference information prompt index corresponding to the target rehabilitation training push application based on the key probability corresponding to each training synchronous information and the historical balance evaluation index corresponding to each training synchronous information, wherein the reference information prompt index is used for prompting the relational rehabilitation training parameter scheme for a user corresponding to the target rehabilitation training push application, the size of the historical balance evaluation index can reflect the confidence coefficient of the training synchronous information in predicting the key probability of the target rehabilitation training strategy, and the larger the historical balance evaluation index is, the higher the confidence coefficient of the training synchronous information in predicting the target rehabilitation training strategy degree is;

the gait feature information comprises a plurality of gait feature information units which take unit time sequence segments as segmentation nodes;

the artificial intelligence network model is obtained by training in advance based on a training sample set, the training sample set comprises gait images corresponding data samples and corresponding user obstacle classification attribute set samples, so that the initial artificial intelligence network model performs feature learning based on the gait images corresponding data samples and the corresponding user obstacle classification attribute set samples to further obtain the artificial intelligence network model which is completed by training in advance, wherein the user obstacle classification attribute set comprises one or more user obstacle classification attributes, the user obstacle classification attributes are used for representing classification attribute conditions of cognitive impairment of the medical user, and the classification attributes comprise probability value distribution of each user obstacle classification;

wherein, the cloud computing service system is further configured to:

acquiring reference positive rehabilitation training motion data and reference negative rehabilitation training motion data when a reference rehabilitation training motion condition is generated, and performing feature extraction on the reference positive rehabilitation training motion data to obtain reference positive rehabilitation training motion features of the reference positive rehabilitation training motion data, wherein the reference positive rehabilitation training motion features are used for indicating positive feedback motion features of the reference rehabilitation training motion condition;

performing feature extraction on the reference negative rehabilitation training motion data to obtain reference negative rehabilitation training motion data of the reference negative rehabilitation training motion data, wherein the reference negative rehabilitation training motion data is used for indicating negative feedback motion features of the reference negative rehabilitation training motion situation;

taking the reference negative rehabilitation training motion data as opponent data of the reference positive rehabilitation training motion characteristics to generate training basic data of the rehabilitation training label learning network;

training the rehabilitation training label learning network by adopting the training basic data, wherein the rehabilitation training label learning network is used for analyzing based on rehabilitation training motion data to obtain a rehabilitation training label set corresponding to the rehabilitation training motion data;

the relation rehabilitation training strategy is a rehabilitation training strategy related to the target rehabilitation training strategy, and the training synchronous information provides related training flow direction information between the target rehabilitation training strategy and the relation rehabilitation training strategy and is used for associating the target rehabilitation training strategy with the relation rehabilitation training strategy;

the rehabilitation training push network represents the relationship between the target rehabilitation training strategy and the training recommendation schemes of other relationship rehabilitation training strategies.

2. The system according to claim 1, wherein the obtaining of the historical trade-off evaluation index corresponding to each training synchronization information comprises:

obtaining an evaluation index of a designated rehabilitation training strategy in all relation rehabilitation training strategies related to each training synchronous data as a first evaluation index, wherein the designated rehabilitation training strategy is a relation rehabilitation training strategy corresponding to a certified rehabilitation training strategy of the relation rehabilitation training parameter scheme;

acquiring a historical balance evaluation index corresponding to each training synchronous information based on a preset index value and the first evaluation index corresponding to each training synchronous information;

the obtaining of the historical trade-off evaluation index corresponding to each training synchronization information based on a preset index value and the first evaluation index corresponding to each training synchronization information includes:

obtaining the opposite number corresponding to the ratio of the preset index value to each first evaluation index to obtain a target evaluation index corresponding to each training synchronization information;

and taking the target evaluation index corresponding to each piece of training synchronous information as an independent variable, acquiring a function value of a preset target function, and obtaining a historical balance evaluation index corresponding to each piece of training synchronous information, wherein the base number value of the preset target function is greater than 1.

