CN117809849A - Analysis method and system for walking postures of old people with cognitive dysfunction - Google Patents
Analysis method and system for walking postures of old people with cognitive dysfunction Download PDFInfo
- Publication number
- CN117809849A CN117809849A CN202410229811.2A CN202410229811A CN117809849A CN 117809849 A CN117809849 A CN 117809849A CN 202410229811 A CN202410229811 A CN 202410229811A CN 117809849 A CN117809849 A CN 117809849A
- Authority
- CN
- China
- Prior art keywords
- thread
- information
- fragment
- original
- evaluation
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 27
- 208000010877 cognitive disease Diseases 0.000 title claims abstract description 13
- 230000036544 posture Effects 0.000 title claims description 64
- 238000011156 evaluation Methods 0.000 claims abstract description 144
- 239000012634 fragment Substances 0.000 claims abstract description 141
- 238000004364 calculation method Methods 0.000 claims abstract description 70
- 238000000034 method Methods 0.000 claims abstract description 43
- 238000012545 processing Methods 0.000 claims abstract description 35
- 230000010354 integration Effects 0.000 claims description 32
- 230000006870 function Effects 0.000 claims description 27
- 230000002159 abnormal effect Effects 0.000 claims description 11
- 230000003993 interaction Effects 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 5
- 238000012804 iterative process Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 238000003491 array Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
According to the analysis method and system for walking gestures of the aged with cognitive dysfunction, continuous type example fragments are projected into the possibility space based on the target calculation thread to obtain the possibility of the example fragments, the example fragments can be scaled into the value range of the possibility while the distribution information of the sizes of the example fragments is maintained, and then when the information evaluation thread is configured by using the example fragment possibility of the example information and the example prediction fragment possibility, errors in the evaluation index processing process can be reduced, and the accuracy and the robustness of the information evaluation thread obtained by configuration are improved, so that the analysis result can be obtained more accurately by using the information evaluation thread.
Description
Technical Field
The application relates to the technical field of data analysis, in particular to a method and a system for analyzing walking postures of old people with cognitive dysfunction.
Background
At present, detection in various aspects and diagnosis of doctors are required for judging the walking postures of the old with cognitive dysfunction, so that the workload of the doctors is huge, and the walking postures of the old with cognitive dysfunction are analyzed through artificial intelligence in order to reduce the workload of the doctors, but how to analyze the walking postures of the old with cognitive dysfunction is a technical problem which is difficult to solve at present.
Disclosure of Invention
In order to improve the technical problems in the related art, the application provides an analysis method and an analysis system for walking postures of the aged with cognitive dysfunction.
In a first aspect, a method for analyzing walking postures of old people with cognitive dysfunction is provided, including:
for each of a plurality of pieces of walking gesture information of the old people to be determined, obtaining a target evaluation coefficient of each piece of walking gesture information of the old people to be determined through regression evaluation of an information evaluation thread configured in advance; the information evaluation thread is configured to obtain the information evaluation thread to be configured based on the example fragment possibility and the example prediction fragment possibility of the example information; the example prediction fragment possibility of the example information is obtained by carrying out regression evaluation on the example information through the information evaluation thread to be configured; the example fragment possibility is obtained by projecting an example fragment of the example information by adopting a target calculation thread built in advance;
distributing the walking gesture information of the plurality of the pending old people based on target evaluation coefficients of the walking gesture information of each pending old people to obtain the walking gesture information distribution situation of the pending old people;
and screening the walking posture information of a specified number of the pending old people from the walking posture information distribution condition of the pending old people, determining the walking posture information as walking abnormal data, and analyzing the walking abnormal data to obtain an analysis result.
In the present application, the target evaluation coefficients include predicted segment likelihood under a segment level and at least one residual index evaluation coefficient under a residual interaction level, and the distribution of the walking posture information of the plurality of pending old people based on the target evaluation coefficients of the walking posture information of each of the pending old people to obtain a distribution situation of the walking posture information of the pending old people includes:
acquiring a first confidence coefficient under the segment level and a second confidence coefficient under the residual interaction level;
for any walking posture information of the old people to be determined, performing first function calculation on the predicted fragment possibility of the walking posture information of the old people to be determined based on the first confidence coefficient to obtain a first calculated value, and performing first function calculation on the residual index evaluation coefficient of the walking posture information of the old people to be determined based on the second confidence coefficient to obtain a second calculated value;
processing a second function of the first calculated value and the second calculated value of each piece of walking gesture information of the old to be determined to be a splicing evaluation coefficient of the walking gesture information of the old to be determined;
and carrying out simplified distribution on the walking posture information of the plurality of the old people to be determined according to the splicing evaluation coefficients to obtain the walking posture information distribution situation of the old people to be determined.
In the present application, the method further comprises:
configuring the information assessment thread by: acquiring an example fragment set corresponding to the example information set; the set of example snippets includes example snippets of each of the example information in the set of example information; projecting the example fragments of each example information into corresponding example fragment possibilities by adopting a target calculation thread built in advance to obtain an example fragment possibility set; wherein the distribution of each of the example segment likelihoods in the example segment likelihoods set is linked to the distribution of the example segment in the example segment set;
acquiring the possibility of an example prediction fragment of each piece of example information output by the information evaluation thread to be configured;
performing evaluation index processing on the information evaluation thread to be configured based on the example fragment probability of the example information and the example prediction fragment probability to obtain an evaluation index processing result;
and optimizing the thread coefficient in the information evaluation thread to be configured based on the evaluation index processing result until the information evaluation thread to be configured converges, so as to obtain the configured information evaluation thread.
