CN110890146A - Bedside intelligent interaction system for intelligent ward - Google Patents
Bedside intelligent interaction system for intelligent ward Download PDFInfo
- Publication number
- CN110890146A CN110890146A CN201911064287.3A CN201911064287A CN110890146A CN 110890146 A CN110890146 A CN 110890146A CN 201911064287 A CN201911064287 A CN 201911064287A CN 110890146 A CN110890146 A CN 110890146A
- Authority
- CN
- China
- Prior art keywords
- neuron
- data
- neural network
- network model
- bedside
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/20—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the management or administration of healthcare resources or facilities, e.g. managing hospital staff or surgery rooms
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H40/00—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices
- G16H40/60—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices
- G16H40/67—ICT specially adapted for the management or administration of healthcare resources or facilities; ICT specially adapted for the management or operation of medical equipment or devices for the operation of medical equipment or devices for remote operation
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Epidemiology (AREA)
- General Business, Economics & Management (AREA)
- Business, Economics & Management (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Pathology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The utility model provides a bedside intelligence interactive system for wisdom ward, includes bedside digital terminal, nurse station host computer and backstage management system, bedside digital terminal provides human-computer interaction interface, including patient function interface, doctor function interface and nurse function interface, nurse station host computer sets up in the nurse station department of each ward of hospital, including information display interface, information management interface and remote call interface, backstage management system includes backstage login interface, operation monitoring module, information statistics analysis platform, data storage module and doctor workstation. The invention has the beneficial effects that: the bedside intelligent interaction system for the intelligent ward combines the internet technology and medical treatment, realizes information interaction among medical treatment, nursing treatment and patients through the bedside digital terminal, the nurse station host and the background management system, and realizes intellectualization, visualization, precision and high efficiency of medical treatment.
Description
Technical Field
The invention relates to the field of intelligent wards, in particular to a bedside intelligent interaction system for an intelligent ward.
Background
With the rapid development of information technology, more and more hospitals in China are also accelerating to implement the overall construction based on an information-based service platform and system so as to improve the service level and the core competitiveness of the hospitals. At present, China has integrally gone into an intelligent era, and in terms of nursing disciplines, important contents such as continuously improving nursing service quality, gradually implementing hospital nursing post management, strengthening nursing informatization construction and the like are also put forward in a nursing career development planning outline, so that the fact that the nursing career goes into informatization and intellectualization becomes a necessary trend of the era development.
Disclosure of Invention
In view of the above problems, the present invention provides a bedside intelligent interactive system for an intelligent ward.
The purpose of the invention is realized by the following technical scheme:
a bedside intelligent interaction system for an intelligent ward comprises a bedside digital terminal, a nurse station host and a background management system, wherein the bedside digital terminal provides a human-computer interaction interface which comprises a patient function interface, a doctor function interface and a nurse function interface, the patient function interface is used for providing system functions for a patient, the doctor function interface is used for providing system functions for a doctor, the nurse function interface is used for providing system functions for a nurse, the nurse station host is arranged at a nurse station of each ward of a hospital and used for providing the human-computer interaction interface for the nurse, the nurse station host comprises an information display interface, an information management interface and a remote call interface, the information display interface is used for displaying information, the information management interface is used for the nurse to enter and modify the information in the bedside digital terminal, and the nurse can establish video connection with the bedside intelligent terminal through the remote call interface, thereby carry out long-range call to the patient, backstage management system includes backstage login interface, operation monitoring module, information statistics analysis platform, data storage module and doctor's workstation, backstage login interface is used for user input account and password to get into backstage management system, operation monitoring module is used for monitoring bedside digital terminal's running state, information statistics analysis platform is used for gathering bedside digital terminal's data, and goes forward the analysis to the data of gathering, data storage module is used for saving the data that information statistics analysis platform gathered, and the doctor can pass through doctor's workstation inquires the information in the bedside digital terminal, also can establish video connection through doctor's workstation and bedside digital terminal, realizes the remote consultation to the patient.
The beneficial effects created by the invention are as follows: the bedside intelligent interaction system for the intelligent ward combines the internet technology and medical treatment, realizes information interaction among medical treatment, nursing treatment and patients through the bedside digital terminal, the nurse station host and the background management system, and realizes intellectualization, visualization, precision and high efficiency of medical treatment.
Drawings
The invention is further described with the aid of the accompanying drawings, in which, however, the embodiments do not constitute any limitation to the invention, and for a person skilled in the art, without inventive effort, further drawings may be derived from the following figures.