3. The system for generating reference information based on artificial intelligence and smart medical care according to claim 1, wherein the acquiring a rehabilitation training push network corresponding to a target rehabilitation training push application of the medical monitoring terminal device according to the rehabilitation training label set corresponding to the rehabilitation training motion data includes:

according to a rehabilitation training label set corresponding to the rehabilitation training motion data, acquiring a rehabilitation activity rule of a rehabilitation subscription object applied by the target rehabilitation training push and a rehabilitation activity rule of a relational rehabilitation training strategy corresponding to the target rehabilitation training push, wherein the rehabilitation activity rule is used for representing a rehabilitation training strategy characteristic of a rehabilitation training strategy;

and generating a rehabilitation training push network corresponding to the target rehabilitation training push application based on the rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the rehabilitation activity rule of the corresponding relation rehabilitation training strategy of the target rehabilitation training push application, taking the rehabilitation subscription object of the target rehabilitation training strategy as the target rehabilitation training strategy and taking the relation rehabilitation training strategy as the relation rehabilitation training strategy.

4. The artificial intelligence and smart medical-based reference information generating system of claim 3, wherein after the obtaining of the rehabilitation activity rules of the rehabilitation subscription subject of the target rehabilitation training push application and the rehabilitation activity rules of the corresponding relational rehabilitation training strategy of the target rehabilitation training push application, the cloud computing service system is further configured to:

comparing the relevance of the rehabilitation activity rule of the rehabilitation subscription object applied by the target rehabilitation training push with a preset rehabilitation activity rule, wherein the preset rehabilitation activity rule corresponds to the relational rehabilitation training parameter scheme;

according to the correlation comparison result, acquiring a rehabilitation activity rule matched with the preset rehabilitation activity rule from the rehabilitation activity rules of the rehabilitation subscription object pushed and applied by the target rehabilitation training as a target rehabilitation activity rule;

the generating of the rehabilitation training push network corresponding to the target rehabilitation training push application based on the rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the rehabilitation activity rule of the corresponding relationship rehabilitation training strategy of the target rehabilitation training push application, with the rehabilitation subscription object of the target rehabilitation training strategy as the target rehabilitation training strategy and with the relationship rehabilitation training strategy as the relationship rehabilitation training strategy, comprises:

and generating a rehabilitation training push network corresponding to the target rehabilitation training push application based on the target rehabilitation activity rule of the rehabilitation subscription object of the target rehabilitation training push application and the target rehabilitation activity rule of the corresponding relation rehabilitation training strategy of the target rehabilitation training push application, taking the rehabilitation subscription object of the target rehabilitation training strategy as the target rehabilitation training strategy and taking the relation rehabilitation training strategy as the relation rehabilitation training strategy.

5. The system for generating reference information based on artificial intelligence and smart medical treatment as claimed in claim 1, wherein before said obtaining the key probability of the key rehabilitation subscription object in the rehabilitation subscription objects corresponding to all the relationship rehabilitation training strategies associated with each training synchronization data based on the rehabilitation training push network, the cloud computing service system is further configured to:

obtaining an evaluation index of a designated rehabilitation training strategy in all relation rehabilitation training strategies related to each training synchronous data as a second evaluation index, wherein the designated rehabilitation training strategy is a relation rehabilitation training strategy corresponding to a rehabilitation training strategy of the authenticated relation rehabilitation training parameter scheme;

and according to a second evaluation index corresponding to each training synchronous information, removing target training synchronous information from all training synchronous information, wherein the second evaluation index corresponding to the target training synchronous information is smaller than a preset evaluation index.

6. The system for generating reference information based on artificial intelligence and intelligent medical treatment as claimed in any one of claims 1 to 5, wherein said obtaining a reference information prompt indicator corresponding to said target rehabilitation training push application based on said key probability corresponding to each training synchronization information and a historical trade-off evaluation indicator corresponding to each training synchronization information comprises:

acquiring a reference information prompt index corresponding to each piece of training synchronization information based on the key probability corresponding to each piece of training synchronization information and a historical balance evaluation index corresponding to each piece of training synchronization information;

and obtaining a reference information prompt index corresponding to the target rehabilitation training push application based on the influence index corresponding to each training synchronous information and the reference information prompt index corresponding to each training synchronous information.