In the present application, the method further comprises:
acquiring an original calculation thread built in advance; determining not less than three target values and weights corresponding to the target values from a plurality of sample fragments;
projecting each target value into a corresponding original fragment possibility by adopting the original calculation thread;
determining function coefficients in the original calculation thread according to the weights corresponding to the target values and the original fragment probability;
and building the target computing thread according to the function coefficients and the original computing thread.
In this application, the original computing thread and the target computing thread both meet the following conditions:
when the example segment is 0, the possibility of the original segment obtained by projecting the example segment through the original computing thread is 0, and the possibility of the example segment obtained by projecting the example segment through the target computing thread is 0;
when the example segment is positive, the possibility of the original segment obtained by projecting the example segment through the original computing thread is 1, and the possibility of the example segment obtained by projecting the example segment through the target computing thread is 1; the original segment likelihood and the example segment likelihood are each in the range of 0 to 1; the original segment likelihood and the example segment likelihood monotonically increase with respect to the example segment.
In the present application, determining not less than three target values from the plurality of example segments, and weights corresponding to the target values includes:
distributing the plurality of example fragments according to the sequence from big to small to form an example fragment distribution condition;
determining the corresponding positioning of each sample fragment in the sample fragment distribution situation; determining not less than three fragment positioning percentages based on a preset target value number;
and determining not less than three target values from the plurality of sample fragments according to the fragment positioning percentages and the positioning corresponding to each sample fragment, and determining the fragment positioning percentage corresponding to the target value as the weight corresponding to the target value.
In this application, the determining, according to the weights corresponding to the target values and the original segment likelihoods, function coefficients in the original computing thread includes:
performing the following integration iterative process on the original computing thread: acquiring the possibility of an original fragment obtained by projecting the target value by the original calculation thread;
performing evaluation index processing on the original calculation thread according to the weight corresponding to the target value and the original fragment probability to obtain an integration result of the original calculation thread;
optimizing coefficients in the original calculation thread based on the integration result to obtain an optimized calculation thread;
when the optimized computing thread does not accord with a preset integration condition, determining the optimized computing thread as an original computing thread in the next integration iteration process;
stopping the integration iteration process when the optimized calculation thread accords with a preset integration condition;
and determining coefficients in a calculation thread when the integration iteration process is stopped as the function coefficients.
In this application, the performing evaluation index processing on the original computing thread according to the weight corresponding to the target value and the original segment likelihood to obtain an integrated result of the original computing thread includes:
building a target cost index strategy based on the original calculation thread;
substituting the weight corresponding to the target value and the original fragment probability into the target cost index strategy to obtain the integration result corresponding to the target value.
In this application, the building a target cost indicator policy based on the original computing thread includes:
determining a comparison between the weights in the target values and the original computing thread;
respectively averaging the comparison results corresponding to the target values to obtain a plurality of average values;
and fusing the average values to obtain the target cost index strategy.
In a second aspect, an analysis system for walking postures of cognitively dysfunctional elderly people is provided, comprising a processor and a memory in communication with each other, the processor being configured to read a computer program from the memory and execute the computer program to implement the method described above.
According to the analysis method and the analysis system for the walking postures of the aged with cognitive dysfunction, when the analysis is performed, the target evaluation coefficients of the walking posture information of all the aged to be determined can be obtained through regression evaluation of the information evaluation threads configured in advance, then the walking posture information of a plurality of aged to be determined is distributed based on the target evaluation coefficients of the walking posture information of all the aged to be determined, the walking posture information distribution situation of the aged to be determined is obtained, the walking abnormal data are determined by screening the walking posture information of a specified number of aged to be determined from the walking posture information distribution situation of the aged to be determined, and the walking abnormal data are analyzed, so that an analysis result is obtained. The information evaluation thread to be configured can be configured based on the example fragment probability of the example information and the example prediction fragment probability, wherein the example prediction fragment probability is obtained by carrying out regression evaluation on the example information through the information evaluation thread to be configured, and the example fragment probability is obtained by adopting a target calculation thread built in advance to project the example fragment of the example information. According to the method and the device for processing the sample fragments, continuous sample fragments are projected into a possibility space based on a target computing thread to obtain the sample fragment possibility, the sample fragments can be scaled into a value range of the possibility while distribution information of the sample fragment sizes is maintained, and then when the sample fragment possibility of sample information and sample prediction fragment possibility configuration information are utilized to evaluate the thread, errors in an evaluation index processing process can be reduced, and accuracy and robustness of the information evaluation thread obtained through configuration are improved, so that the accuracy of analysis can be improved by using the information evaluation thread, and an analysis result can be obtained more accurately.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered limiting the scope, and that other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of an analysis method for walking postures of the aged with cognitive dysfunction according to an embodiment of the present application.