FIG. 1 is a schematic diagram of the present invention.
Reference numerals:
a bedside digital terminal; a nurse station host; and (4) a background management system.
Detailed Description
The invention is further described with reference to the following examples.
Referring to fig. 1, the bedside intelligent interaction system for the intelligent ward of the embodiment includes a bedside digital terminal, a nurse station host and a background management system, the bedside digital terminal provides a human-computer interaction interface, which includes a patient function interface, a doctor function interface and a nurse function interface, the patient function interface is used for providing system functions for patients, the doctor function interface is used for providing system functions for doctors, the nurse function interface is used for providing system functions for nurses, the nurse station host is disposed at a nurse station of each ward of a hospital and is used for providing a human-computer interaction interface for nurses, which includes an information display interface, an information management interface and a remote call interface, the information display interface is used for displaying information, and the information management interface is used for nurses to enter and modify information in the bedside digital terminal, the nurse can establish video connection with the bedside intelligent terminal through a remote call interface so as to remotely call a patient, the background management system comprises a background login interface, an operation monitoring module, an information statistical analysis platform, a data storage module and a doctor workstation, the background login interface is used for inputting an account number and a password by a user so as to enter the background management system, the operation monitoring module is used for monitoring the operation state of the bedside digital terminal, the information statistical analysis platform is used for acquiring the data of the bedside digital terminal and analyzing the acquired data, the data storage module is used for storing the data acquired by the information statistical analysis platform, the doctor can inquire the information in the bedside digital terminal through the doctor workstation and establish video connection with the bedside digital terminal through the doctor workstation, and remote consultation of the patient is realized.
Preferably, the patient function interface is used for providing system functions for the patient, and comprises an intelligent bedside card, self-service business handling and information display, wherein the intelligent bedside card is used for displaying patient information, hospital departments, main doctors, responsibility nurses, nursing levels, attention points and the like, the self-service business handling comprises a self-service inquiry unit, a payment recharging unit, a nutrition ordering unit, an insurance service unit and a leisure service unit, the self-service handling of services such as self-service inquiry, payment recharging, remote calling and leisure service is performed on the patient, and the information display comprises a hospital introduction unit, a health promotion and education unit, an intelligent reminding unit, a satisfaction evaluation unit and a questionnaire investigation unit.
Preferably, doctor function interface is used for providing system function to the doctor, including doctor's advice inquiry unit, record unit, call nurse unit, vital sign inquiry unit and remote consultation unit of making an appointment, doctor's advice inquiry unit is used for the doctor to look over the doctor's advice and the execution conditions, record unit is used for the doctor to record this patient's condition of making an appointment, and when the nurse was not in the ward, the doctor can reach the ward through calling nurse unit real-time call nurse, vital sign inquiry unit is used for the doctor to look over patient's vital sign condition at any time, remote consultation unit realizes the remote consultation of bedside through the doctor workstation of connecting backstage management system.
This preferred embodiment provides a bedside intelligence interactive system for wisdom ward, combines internet technology and medical treatment, through bedside digital terminal, nurse station host computer and backstage management system, has realized the information interaction between doctor, nurse, the sick, has realized intelligent, visual, accurate and the high-efficient of medical treatment.
Preferably, nurse functional interface is used for providing system function to the nurse, including doctor's advice inquiry unit, call doctor's unit, vital sign enter unit and risk assessment unit, doctor's advice inquiry unit is used for nurse's inquiry doctor's advice information, and the nurse can pass through call doctor's unit and doctor workstation establish video connection to visually talkback with the doctor, vital sign enter unit is used for the nurse to look over or record patient vital sign data, risk assessment unit is used for according to the patient vital sign data that vital sign entered the unit and assesses patient's current health status.
Preferably, the risk assessment unit adopts a neural network model to assess the health state of the patient, the neural network model adopted by the risk assessment unit is marked as a second neural network model, the patient vital sign data input by the vital sign input unit is taken as an input value of the second neural network model, and the output result of the output layer of the second neural network model is the health state of the patient obtained through assessment; the method comprises the steps of adopting a collected patient vital sign sample set to train a second neural network model, wherein the data preprocessing part adopts the neural network model to cluster the collected patient vital sign sample set, the neural network model adopted by the data preprocessing part is marked as a first neural network model, and the clustering result of the first neural network model is used as the training sample set of the second neural network model.