7. The artificial intelligence and smart medical-based reference information generating system according to any one of claims 1-5, wherein the cloud computing service system is further configured to:

acquiring a relational rehabilitation training parameter scheme of user prompts corresponding to the target rehabilitation training push application based on the reference information prompt indexes, and acquiring rehabilitation plan execution information aiming at the relational rehabilitation training parameter scheme;

analyzing the acquired rehabilitation plan execution information to obtain a current key execution action and limb part information related to the current key action, and obtaining missing plan execution information and missing rehabilitation plan flow information through the rehabilitation plan execution information, the current key execution action and the limb part information related to the current key action;

processing the missing plan execution information and the missing rehabilitation plan flow information by adopting a preset supplementary training extension scheme, and generating training extension suggestion information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the preset supplementary training extension scheme to obtain supplementary training extension suggestion information corresponding to the rehabilitation plan execution information;

the analyzing the acquired rehabilitation plan execution information to obtain a current key execution action and limb part information related to the current key action, and obtaining missing plan execution information and missing rehabilitation plan flow information through the rehabilitation plan execution information, the current key execution action and the limb part information related to the current key action, including:

determining basic execution action information of the rehabilitation plan execution information; the basic execution action information comprises a basic execution action and basic action related limb part information;

selecting a current key execution action from current reference training extended suggestion information corresponding to the rehabilitation plan execution information, and acquiring corresponding limb part information related to the current key action based on the rehabilitation plan execution information;

detecting missing plan execution information based on the current key execution action, the limb part information related to the current key action and the basic execution action information to obtain missing plan execution information;

selecting target action related limb part information from the current reference training extended suggestion information according to the missing plan execution information, and determining missing rehabilitation plan flow information corresponding to the current reference training extended suggestion information according to the target action related limb part information and the current key execution action;

the method for processing the missing plan execution information and the missing rehabilitation plan flow information by adopting the preset supplementary training extension scheme, generating training extension suggestion information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the preset supplementary training extension scheme, and obtaining supplementary training extension suggestion information corresponding to the rehabilitation plan execution information comprises the following steps:

processing the missing plan execution information and the missing rehabilitation plan flow information by adopting a preset supplementary training extension scheme, and generating training extension suggestion information based on the missing rehabilitation plan flow information and the missing rehabilitation plan execution information matched with the preset supplementary training extension scheme to obtain supplementary training extension suggestion information corresponding to the rehabilitation plan execution information:

performing data association on the target action related limb part information and the current key execution action based on the missing plan execution information to obtain basic data associated information, updating the current key execution action and the current key action related limb part information according to the basic data associated information and first comparison information of the basic execution action information, and returning to the missing plan execution information detection mode until a first supplementary training expansion scheme is matched;

generating training extended suggestion information based on the missing rehabilitation plan flow information and the missing plan execution information matched with the first supplementary training extended scheme to obtain supplementary training extended suggestion information corresponding to the rehabilitation plan execution information.

8. A reference information generation method based on artificial intelligence and smart medical treatment is applied to a cloud computing service system, the cloud computing service system is in communication connection with a plurality of medical monitoring terminal devices, and the method comprises the following steps:

the method further comprises the following steps:

the step of performing further reference information prompting on the medical user corresponding to the gait image corresponding data according to the rehabilitation training label set corresponding to the rehabilitation training motion data comprises the following steps:

based on the rehabilitation training push network, acquiring the key probability of a key rehabilitation subscription object in rehabilitation subscription objects corresponding to all relationship rehabilitation training strategies associated with each training synchronous data, wherein the key rehabilitation subscription object is a rehabilitation subscription object with a prescription matching relationship when a relationship rehabilitation training parameter scheme is matched, and the key probability is the ratio of the rehabilitation subscription object with the prescription matching relationship when the relationship rehabilitation training parameter scheme is matched to all rehabilitation subscription objects under a certain rehabilitation activity rule corresponding to the rehabilitation subscription object;

wherein the method further comprises:

acquiring reference positive rehabilitation training motion data and reference negative rehabilitation training motion data when a reference rehabilitation training motion situation is generated, and performing feature extraction on the reference positive rehabilitation training motion data to obtain reference positive rehabilitation training motion features of the reference positive rehabilitation training motion data, wherein the reference positive rehabilitation training motion features are used for indicating positive feedback motion features of the reference rehabilitation training motion situation;