Detailed Description
In order to better understand the technical solutions described above, the following detailed description of the technical solutions of the present application is provided through the accompanying drawings and specific embodiments, and it should be understood that the specific features of the embodiments and embodiments of the present application are detailed descriptions of the technical solutions of the present application, and not limit the technical solutions of the present application, and the technical features of the embodiments and embodiments of the present application may be combined with each other without conflict.
Referring to fig. 1, a method for analyzing walking postures of the elderly with cognitive dysfunction is shown, and the method may include the following technical solutions described in step S101 to step S103.
Step S101, for each piece of walking gesture information of the plurality of pieces of walking gesture information of the old, obtaining a target evaluation coefficient of the walking gesture information of each piece of old through regression evaluation of an information evaluation thread configured in advance.
The information evaluation thread is configured based on the example fragment probability and the example prediction fragment probability of the example information. The example prediction segment possibility of the example information is obtained by carrying out regression evaluation on the example information through an information evaluation thread to be configured. The example fragment possibility is obtained by projecting an example fragment of the example information by using a target calculation thread built in advance. Wherein, the example segment may be understood as a sample time, and the example information may be understood as sample information.
Step S102, based on target evaluation coefficients of the walking gesture information of each of the pending old people, distributing the walking gesture information of a plurality of the pending old people to obtain the walking gesture information distribution situation of the pending old people.
In the embodiment of the application, after determining the target evaluation coefficients of the walking gesture information of each pending old person, the walking gesture information of a plurality of pending old persons can be ascending or simplified distributed based on the target evaluation coefficients, so as to obtain the walking gesture information distribution situation of the pending old persons.
In some possible embodiments, step S102 may be implemented by the following steps S1021 to S1024:
the target evaluation coefficients include predicted segment likelihoods at the segment level and not less than one remaining index evaluation coefficient at the remaining interaction level.
Step S1021, obtaining a first confidence level under the segment level and a second confidence level under the residual interaction level.
In some possible embodiments, the information evaluation thread may also be a multi-objective evaluation thread, through which the objective evaluation coefficients of the walking posture information of each pending elderly person under several interaction levels can be evaluated by regression. The target evaluation coefficients include predicted segment likelihoods at the segment level and not less than one remaining index evaluation coefficient at the remaining interaction level.
Step S1022, for any walking posture information of the old to be determined, performing a first function calculation on the predicted segment possibility of the walking posture information of the old to be determined based on the first confidence coefficient to obtain a first calculated value, and performing a first function calculation on the residual index evaluation coefficient of the walking posture information of the old to be determined based on the second confidence coefficient to obtain a second calculated value.
In this embodiment, after obtaining the predicted segment likelihood of the walking posture information of any one of the to-be-determined old people, the first confidence coefficient may be used as an index to perform a first confidence coefficient power operation of the predicted segment likelihood to obtain a first calculated value. For any walking posture information of the old people to be determined, after the residual index evaluation coefficient of the walking posture information of the old people to be determined is obtained, a second confidence coefficient corresponding to one index evaluation coefficient is used as an index, and a second confidence coefficient power operation of the index evaluation coefficient is performed to obtain a second calculated value.
Step S1023, processing the second function of the first calculated value and the second calculated value of the walking gesture information of each old person to be determined as a splicing evaluation coefficient of the walking gesture information of the old person to be determined.
Step S1024, the walking gesture information of the plurality of the old people to be determined is simplified and distributed according to the splicing evaluation coefficients, and the walking gesture information distribution situation of the old people to be determined is obtained.
In the embodiment of the application, after the splicing evaluation coefficients of the walking posture information of each undetermined old person are obtained, the walking posture information of a plurality of undetermined old persons can be simplified and distributed according to the size of the splicing evaluation coefficients, so that the walking posture information distribution situation of the undetermined old persons is obtained.
In the embodiment of the application, when the information evaluation thread is used for carrying out online distribution on the walking gesture information of the old people to be determined, if the target evaluation coefficient predicted by the information evaluation thread comprises the predicted segment possibility under the segment level and the residual index evaluation coefficient under the residual interaction level, for each of the walking gesture information of the old people to be determined, the predicted segment possibility can be multiplied into the index evaluation coefficient in a power-of-power mode to obtain a new splicing evaluation coefficient, and then the walking gesture information of all the old people to be determined is distributed from large to small according to the splicing evaluation coefficient, so that the walking gesture information of a certain old people to be determined to be arranged in front can be determined as a recommendation result returned to a user. According to the embodiment of the application, the recommendation result is considered by integrating the possibility of predicting the fragments, and the recommendation accuracy of analysis is improved.
Step S103, screening the walking posture information of the appointed number of the undetermined old people from the walking posture information distribution situation of the undetermined old people, determining the walking posture information as walking abnormal data, and analyzing the walking abnormal data to obtain an analysis result.
Here, the walking posture information of the specified number of the pending old people can be screened out from the first walking posture information of the pending old people in the distribution situation of the walking posture information of the pending old people, and the walking abnormal data can be determined.
According to the method and the device for processing the sample fragments, continuous sample fragments are projected into a possibility space based on a target computing thread to obtain the sample fragment possibility, the sample fragments can be scaled into a value range of the possibility while distribution information of the sample fragment sizes is maintained, and then when the sample fragment possibility of sample information and sample prediction fragment possibility configuration information are utilized to evaluate the thread, errors in an evaluation index processing process can be reduced, and accuracy and robustness of the information evaluation thread obtained through configuration are improved, so that the accuracy of analysis can be improved by using the information evaluation thread, and an analysis result can be obtained more accurately.