In the preferred embodiment, the collected patient vital sign data set is adopted to train the second neural network model adopted by the risk assessment unit, and the vital sign sample set contains a small amount of labeled vital sign data and a large amount of unlabeled vital sign data, so that before the second neural network model is trained, the first neural network model is introduced to carry out semi-supervised clustering preprocessing on the collected patient vital sign data set, and the obtained clustering result and the labels corresponding to the clustering result are adopted to carry out supervised training on the second neural network model, thereby improving the accuracy of the training result of the second neural network model.
Preferably, the first neural network model is used for clustering a collected patient vital sign sample set, and the collected patient vital sign sample set is set as D, where the sample set D includes a labeled sample subset D1And unlabeled exemplar subset D2Given an initial cluster center number c0In the sample subset D1In turn select c0An initial cluster center defining a subset of samples D1The priority of data points competing for the initial clustering center is rho, when the data points are selected as the clustering centers, the data points do not participate in the competition of the next initial clustering center any more, and a sample subset D is defined1Middle data point xiCompeting initial cluster centers v1Has a priority of ρ (x)i,v1) Then ρ (x)i,v1) The calculation formula of (2) is as follows:
in the formula, xj、xmAnd xnAre the data points, x, in the sample set D, respectivelyiAs a subset of samples D1N is the number of data points in the sample set D, N1As a subset of samples D1Number of data points in, v1Selecting a first initial clustering center;
define the remaining data points xlCompeting initial cluster centers vk(k=2,...,c0) Has a priority of ρ (x)l,vk) Then ρ (x)l,vk) The calculation formula of (2) is as follows:
in the formula, xj、xmAnd xnAre the data points, x, in the sample set D, respectivelylAnd xzAs a subset of samples D1N is the number of data points in the sample set D, N1As a subset of samples D1Number of data points in, vkFor the selected k-th initial cluster center, vk-1The selected (k-1) th initial clustering center is obtained;
c to be selected0Inputting data points corresponding to the initial clustering centers into the first neural network model, and inputting c0And taking the data points as initial weights of the connection of hidden layer neurons and input layer neurons of the first neural network model, and taking the labels of the input data points as the labels of the corresponding neurons.
The preferred embodiment defines the priority of selecting the initial clustering center by the data points, selects the data point with the highest priority as the initial clustering center, and the defined priority calculation method comprehensively considers the distance information between the current data point and all the data points in the sample set and the local information of the current data point, so that the global distribution information of the sample space and the local information of the data points are comprehensively utilized, the selected clustering center is more consistent with the distribution condition of the sample data set, and the accuracy of the clustering result is improved.
Preferably, c0After the data points corresponding to the initial clustering centers are input into the first neural network model, sequentially inputting the data points in the sample set D into the first neural network model, and setting xiFor an input data point, at data point xiWhen an already existing neuron is fallen into, calculate data point xiWith the nerve falling intoDistance of element, when data point xiIf the data is label-free data, then selecting a distance data point xiRecent neuron addition; when data point xiWhen the label data exists, a distance data point x is selectediNearest neuron uaIs a data point xiSelecting a distance data point xiSecond proximal neuron ubIs a data point xiWhen the neuron u is a second candidate neuronaAnd ubIs the same, and data point xiOf (3) and neurons uaIs the same, data point xiAdding neuron uaAnd applying the following formula to neurons uaUpdating the weight value:
in the formula (I), the compound is shown in the specification,represents the neuron u at the (t-1) th iterationaThe weight of (a) is calculated,to adjust the posterior neuron uaβ represents the learning rate;
when neuron uaAnd ubIf the labels of (1) are different, the following formula is adopted to treat the neuron uaAnd ubUpdating the weight value:
in the formula, q (x)i,ua) Is the first candidate neuron uaQ (x) ofi,ub) Is the first candidate neuron ubF (x) ofi,ua) To determine the function, at data point xiAnd neuron uaIf the label of (b) is the same, f (x)i,ua) 1, at data point xiAnd neuron uaIf the labels of (1) are different, f (x)i,ua)=0;f(xi,ub) To determine the function, at data point xiAnd neuron ubIf the label of (b) is the same, f (x)i,ub) 1, at data point xiAnd neuron ubIf the labels of (1) are different, f (x)i,ub)=0;
When data point xiIf not in any existing neuron, data point x is assignediAs a new hidden layer neuron, let data point xiAs the initial weight of the neuron, at data point xiIf there is label data, let data point xiAs a label for the neuron; when data point xiWhen there is no label data, the label of the neuron is determined by the following formula:
B(xi)=B(ul)
in the formula, B (x)i) And B (u)l) Respectively represent data points xiAnd neuron ulThe label of (a) is used,andare respectively neuron ulAnd ujN is a distance data point xiThe nearest n neurons, m (u)l) Is neuron ulThe data point in (a) belongs to the sample subset D1Number of data points of, M (u)l) Is neuron ulThe total number of data points.
In the preferred embodiment, after the initial clustering center is determined, the data points in the sample set D are sequentially input into the first neural network model, the weights of the neurons in the hidden layer in the first neural network model are adjusted according to the characteristics of the input data points, the first candidate neuron and the second candidate neuron are selected according to the condition that the data points fall into the existing neurons, a new weight adjustment mode is provided for the condition that the first candidate neuron and the second candidate neuron have label conflict, and a new weight adjustment mode is provided by the data point xiThe distance between the label conflict area and the first candidate neuron is measured according to the distance of the first candidate neuron, so that the adjustment parameter of the weight of the first candidate neuron is determined, and the adjustment parameter of the second candidate neuron is determined in the same way, so that the data point labels in the adjusted neurons have higher uniformity, the sample space can be better adapted, and the accuracy of a clustering result is improved; when the label of the non-label neuron is predicted, the label of the neuron with a short distance is determined, reliable data distribution characteristics are recorded in the neuron, and compared with the traditional mode that the label of the neuron is determined through a single data point, the method has high reliability; in addition, when determining the label of the unlabeled neuron by using the neuron, compared with the conventional method of only considering the distance factor, the preferred embodiment comprehensively introduces the number of labeled data points in the neuron with a short distance, and the neuron has more labeled data, that is, the reliability of the label of the neuron is higher, thereby improving the reliability of label prediction.
Preferably, the clustering result of the first neural network model is used as a training sample set of a second neural network model, the second neural network model adopts a radial basis function neural network and comprises an input layer, a hidden layer and an output layer, the number of neurons of the given hidden layer is S, and the training of the second neural network model is setThe training sample set is H { (c { (C)i,yi) 1, 2.., M }, wherein ciAs the output class of the first neural network model, yiIs of class ciAnd (3) corresponding expected values, wherein M is the number of samples in the sample set H, the samples in the sample set H are input into an input layer of the second neural network model, so that the neurons of the hidden layer of the second neural network model are iteratively trained, the priority g of the samples participating in the neuron training of the hidden layer is defined, and then the samples ciThe priority participating in the training of the s-th neuron of the hidden layer at the t-th iteration isAnd isqt(s) is the connection weight of the s-th neuron of the hidden layer and the output layer in the t-th iteration, and the weight q is obtained by adopting the following formulat(s) and priorityCarrying out iterative updating;
in the formula (I), the compound is shown in the specification,is a normalization factor, and for the s-th neuron of the hidden layer, the sample c is aligned at the t-th iterationiPredicted value of (a), yiIs a sample ciCorrespond toThe expected value of (c) is,represents the number of samples, ρ, involved in training of hidden layer neurons s at the t-th iterationt-1(ci) Represents the sample c up to the (t-1) th iterationiThe total number of times that the neuron is misclassified in all neurons of the hidden layer;
when the sample ciPriority of participation in training of s-th neuron of hidden layer at t-th iterationThen sample ciParticipating in the training of the (t +1) th iteration of the s-th neuron of the hidden layer whenThen sample ciDoes not participate in training of the (t +1) th iteration of the s-th neuron of the hidden layer, where Rt(s) sample control threshold corresponding to the set s-th neuron of the hidden layer at the t-th iteration, and
and stopping the training of the second neural network model when the prediction error values of the hidden layer neurons are all less than 0.3.