Step S201, an example clip set corresponding to the example information set is obtained.
The set of instance segments includes instance segments of respective instance information in the set of instance information.
Step S202, projecting the example fragments of each example information into corresponding example fragment possibilities by using a target calculation thread built in advance, thereby obtaining an example fragment possibility set.
Wherein the distribution of the respective example segment likelihoods in the example segment likelihoods set is associated with the distribution of the example segment in the example segment set.
In the embodiment of the application, the target calculation thread is a distribution function which is built in advance and aims at the fragments. The target computing thread may project any fragment as a fragment likelihood within the [0,1] interval. For example segments of arbitrary example information, the target computing thread may project the example segment as example segment likelihoods within the [0,1] interval. The example segment likelihoods obtained after the example segment projection of each example information may be built as an example segment likelihood set. The target computing thread monotonically increases with respect to the segments, and thus, the distribution of the respective example segment likelihoods in the example segment likelihood set is linked to the distribution of the example segments in the example segment set. That is, the larger the example fragment, the greater the likelihood that the example fragment will be obtained after projection through the target computing thread.
In some possible embodiments, the construction process of the target computing thread in step S202 may be implemented by the following steps S301 to S305:
step S301, an original calculation thread built in advance is obtained.
In step S302, at least three target values and weights corresponding to the target values are determined from the plurality of sample segments.
In some possible embodiments, step S302 may be implemented by the following steps S3021 to S3024:
in step S3021, a plurality of example segments are distributed in order from large to small, so as to form an example segment distribution situation.
In the embodiment of the application, a plurality of example segments can be arranged from large to small according to the values of the example segments to obtain the distribution situation of the example segments.
In step S3022, the corresponding locations of the respective example clips in the example clip distribution are determined.
In the embodiment of the application, after the example fragment distribution situation is obtained, the location of each example fragment arranged in the example fragment distribution situation can be determined.
Step S3023, determining not less than three segment positioning percentages based on the preset target value number.
In step S3024, according to the segment positioning percentages and the positioning corresponding to each exemplary segment, at least three target values are determined from the plurality of exemplary segments, and the segment positioning percentage corresponding to the target value is determined as the weight corresponding to the target value.
In the embodiment of the application, by determining the corresponding positioning of each sample fragment in the sample fragment distribution situation, the target value can be determined from a plurality of sample fragments according to the fragment positioning percentage and the corresponding positioning of each sample fragment, so that the target calculation thread can be built based on the actual corresponding weight of the target value and the original fragment possibility obtained by projecting the target value by the original calculation thread, the target calculation thread can project the sample fragment as the sample fragment possibility basically consistent with the actual target value, and the projection accuracy of the target calculation thread is improved.
In step S303, each target value is projected as a corresponding original segment likelihood using the original calculation thread.
Step S304, determining the function coefficients in the original calculation thread according to the weights corresponding to the target values and the original fragment probability.
In some possible embodiments, step S304 may be implemented by the following steps S3041 to S3045:
the following integration iterative process is performed on the original computing thread:
step S3041, obtaining an original segment probability obtained by projecting the target value by the original calculation thread.
And step S3042, performing evaluation index processing on the original calculation thread according to the weight corresponding to the target value and the original fragment probability to obtain an integration result of the original calculation thread.
In the embodiment of the application, evaluation index processing can be performed based on the weight corresponding to each target value, the possibility of the original segment and the function coefficient in the current original calculation thread, so as to obtain the integration result of the original calculation thread.
In some possible embodiments, step S3042 may be implemented by the following steps S30421 and S30422:
and step S30421, constructing a target cost index strategy based on the original calculation thread.
In some possible embodiments, step S30421 may be implemented by: a comparison between the weights in the target values and the original computing thread is determined. And respectively averaging comparison results corresponding to the target values to obtain a plurality of average values. And fusing the average values to obtain a target cost index strategy.
In the embodiment of the present application, the target cost indicator policy is an average cost indicator policy.
And S30422, substituting the weight corresponding to the target value and the original fragment probability into a target cost index strategy to obtain an integration result corresponding to the target value.
And step S3043, optimizing coefficients in the original calculation thread based on the integration result to obtain an optimized calculation thread.
In step S3044, when the optimized computing thread does not meet the preset integration condition, the optimized computing thread is determined as the original computing thread in the next integration iteration process.
In some possible embodiments, the preset integration condition may be a preset number of iterations, and after the preset number of iterations has been reached, the iteration process is terminated, and the coefficient obtained by optimization is determined as a function coefficient, so as to obtain the target computing thread.
Step S3045, stopping the integration iteration process when the optimized calculation thread accords with a preset integration condition; and determining coefficients in the calculation thread when the integration iteration process is stopped as function coefficients.
In the embodiment of the application, when the integration result obtained by calculation of the optimized calculation thread is the minimum value, the integration iteration process is stopped, and the coefficient in the calculation thread at this time is determined as the function coefficient.
And step S305, building a target calculation thread according to the function coefficients and the original calculation thread.
Here, after determining the function coefficient when the integration iterative process is terminated, the function coefficient is replaced with the coefficient in the original calculation thread, so as to obtain the target calculation thread.
In the embodiment of the application, the target value is determined from a plurality of sample fragments, and based on the weight actually corresponding to the target value and the original fragment possibility obtained by projecting the target value by the original calculation thread, the function coefficient in the original calculation thread is determined, so that the target calculation thread built according to the function coefficient and the original calculation thread can project the sample fragment as the sample fragment possibility basically consistent with the actual target value, the projection accuracy of the target calculation thread is improved, the accuracy of configuration input data can be improved in the subsequent iterative configuration of the information evaluation thread, and the recommendation accuracy of the information evaluation thread obtained by final configuration is improved.
In step S203, the likelihood of the sample prediction segment of each sample information outputted by the information evaluation thread to be configured is obtained.
Step S204, performing evaluation index processing on the information evaluation thread to be configured based on the example segment probability and the example prediction segment probability of the example information to obtain an evaluation index processing result.
In the embodiment of the application, the evaluation index processing result is used for representing the difference between the actual example fragment probability of the example information and the example prediction fragment probability obtained by carrying out regression evaluation on the example information through the information evaluation thread to be configured, so as to represent the accuracy of the information evaluation thread. And calculating the example fragment probability and the example prediction fragment probability of the example information through a preset cost index strategy to obtain an evaluation index processing result of the information evaluation thread to be configured. It should be noted that, in the embodiment of the present application, the cost index policy, that is, the evaluation index processing method is not specifically limited, and for example, average loss, cross entropy loss, etc. may all conform to the foregoing evaluation index processing procedure.
Step S205, optimizing the thread coefficient in the information evaluation thread to be configured based on the evaluation index processing result until the information evaluation thread to be configured converges, and obtaining the configured information evaluation thread.
After the evaluation index processing result is obtained through calculation, the evaluation index processing result can be fed back to the information evaluation thread to be configured, so that the thread coefficient in the information evaluation thread to be configured is optimized to be the residual coefficient value, the optimized information evaluation thread is obtained, and the evaluation index processing result obtained through calculation by the optimized information evaluation thread is smaller.
When the information evaluation thread is configured, the method can firstly adopt a target calculation thread built in advance, project the example fragment of each example information into the corresponding example fragment probability, acquire the example prediction fragment probability of each example information output by the information evaluation thread to be configured, so that the information evaluation thread to be configured is subjected to evaluation index processing based on the example fragment probability and the example prediction fragment probability of the example information, and the thread coefficient in the information evaluation thread is optimized based on the evaluation index processing result until the information evaluation thread to be configured converges, and the configured information evaluation thread is obtained. In the embodiment of the application, since the distribution of the example fragment possibilities obtained by the projection of the target computing thread in the example fragment possibility set is related to the distribution of the example fragments in the example fragment set, the continuous example fragments are projected into the possibility space to obtain the example fragment possibilities based on the target computing thread, so that the evaluation index processing can be performed by expanding the fragments into the possible value range while the distribution information of the example fragment size is maintained, the error of the evaluation index processing process is further reduced, and the accuracy and the robustness of the information evaluation thread obtained by configuration are improved.
On the basis of the above, an analysis system of walking postures of cognitively dysfunctional elderly people is shown, comprising a processor and a memory in communication with each other, the processor being adapted to read a computer program from the memory and execute it, so as to implement the above-mentioned method.
On the basis of the above, there is also provided a computer readable storage medium on which a computer program stored which, when run, implements the above method.
In summary, based on the above scheme, when analysis is performed, the target evaluation coefficient of the walking posture information of each of the pending old people can be obtained by performing regression evaluation through the information evaluation thread configured in advance, then the walking posture information of a plurality of the pending old people is distributed based on the target evaluation coefficient of the walking posture information of each of the pending old people, the walking posture information distribution situation of the pending old people is obtained, the walking posture information of a specified number of the pending old people is screened from the walking posture information distribution situation of the pending old people to be determined as walking abnormal data, and the walking abnormal data is analyzed, so that an analysis result is obtained. The information evaluation thread to be configured can be configured based on the example fragment probability of the example information and the example prediction fragment probability, wherein the example prediction fragment probability is obtained by carrying out regression evaluation on the example information through the information evaluation thread to be configured, and the example fragment probability is obtained by adopting a target calculation thread built in advance to project the example fragment of the example information. According to the method and the device for processing the sample fragments, continuous sample fragments are projected into a possibility space based on a target computing thread to obtain the sample fragment possibility, the sample fragments can be scaled into a value range of the possibility while distribution information of the sample fragment sizes is maintained, and then when the sample fragment possibility of sample information and sample prediction fragment possibility configuration information are utilized to evaluate the thread, errors in an evaluation index processing process can be reduced, and accuracy and robustness of the information evaluation thread obtained through configuration are improved, so that the accuracy of analysis can be improved by using the information evaluation thread, and an analysis result can be obtained more accurately.