In the preferred embodiment, the output set of the first neural network model is used for training the second neural network model, the priority of each sample in the sample set H participating in training hidden layer neurons is defined, the priority of the sample participating in the hidden layer neuron training is iteratively updated, the priority of the sample participating in the neuron training is reduced when the sample is correctly classified by the neurons, the priority of the sample participating in the neuron training is increased when the sample is incorrectly classified by the neurons, a sample control threshold is set for screening the samples participating in the hidden layer neuron training, the samples with higher priorities are selected to participate in the neuron training, and the samples which are difficult to classify for the neurons are used for carrying out re-classification on the neuronsThe point training is carried out, thereby improving the classification performance of the neuron, and in addition, when the priority of the sample is calculated, the preferred embodiment introduces an inhibition item in consideration of the condition of the sample itselfAdjusting the updating of the priority of the sample, counting the times of the sample being classified incorrectly in the neurons of the whole hidden layer, judging that the sample has a problem when the times of the sample being classified incorrectly in the neurons of the whole hidden layer are more, and inhibiting the updating of the priority of the sample, so that the phenomenon that the prediction error rate of the neurons rises when the problem sample is used for repeatedly training the neurons is avoided, the connection weight of the neurons is influenced, and the performance and the weight of each neuron of the hidden layer are more balanced by introducing an inhibition item, so that the accuracy of the evaluation result of the second neural network model is improved; the set sample control threshold value is adaptively changed along with the self condition of the neuron, when the prediction error of the neuron is higher, the value of the neuron sample control threshold value is reduced, so that more samples participate in the training of the neuron, the accuracy of the neuron training result is improved, when the prediction error of the neuron is smaller, the value of the sample control threshold value is increased, the number of the samples participating in the training of the neuron is reduced, and the convergence speed of the second neural network model is improved.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the protection scope of the present invention, although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (7)
1. A bedside intelligent interaction system for a smart ward is characterized by comprising a bedside digital terminal, a nurse station host and a background management system, wherein the bedside digital terminal provides a human-computer interaction interface which comprises a patient function interface, a doctor function interface and a nurse function interface, the patient function interface is used for providing system functions for patients, the doctor function interface is used for providing system functions for doctors, the nurse function interface is used for providing system functions for nurses, the nurse station host is arranged at a nurse station of each ward area of a hospital and comprises an information display interface, an information management interface and a remote call interface, the information display interface is used for displaying information, the information management interface is used for allowing nurses to enter and modify information in the bedside digital terminal, and the nurses can establish video connection through the remote call interface and the bedside intelligent terminal, thereby carry out long-range call to the patient, backstage management system includes backstage login interface, operation monitoring module, information statistics analysis platform, data storage module and doctor's workstation, backstage login interface is used for user input account and password to get into backstage management system, operation monitoring module is used for monitoring bedside digital terminal's running state, information statistics analysis platform is used for gathering bedside digital terminal's data, and goes forward the analysis to the data of gathering, data storage module is used for saving the data that information statistics analysis platform gathered, and the doctor can pass through doctor's workstation inquires the information in the bedside digital terminal, also can establish video connection through doctor's workstation and bedside digital terminal, realizes the remote consultation to the patient.
2. The bedside intelligent interactive system for intelligent wards as claimed in claim 1, wherein the bedside digital terminal, the nurse station host and the background management system are connected through an IP network or a WiFi network.
3. The bedside intelligent interactive system for the intelligent ward of claim 2, wherein the nurse functional interface comprises an order inquiry unit, a doctor calling unit, a vital sign entry unit and a risk assessment unit, the order inquiry unit is used for nurses to inquire order information, nurses can establish video connection with a doctor workstation by calling the doctor calling unit to realize visual talkback with doctors, the vital sign entry unit is used for nurses to check or record vital sign data of patients, and the risk assessment unit is used for assessing the current physical health status of patients according to the vital sign data of the patients entered by the vital sign entry unit.
4. The bedside intelligent interactive system for an intelligent ward according to claim 3, wherein the risk assessment unit assesses the health state of a patient using a neural network model, the neural network model used by the risk assessment unit is recorded as a second neural network model, patient vital sign data recorded by a vital sign recording unit is used as an input value of the second neural network model, and an output result of an output layer of the second neural network model is the assessed health state of the patient; the method comprises the steps of adopting a collected patient vital sign sample set to train a second neural network model, wherein the data preprocessing part adopts the neural network model to cluster the collected patient vital sign sample set, the neural network model adopted by the data preprocessing part is marked as a first neural network model, and the clustering result of the first neural network model is used as the training sample set of the second neural network model.