It should be appreciated that the systems and modules thereof shown above may be implemented in a variety of ways. For example, in some embodiments, the system and its modules may be implemented in hardware, software, or a combination of software and hardware. Wherein the hardware portion may be implemented using dedicated logic; the software portions may then be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or special purpose design hardware. Those skilled in the art will appreciate that the methods and systems described above may be implemented using computer executable instructions and/or embodied in processor control code, such as provided on a carrier medium such as a magnetic disk, CD or DVD-ROM, a programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The system and its modules of the present application may be implemented not only with hardware circuitry, such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but also with software, such as executed by various types of processors, and with a combination of the above hardware circuitry and software (e.g., firmware).
It should be noted that, the advantages that may be generated by different embodiments may be different, and in different embodiments, the advantages that may be generated may be any one or a combination of several of the above, or any other possible advantages that may be obtained.
Claims (10)
1. An analysis method of walking postures of the aged with cognitive dysfunction is characterized by comprising the following steps:
for each of a plurality of pieces of walking gesture information of the old people to be determined, obtaining a target evaluation coefficient of each piece of walking gesture information of the old people to be determined through regression evaluation of an information evaluation thread configured in advance; the information evaluation thread is configured to obtain the information evaluation thread to be configured based on the example fragment possibility and the example prediction fragment possibility of the example information; the example prediction fragment possibility of the example information is obtained by carrying out regression evaluation on the example information through the information evaluation thread to be configured; the example fragment possibility is obtained by projecting an example fragment of the example information by adopting a target calculation thread built in advance;
distributing the walking gesture information of the plurality of the pending old people based on target evaluation coefficients of the walking gesture information of each pending old people to obtain the walking gesture information distribution situation of the pending old people;
and screening the walking posture information of a specified number of the pending old people from the walking posture information distribution condition of the pending old people, determining the walking posture information as walking abnormal data, and analyzing the walking abnormal data to obtain an analysis result.
2. The method of claim 1, wherein the target evaluation coefficients include a predicted segment likelihood at a segment level and a remaining index evaluation coefficient at not less than one remaining interaction level, the distributing the plurality of pending elderly walking posture information based on the target evaluation coefficients of each of the pending elderly walking posture information to obtain a pending elderly walking posture information distribution, including:
acquiring a first confidence coefficient under the segment level and a second confidence coefficient under the residual interaction level;
for any walking posture information of the old people to be determined, performing first function calculation on the predicted fragment possibility of the walking posture information of the old people to be determined based on the first confidence coefficient to obtain a first calculated value, and performing first function calculation on the residual index evaluation coefficient of the walking posture information of the old people to be determined based on the second confidence coefficient to obtain a second calculated value;
processing a second function of the first calculated value and the second calculated value of each piece of walking gesture information of the old to be determined to be a splicing evaluation coefficient of the walking gesture information of the old to be determined;
and carrying out simplified distribution on the walking posture information of the plurality of the old people to be determined according to the splicing evaluation coefficients to obtain the walking posture information distribution situation of the old people to be determined.
3. The method of claim 1, wherein the method further comprises:
configuring the information assessment thread by: acquiring an example fragment set corresponding to the example information set; the set of example snippets includes example snippets of each of the example information in the set of example information; projecting the example fragments of each example information into corresponding example fragment possibilities by adopting a target calculation thread built in advance to obtain an example fragment possibility set; wherein the distribution of each of the example segment likelihoods in the example segment likelihoods set is linked to the distribution of the example segment in the example segment set;
acquiring the possibility of an example prediction fragment of each piece of example information output by the information evaluation thread to be configured;
performing evaluation index processing on the information evaluation thread to be configured based on the example fragment probability of the example information and the example prediction fragment probability to obtain an evaluation index processing result;
and optimizing the thread coefficient in the information evaluation thread to be configured based on the evaluation index processing result until the information evaluation thread to be configured converges, so as to obtain the configured information evaluation thread.
4. A method as claimed in claim 3, wherein the method further comprises:
acquiring an original calculation thread built in advance; determining not less than three target values and weights corresponding to the target values from a plurality of sample fragments;
projecting each target value into a corresponding original fragment possibility by adopting the original calculation thread;
determining function coefficients in the original calculation thread according to the weights corresponding to the target values and the original fragment probability;
and building the target computing thread according to the function coefficients and the original computing thread.
5. The method of claim 4, wherein the original computing thread and the target computing thread each satisfy the following conditions:
when the example segment is 0, the possibility of the original segment obtained by projecting the example segment through the original computing thread is 0, and the possibility of the example segment obtained by projecting the example segment through the target computing thread is 0;
when the example segment is positive, the possibility of the original segment obtained by projecting the example segment through the original computing thread is 1, and the possibility of the example segment obtained by projecting the example segment through the target computing thread is 1; the original segment likelihood and the example segment likelihood are each in the range of 0 to 1; the original segment likelihood and the example segment likelihood monotonically increase with respect to the example segment.
6. The method of claim 4 or 5, wherein determining not less than three target values from the number of example segments, and weights corresponding to the target values, comprises:
distributing the plurality of example fragments according to the sequence from big to small to form an example fragment distribution condition;
determining the corresponding positioning of each sample fragment in the sample fragment distribution situation; determining not less than three fragment positioning percentages based on a preset target value number;
and determining not less than three target values from the plurality of sample fragments according to the fragment positioning percentages and the positioning corresponding to each sample fragment, and determining the fragment positioning percentage corresponding to the target value as the weight corresponding to the target value.
7. The method of claim 5, wherein said determining function coefficients in said original computing thread based on said weights and said original segment likelihoods for each of said target values comprises:
performing the following integration iterative process on the original computing thread: acquiring the possibility of an original fragment obtained by projecting the target value by the original calculation thread;
performing evaluation index processing on the original calculation thread according to the weight corresponding to the target value and the original fragment probability to obtain an integration result of the original calculation thread;
optimizing coefficients in the original calculation thread based on the integration result to obtain an optimized calculation thread;
when the optimized computing thread does not accord with a preset integration condition, determining the optimized computing thread as an original computing thread in the next integration iteration process;
stopping the integration iteration process when the optimized calculation thread accords with a preset integration condition;
and determining coefficients in a calculation thread when the integration iteration process is stopped as the function coefficients.
8. The method of claim 7, wherein the performing evaluation index processing on the original computing thread according to the weight corresponding to the target value and the original segment likelihood to obtain an integrated result of the original computing thread comprises:
building a target cost index strategy based on the original calculation thread;
substituting the weight corresponding to the target value and the original fragment probability into the target cost index strategy to obtain the integration result corresponding to the target value.
9. The method of claim 8, wherein the building a target cost indicator policy based on the original computing thread comprises:
determining a comparison between the weights in the target values and the original computing thread;
respectively averaging the comparison results corresponding to the target values to obtain a plurality of average values;
and fusing the average values to obtain the target cost index strategy.
10. An analysis system of walking attitudes of cognitively dysfunctional elderly persons, characterized in that it comprises a processor and a memory in communication with each other, said processor being adapted to read a computer program from said memory and to execute it in order to implement the method according to any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410229811.2A CN117809849B (en) | 2024-02-29 | 2024-02-29 | Analysis method and system for walking postures of old people with cognitive dysfunction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410229811.2A CN117809849B (en) | 2024-02-29 | 2024-02-29 | Analysis method and system for walking postures of old people with cognitive dysfunction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117809849A true CN117809849A (en) | 2024-04-02 |
| CN117809849B CN117809849B (en) | 2024-05-03 |
Family
ID=90423887
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410229811.2A Active CN117809849B (en) | 2024-02-29 | 2024-02-29 | Analysis method and system for walking postures of old people with cognitive dysfunction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117809849B (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110584601A (en) * | 2019-08-26 | 2019-12-20 | 首都医科大学 | Method for monitoring and evaluating cognitive function of old people |
| CN110853753A (en) * | 2019-11-26 | 2020-02-28 | 上海健康医学院 | Cognitive dysfunction old man rehabilitation and nursing system at home |
| CN115346670A (en) * | 2022-08-15 | 2022-11-15 | 浙江工业大学 | Parkinson's disease rating method based on posture recognition, electronic device and medium |
| CN116230226A (en) * | 2023-02-01 | 2023-06-06 | 武汉科技大学 | Device for preventing and monitoring cognitive dysfunction of old people |
| CN116453058A (en) * | 2023-04-27 | 2023-07-18 | 郑州轻工业大学 | Home old man behavior monitoring method and system based on deep learning and digital body separation |
| CN116439656A (en) * | 2022-12-09 | 2023-07-18 | 大连大学 | Gait-based cognitive impairment evaluation method |
| CN116597507A (en) * | 2023-04-23 | 2023-08-15 | 电子科技大学 | Human body action normalization evaluation method and system |
| CN116682171A (en) * | 2023-05-15 | 2023-09-01 | 深圳和而泰智能控制股份有限公司 | Method for training a panic-tense gait recognition model, gait recognition method and related device |
| CN116737975A (en) * | 2023-05-15 | 2023-09-12 | 重庆海辉星科技有限公司 | Public health data query method and system applied to image analysis |
| CN116779160A (en) * | 2023-05-26 | 2023-09-19 | 光翰科技(上海)有限公司 | Self-adaptive cognitive disorder screening system |
| CN116879684A (en) * | 2023-09-06 | 2023-10-13 | 成都汉度科技有限公司 | Abnormal line judging method and system |
| US20230355135A1 (en) * | 2022-05-03 | 2023-11-09 | Asia University | Intelligent gait analyzing apparatus |
| CN117059276A (en) * | 2023-08-23 | 2023-11-14 | 重庆市中医院 | Patient vital sign monitoring method and system based on artificial intelligence |
| CN117370767A (en) * | 2023-12-07 | 2024-01-09 | 四川经科企业管理服务有限公司 | User information evaluation method and system based on big data |
| CN117612712A (en) * | 2024-01-23 | 2024-02-27 | 首都医科大学宣武医院 | Method and system for detecting and improving cognition evaluation diagnosis precision |