5. The bedside intelligent interactive system for intelligent wards, as claimed in claim 4, wherein the first neural network model is used for clustering the collected patient vital sign sample sets, and the collected patient vital sign sample sets are set asThe sample setIncluding a subset of labeled exemplarsAnd unlabeled exemplar subsetsGiven initial cluster center numberIn the sample subsetIn turn selectAn initial cluster center defining a subset of samplesThe priority of data points competing for the initial cluster center in (1) isWhen the data point is selected as the clustering center, the data point does not participate in the competition of the next initial clustering center any more, and the sample subset is definedData point inCompeting initial cluster centersHas a priority ofThen, thenThe calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,、andare respectively a sample setThe data point in (a) is,as a subset of samplesThe data point in (a) is,as a sample setThe number of data points in (1),as a subset of samplesThe number of data points in (1),to be selectedA first initial cluster center;
defining remaining data pointsCompeting initial cluster centersHas a priority ofThen, thenThe calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,、andare respectively a sample setThe data point in (a) is,andis a sampleSubsetsThe data point in (a) is,as a sample setThe number of data points in (1),as a subset of samplesThe number of data points in (1),to a selected oneThe center of each initial cluster is determined,to a selected oneAn initial clustering center;
to be selectedInputting data points corresponding to the initial clustering centers into the first neural network model, and inputtingThe data points are used as initial weight values of the connection of hidden layer neurons and input layer neurons of the first neural network model, and the labels of the input data points are used as corresponding nervesA meta tag.
6. The bedside intelligent interactive system for intelligent wards of claim 5, wherein,inputting data points corresponding to the initial clustering centers into the first neural network model, and then collecting the sample setThe data points in (1) are sequentially input into the first neural network modelFor an input data point, when the data pointWhen an existing neuron is fallen into, a data point is calculatedDistance from the neuron that falls in, as data pointsWhen the data is label-free data, selecting a distance data pointRecent neuron addition; when the data point isSelecting distance data points when the data is labeledNearest neuronAre data pointsSelecting distance data pointsSecond proximal neuronAre data pointsA second candidate neuron of (1), when the neuron isAndare identical, and data pointsOf (3) and neuronsIs the same, data pointAdding neuronsAnd applying the following formula to neuronsUpdating the weight value:
in the formula (I), the compound is shown in the specification,is shown asSub-iterative time neuronThe weight of (a) is calculated,for regulating posterior neuronThe weight of (a) is calculated,represents a learning rate;
when neuronAndif the labels are different, the neuron is treated by the following formulaAndupdating the weight value:
in the formula (I), the compound is shown in the specification,to determine the function, when the data pointAnd neuronsWhen the labels are the same, thenWhen a data pointAnd neuronsWhen the labels are different, then;To determine the function, when the data pointAnd neuronsWhen the labels are the same, thenWhen a data pointAnd neuronsWhen the labels are different, then;
When the data point isIf not in any existing neuron, the data point is assignedAs a new hidden layer neuron, let the data pointAs an initial weight for the neuron, at the data pointWhen the data is tagged, the data point is orderedAs a label for the neuron; when the data point isWhen there is no label data, the label of the neuron is determined by the following formula:
in the formula (I), the compound is shown in the specification,andrespectively represent data pointsAnd neuronsThe label of (a) is used,andare neurons respectivelyAndthe weight of (a) is calculated,as distance data pointsMore recentThe number of the nerve cells is one,is a neuronThe data point in (1) belongs to the sample subsetThe number of data points of (a),is a neuronThe total number of data points.
7. The bedside intelligent interactive system for intelligent ward of claim 6, wherein the clustering result of the first neural network model is used as a training sample set of the second neural network model, the second neural network model is a radial basis neural network, and comprises an input layer, an implicit layer and an output layer, the number of neurons in the given implicit layer isThe set of training samples for the second neural network model isWherein, in the step (A),for the output class of the first neural network model,is a categoryThe corresponding desired value is set to the desired value,as a set of samplesNumber of samples in (1), set the samples into a setThe samples in (1) are input into an input layer of a second neural network model, so that the neurons of the hidden layer of the second neural network model are subjected to iterative training, and the priority of the samples participating in the training of the neurons of the hidden layer is definedThen sampleIn the first placeParticipation in hidden layer number one at sub-iterationThe training of each neuron has priority ofAnd is and,is at the firstImplicit layer number at sub-iterationThe connection weight of each neuron and the output layer adopts the following formula to the weightAnd priorityCarrying out iterative updating;
in the formula (I), the compound is shown in the specification,is a normalization factor, and,as a hidden layerOne neuron is in the first placePair samples at sub-iterationThe predicted value of (a) is determined,is a sampleThe corresponding desired value is set to the desired value,is shown asInvolvement of hidden layer neurons at sub-iterationThe number of samples to be trained is,is shown toSample to sub-iterationThe total number of times that the neuron is misclassified in all neurons of the hidden layer;
when the sample isIn the first placeParticipation in hidden layer number one at sub-iterationPriority of individual neuron trainingThen sampleParticipating in the hidden layer oneThe first of each neuronTraining in a sub-iteration whenThen sampleDoes not participate in the hidden layerThe first of each neuronAnd (c) training of a sub-iteration, wherein,for setting the hidden layerOne neuron is in the first placeThe corresponding sample at the time of the sub-iteration controls the threshold, an;
And stopping the training of the second neural network model when the prediction error values of the hidden layer neurons are all less than 0.3.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911064287.3A CN110890146B (en) | 2019-11-04 | 2019-11-04 | Bedside intelligent interaction system for intelligent ward |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911064287.3A CN110890146B (en) | 2019-11-04 | 2019-11-04 | Bedside intelligent interaction system for intelligent ward |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110890146A true CN110890146A (en) | 2020-03-17 |
CN110890146B CN110890146B (en) | 2020-07-14 |
Family
ID=69746787
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911064287.3A Active CN110890146B (en) | 2019-11-04 | 2019-11-04 | Bedside intelligent interaction system for intelligent ward |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110890146B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112542253A (en) * | 2020-12-11 | 2021-03-23 | 聊城大学 | Intelligent medical care patient interaction system |
CN112669995A (en) * | 2020-12-11 | 2021-04-16 | 边缘智能研究院南京有限公司 | Intelligent medical management system and method |
CN112837799A (en) * | 2021-01-29 | 2021-05-25 | 广东德澳智慧医疗科技有限公司 | Remote internet big data intelligent medical system based on block chain |
CN112860800A (en) * | 2021-02-22 | 2021-05-28 | 深圳市星网储区块链有限公司 | Trusted network application method and device based on block chain and federal learning |
CN113284595A (en) * | 2021-06-16 | 2021-08-20 | 广东德澳智慧医疗科技有限公司 | Intelligent scheduling and patient reservation management system for hospital and clinic operators |
CN114519176A (en) * | 2020-11-18 | 2022-05-20 | 京东方科技集团股份有限公司 | Interaction method, electronic device and storage medium |
CN115473925A (en) * | 2022-11-02 | 2022-12-13 | 四川港通医疗设备集团股份有限公司 | Intelligent medical call management method and system based on cloud computing |
CN117275699B (en) * | 2023-11-23 | 2024-02-13 | 四川省医学科学院·四川省人民医院 | Wisdom ward system |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102592063A (en) * | 2012-03-25 | 2012-07-18 | 河北普康医疗设备有限公司 | Digitalized information management system for nursing stations in hospitals and method for realizing same |
CN106202922A (en) * | 2016-07-06 | 2016-12-07 | 吴本刚 | A kind of transformer fault diagnosis system based on clustering algorithm |
CN108416392A (en) * | 2018-03-16 | 2018-08-17 | 电子科技大学成都研究院 | Building clustering method based on SOM neural networks |
CN108597561A (en) * | 2018-02-01 | 2018-09-28 | 南京滕健医疗科技有限公司 | Wisdom ward interactive system |
CN109472490A (en) * | 2018-11-06 | 2019-03-15 | 北京京航计算通讯研究所 | Military project group personal information labeling system based on cluster |
EP3562047A1 (en) * | 2018-04-27 | 2019-10-30 | Nokia Solutions and Networks Oy | Method of training a deep neural network classifier, a training module and a network analyzer |
-
2019
- 2019-11-04 CN CN201911064287.3A patent/CN110890146B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102592063A (en) * | 2012-03-25 | 2012-07-18 | 河北普康医疗设备有限公司 | Digitalized information management system for nursing stations in hospitals and method for realizing same |
CN106202922A (en) * | 2016-07-06 | 2016-12-07 | 吴本刚 | A kind of transformer fault diagnosis system based on clustering algorithm |
CN108597561A (en) * | 2018-02-01 | 2018-09-28 | 南京滕健医疗科技有限公司 | Wisdom ward interactive system |