-
2024
- 2024-02-29 CN CN202410229811.2A patent/CN117809849B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110584601A (en) * | 2019-08-26 | 2019-12-20 | 首都医科大学 | Method for monitoring and evaluating cognitive function of old people |
| CN110853753A (en) * | 2019-11-26 | 2020-02-28 | 上海健康医学院 | Cognitive dysfunction old man rehabilitation and nursing system at home |
| US20230355135A1 (en) * | 2022-05-03 | 2023-11-09 | Asia University | Intelligent gait analyzing apparatus |
| CN115346670A (en) * | 2022-08-15 | 2022-11-15 | 浙江工业大学 | Parkinson's disease rating method based on posture recognition, electronic device and medium |
| CN116439656A (en) * | 2022-12-09 | 2023-07-18 | 大连大学 | Gait-based cognitive impairment evaluation method |
| CN116230226A (en) * | 2023-02-01 | 2023-06-06 | 武汉科技大学 | Device for preventing and monitoring cognitive dysfunction of old people |
| CN116597507A (en) * | 2023-04-23 | 2023-08-15 | 电子科技大学 | Human body action normalization evaluation method and system |
| CN116453058A (en) * | 2023-04-27 | 2023-07-18 | 郑州轻工业大学 | Home old man behavior monitoring method and system based on deep learning and digital body separation |
| CN116682171A (en) * | 2023-05-15 | 2023-09-01 | 深圳和而泰智能控制股份有限公司 | Method for training a panic-tense gait recognition model, gait recognition method and related device |
| CN116737975A (en) * | 2023-05-15 | 2023-09-12 | 重庆海辉星科技有限公司 | Public health data query method and system applied to image analysis |
| CN116779160A (en) * | 2023-05-26 | 2023-09-19 | 光翰科技(上海)有限公司 | Self-adaptive cognitive disorder screening system |
| CN117059276A (en) * | 2023-08-23 | 2023-11-14 | 重庆市中医院 | Patient vital sign monitoring method and system based on artificial intelligence |
| CN116879684A (en) * | 2023-09-06 | 2023-10-13 | 成都汉度科技有限公司 | Abnormal line judging method and system |
| CN117370767A (en) * | 2023-12-07 | 2024-01-09 | 四川经科企业管理服务有限公司 | User information evaluation method and system based on big data |
| CN117612712A (en) * | 2024-01-23 | 2024-02-27 | 首都医科大学宣武医院 | Method and system for detecting and improving cognition evaluation diagnosis precision |
Non-Patent Citations (4)
| Title |
|---|
| ANA PAULA OLIVEIRA BORGES 等: ""Evaluation of postural balance in mild cognitive impairment through a three-dimensional electromagnetic system"", 《 BRAZILIAN JOURNAL OF OTORHINOLARYNGOLOGY》, vol. 82, no. 4, 15 June 2016 (2016-06-15), pages 433 - 441 * |
| 姚小慧;邵堃;孙传能;金柳颀;: "基于隐式感知的老人认知健康评估系统", 计算机应用与软件, vol. 35, no. 03, 15 March 2018 (2018-03-15), pages 70 - 74 * |
| 陈岩 等: ""轻度认知障碍老年人在不同站立姿势—空间工作记忆任务中姿势控制能力研究——基于COP样本熵"", 《墙报交流（运动生物力学分会）》, 3 November 2023 (2023-11-03), pages 19315 - 19317 * |
| 魏来: ""中老年人群跌倒风险评估的数据采集系统"", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》, 15 July 2020 (2020-07-15), pages 140 - 422 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117809849B (en) | 2024-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9329981B2 (en) | Testing program, testing method, and testing device | |
| US20090235251A1 (en) | Systems, methods and computer program products for improving placement performance of message transforms by exploiting guided replication | |
| CN109871315B (en) | Diagnosis method and device for system upgrade failure based on machine learning | |
| US8775338B2 (en) | Computer-implemented systems and methods for constructing a reduced input space utilizing the rejected variable space | |
| CN111145076B (en) | Data parallelization processing method, system, equipment and storage medium | |
| US20200250559A1 (en) | Identifying and Remediating System Anomalies Through Machine Learning Algorithms | |
| US20200150941A1 (en) | Heterogenous computer system optimization | |
| KR102167747B1 (en) | Apparatus and Method of managing Mobile device memory for analyzing a user utilization pattern by a neural network algorithm to predict a next application | |
| US9396095B2 (en) | Software verification | |
| CN112612887A (en) | Log processing method, device, equipment and storage medium | |
| US20100042965A1 (en) | Method and System for Scalable Reduction in Registers With Sat-Based Resubstitution | |
| US20190179692A1 (en) | Adaptive System and a Method for Application Error Prediction and Management | |
| JPWO2019189249A1 (en) | Learning devices, learning methods, and programs | |
| CN113886237A (en) | Analysis report generation method and device, electronic equipment and storage medium | |
| CN117809849B (en) | Analysis method and system for walking postures of old people with cognitive dysfunction | |
| US20230035666A1 (en) | Anomaly detection in storage systems | |
| CN115828244A (en) | Memory leak detection method and device and related equipment | |
| CN113850428A (en) | Job scheduling prediction processing method and device and electronic equipment | |
| KR102135719B1 (en) | Parts management device and operating method of the same | |
| Al‐Najjar et al. | Scheduling of workflow jobs based on twostep clustering and lowest job weight | |
| US9672321B2 (en) | Virtual positive slack in physical synthesis | |
| US11995485B2 (en) | Automatically switching between quantum services and classical services based on detected events | |
| CN114615144B (en) | Network optimization method and system | |
| EP4184328A1 (en) | Medical imaging device fault handling | |
| CN110647519B (en) | Method and device for predicting missing attribute value in test sample |
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 |