CN108416392A (en) * | 2018-03-16 | 2018-08-17 | 电子科技大学成都研究院 | Building clustering method based on SOM neural networks |
EP3562047A1 (en) * | 2018-04-27 | 2019-10-30 | Nokia Solutions and Networks Oy | Method of training a deep neural network classifier, a training module and a network analyzer |
CN109472490A (en) * | 2018-11-06 | 2019-03-15 | 北京京航计算通讯研究所 | Military project group personal information labeling system based on cluster |
Non-Patent Citations (1)
Title |
---|
王震洲 等: "基于BP神经网络的pH试纸自动检测系统", 《电子测量与仪器学报》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114519176A (en) * | 2020-11-18 | 2022-05-20 | 京东方科技集团股份有限公司 | Interaction method, electronic device and storage medium |
WO2022120873A1 (en) * | 2020-12-11 | 2022-06-16 | 边缘智能研究院南京有限公司 | Intelligent medical management system and method |
CN112669995A (en) * | 2020-12-11 | 2021-04-16 | 边缘智能研究院南京有限公司 | Intelligent medical management system and method |
CN112542253A (en) * | 2020-12-11 | 2021-03-23 | 聊城大学 | Intelligent medical care patient interaction system |
CN112837799A (en) * | 2021-01-29 | 2021-05-25 | 广东德澳智慧医疗科技有限公司 | Remote internet big data intelligent medical system based on block chain |
CN113889252B (en) * | 2021-01-29 | 2023-04-11 | 丁跃辉 | Remote internet big data intelligent medical system based on vital sign big data clustering core algorithm and block chain |
CN112837799B (en) * | 2021-01-29 | 2021-10-15 | 广东德澳智慧医疗科技有限公司 | Remote internet big data intelligent medical system based on block chain |
CN113889252A (en) * | 2021-01-29 | 2022-01-04 | 广东德澳智慧医疗科技有限公司 | Remote internet big data intelligent medical system based on vital sign big data clustering core algorithm and block chain |
CN112860800A (en) * | 2021-02-22 | 2021-05-28 | 深圳市星网储区块链有限公司 | Trusted network application method and device based on block chain and federal learning |
CN113284595A (en) * | 2021-06-16 | 2021-08-20 | 广东德澳智慧医疗科技有限公司 | Intelligent scheduling and patient reservation management system for hospital and clinic operators |
CN115473925A (en) * | 2022-11-02 | 2022-12-13 | 四川港通医疗设备集团股份有限公司 | Intelligent medical call management method and system based on cloud computing |
CN115473925B (en) * | 2022-11-02 | 2023-02-03 | 四川港通医疗设备集团股份有限公司 | Intelligent medical call management method and system based on cloud computing |
CN117275699B (en) * | 2023-11-23 | 2024-02-13 | 四川省医学科学院·四川省人民医院 | Wisdom ward system |
Also Published As
Publication number | Publication date |
---|---|
CN110890146B (en) | 2020-07-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110890146B (en) | Bedside intelligent interaction system for intelligent ward | |
US20200265956A1 (en) | Machine learning clinical decision support system for risk categorization | |
CN109949936B (en) | Re-hospitalization risk prediction method based on deep learning mixed model | |
US9861308B2 (en) | Method and system for monitoring stress conditions | |
US20210375392A1 (en) | Machine learning platform for generating risk models | |
JP2021525411A (en) | Personalized device recommendations for proactive health monitoring and management | |
Morid et al. | Learning hidden patterns from patient multivariate time series data using convolutional neural networks: A case study of healthcare cost prediction | |
CN112992370B (en) | Unsupervised electronic medical record-based medical behavior compliance assessment method | |
WO2017079012A1 (en) | Medical protocol evaluation | |
Bhadra et al. | Enhancing the efficacy of depression detection system using optimal feature selection from EHR | |
CN117497149A (en) | Hospital management method and system based on artificial intelligence | |
El-Darzi et al. | Length of stay-based clustering methods for patient grouping | |
US20230068453A1 (en) | Methods and systems for determining and displaying dynamic patient readmission risk and intervention recommendation | |
US20230196116A1 (en) | Machine learning for classification of users | |
Khademolqorani et al. | Development of a decision support system for handling health insurance deduction | |
US11145401B1 (en) | Systems and methods for generating a sustenance plan for managing genetic disorders | |
AU2021102593A4 (en) | A Method for Detection of a Disease | |
US11544626B2 (en) | Methods and systems for classifying resources to niche models | |
US11289206B2 (en) | Artificial intelligence methods and systems for constitutional analysis using objective functions | |
US20220027783A1 (en) | Method of and system for generating a stress balance instruction set for a user | |
US11275595B1 (en) | System and method for programming a monitoring device | |
US11416776B2 (en) | Method of and system for identifying and enumerating cross-body degradations | |
Klemm et al. | Interactive visual analysis of lumbar back pain-what the lumbar spine tells about your life | |
US11887720B1 (en) | Apparatus and method for using a feedback loop to optimize meals | |
US12009085B2 (en) | Systems and methods for scheduling alimentary combinations |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |