Artificial intelligence-based cardiopulmonary resuscitation training operation evaluation method

Abstract

The application discloses an artificial intelligence-based cardiopulmonary resuscitation training operation evaluation method, which comprises the following steps: acquiring running state data of a simulation model of a current round of cardiopulmonary resuscitation training operation of a training personnel; acquiring operation data of the current round of cardiopulmonary resuscitation training operation of the training personnel collected by a monitoring device; evaluating the operation data according to evaluation constituent element categories of cardiopulmonary resuscitation operation lifesaving standards and judgment standards, to obtain real-time evaluation of each evaluation constituent element of the training personnel; obtaining predicted evaluation of each evaluation constituent element of the training personnel in the current round based on a performance prediction model; and comparing the real-time evaluation with the predicted evaluation to evaluate the cardiopulmonary resuscitation training operation of the training personnel. The application can exclude invalid evaluation, realize comprehensive and objective evaluation of each evaluation constituent element of the training personnel, and provide targeted guidance and suggestions to improve the training effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical simulation, in particular to a cardiopulmonary resuscitation training operation evaluation method based on artificial intelligence. BACKGROUND

[0002] The existing cardiopulmonary resuscitation (CPR) training is mainly for the current operation, and the training personnel are evaluated and scored according to the cardiopulmonary resuscitation operation rescue standard and evaluation standard, which lacks tracking of the training personnel in the whole learning process for the cardiopulmonary resuscitation operation. At the same time, the cardiopulmonary resuscitation manikin used for operation generally has a hardware structure simulating the chest movement of a human body, and corresponding monitoring sensors and electronic devices for monitoring the cardiopulmonary resuscitation operation performed by the training personnel, so as to obtain the compression depth, compression frequency, simulated chest recoil data, artificial respiration tidal volume and other data of the training personnel. However, the health status of these hardware running states and electronic devices is not only affected by the use frequency of the manikin and the operation behavior of the training students, but also affects the cardiopulmonary resuscitation operation of the training personnel. However, this influencing factor is not considered in the existing cardiopulmonary resuscitation training process, which may lead to an incomplete and inaccurate evaluation of the training personnel's cardiopulmonary resuscitation. SUMMARY

[0003] In order to achieve the above-mentioned purpose, the present application provides a cardiopulmonary resuscitation training operation evaluation method based on artificial intelligence, comprising:

[0004] S1, obtaining the running state data of the simulation model for the training personnel to perform the cardiopulmonary resuscitation training operation in the current round, including the running state data of the hardware for simulating the human body structure and the running state data of the monitoring device for collecting the operation data of the cardiopulmonary resuscitation training operation;

[0005] S2, obtaining the operation data of the cardiopulmonary resuscitation training operation performed by the training personnel in the current round collected by the monitoring device;

[0006] S3, evaluating the operation data according to the evaluation constituent element categories of the cardiopulmonary resuscitation operation rescue standard and evaluation standard respectively, to obtain the real-time evaluation of each evaluation constituent element category of the training personnel;

[0007] S4, obtaining the running state data of the simulation model for the training personnel to perform the cardiopulmonary resuscitation training operation in the historical round, the operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation constituent element category, and obtaining the predicted evaluation of each evaluation constituent element category of the training personnel in the current round based on a performance prediction model, wherein the construction steps of the performance prediction model include:

[0008] S41, obtain the running state data of the historical simulation model of the historical round of cardiopulmonary resuscitation training operation of the historical training personnel, the historical operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation element category, construct a data set T={Xi, i=1, 2,…n}, Xi={JM1, JM2, JM3}, wherein Xi is the data set of the historical round of cardiopulmonary resuscitation training operation of the ith historical training personnel, n is the number of historical training personnel, J represents the round of cardiopulmonary resuscitation training operation, JM1, JM2, and JM3 represent the running state data of the J round simulation model, the operation data of the cardiopulmonary resuscitation training operation, and the evaluation of each evaluation element category, respectively;

[0009] S42, process the historical evaluation of each evaluation element category, filter and mark the effective evaluation, including:

[0010] S421, extract the historical evaluation arranged in time sequence according to the evaluation element category, set a threshold, and calculate the variance of adjacent historical evaluations in turn;

[0011] S422, determine whether the variance meets the threshold range,

[0012] Yes, mark the historical evaluation as normal evaluation;

[0013] No, mark the historical evaluation as abnormal evaluation, and go to step S423;

[0014] S423, based on the corresponding relationship between the operation data and the evaluation element category, determine the monitoring device and hardware corresponding to the operation data of the evaluation element category according to the abnormal evaluation, obtain the running state data of the monitoring device and the hardware corresponding to the round of the abnormal evaluation, and obtain the health status of the monitoring device and the hardware;

[0015] S424, determine whether the monitoring device and the hardware are in a healthy state,

[0016] Yes, mark the abnormal evaluation as unstable evaluation,

[0017] No, mark the abnormal evaluation as invalid evaluation;

[0018] S43, predict the prediction evaluation of the current round of the training personnel based on the effective evaluation;

[0019] S5, compare the real-time evaluation with the prediction evaluation, and evaluate the cardiopulmonary resuscitation training operation of the training personnel.

[0020] Further, the monitoring device includes a pressure sensor, a distance measuring sensor, and a flow monitoring device.

[0021] Further, the hardware includes an elastic module simulating the chest of a human body.

[0022] Further, the evaluation component categories include pressing position, pressing depth, chest rebound, ventilation volume, ventilation duration, airway opening, interruption frequency, pressing duration ratio, and pressing frequency to ventilation frequency ratio.

[0023] Further, after step S423, the method further includes verifying the health status of the monitoring device and the hardware by using historical evaluations of other historical training personnel who use the same simulation model at the same time in a time sequence, and specifically includes:

[0024] S4231, extracting a data set of cardiopulmonary resuscitation training operation of other historical training personnel who use the same simulation model to perform cardiopulmonary resuscitation training operation at the same time in a time sequence as the training personnel;

[0025] S4232, repeating steps S421-S422;

[0026] S4233, determining whether the historical evaluation of the same evaluation component category of the other historical training personnel who use the same simulation model at the same time is a normal evaluation,

[0027] Yes, determining that the hardware and the monitoring device on the simulation model are in a healthy state;

[0028] No, determining that at least one of the hardware and the monitoring device on the simulation model is in an unhealthy state.

[0029] Further, step S43, based on the valid evaluation, predicting the prediction evaluation of the current round of the training personnel includes:

[0030] S431, counting the normal evaluation, and calculating the interval range of the prediction evaluation according to the confidence interval logic of the normal distribution in statistics;

[0031] S432, calculating the probability of unstable evaluation in the prediction evaluation according to the frequency change trend of unstable evaluation in the valid evaluation.

[0032] The beneficial effects of the present application are:

[0033] By obtaining the historical cardiopulmonary resuscitation training operation data of the training personnel, the training personnel's mastery of cardiopulmonary resuscitation operation can be comprehensively and objectively evaluated, and each evaluation component is processed and analyzed to obtain the evaluation of each evaluation component, so that the analysis is more comprehensive, and the training personnel can quickly capture the deficiencies and then train and learn accordingly.

[0034] By constructing the performance prediction model, the historical cardiopulmonary resuscitation training operation data of the training personnel is processed, and invalid performance caused by factors such as wear and failure of the simulation model hardware and electronic monitoring devices is removed. The mastery of each evaluation component of cardiopulmonary resuscitation of the training personnel can be objectively predicted, and the real-time evaluation obtained by the real-time collected cardiopulmonary resuscitation training operation data is compared, so as to comprehensively and objectively analyze the cardiopulmonary resuscitation mastery of the training personnel and improve the skill mastery efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a flowchart of the cardiopulmonary resuscitation training operation evaluation method based on artificial intelligence according to an embodiment of the present application.

[0036] Figure 2 is a flowchart of the performance prediction model construction step according to an embodiment of the present application.

[0037] Figure 3 is a flowchart of step S42 according to an embodiment of the present application.

[0038] Figure 4 is a flowchart of the step of verifying the health status of the monitoring device and the hardware by using the historical evaluation of other historical training personnel using the same simulation model at the same time according to an embodiment of the present application.

[0039] Figure 5 is a schematic diagram of the use of the simulation model by part of the historical training personnel according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme in the present application will be described clearly and completely in combination with the drawings of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0041] As shown in Figures 1-4 , the present application provides a cardiopulmonary resuscitation training operation evaluation method based on artificial intelligence, comprising:

[0042] S1, obtaining the running state data of the simulation model used by the training personnel in the current round of cardiopulmonary resuscitation training operation, including the running state data of the hardware used for simulating human body structure and the running state data of the monitoring device used for collecting operation data of cardiopulmonary resuscitation training operation.

[0043] The existing cardiopulmonary resuscitation operation training usually uses a simulation model with a simulated human body structure, including a hardware structure capable of simulating human thoracic fluctuation, compression feedback, and a monitoring device capable of monitoring the compression operation performed by the training personnel. Specifically, the hardware structure can be provided as an elastic module simulating human thoracic fluctuation, such as a gas bag, a spring, a gas cylinder, a piston, etc.; the monitoring device is provided to monitor, but not limited to, compression position, compression force, compression displacement, intubation position, breathing volume, etc., which can include various pressure sensors, distance measuring sensors, flow monitoring devices, etc., such as a matrix type thin film pressure sensor for monitoring the compression position, an infrared distance measuring sensor for monitoring the compression depth, thoracic rebound, and intubation position, and a flow meter for monitoring the breathing volume of the respiratory tract. In the long-term training operation process, the wear and tear of the hardware structure itself and the damage of the hardware structure or the monitoring device caused by human training operation errors will cause the hardware structure simulation to be distorted, or the monitoring data of the monitoring device to be inaccurate, etc. In this step, by acquiring the running state data of the simulation model hardware and the monitoring device, the running status of each structure is monitored and analyzed to obtain the health status of the hardware structure and each monitoring device, and to determine whether the equipment is operating normally.

[0044] S2, acquiring operation data of the cardiopulmonary resuscitation training operation performed by the training personnel in the current round, collected by the monitoring device.

[0045] In this step, the operation data of the cardiopulmonary resuscitation training operation performed by the training personnel is collected by the monitoring device provided on the simulation model, mainly including operation data such as compression position, compression depth, thoracic rebound, ventilation volume, and intubation position obtained by pressure sensors, distance measuring sensors, and flow monitoring devices, etc., to provide data basis for evaluating various indicators of the cardiopulmonary resuscitation training operation of the training personnel.

[0046] S3, evaluating the operation data according to the evaluation constituent element categories of the cardiopulmonary resuscitation operation rescue standard and judgment standard, to obtain real-time evaluation of each evaluation constituent element category of the training personnel.

[0047] The evaluation constituent element categories mainly include compression position, compression depth, thoracic rebound, ventilation volume, ventilation duration, airway opening, interruption frequency, compression duration ratio, and compression frequency to ventilation frequency ratio. In the usual cardiopulmonary resuscitation training operation training, the rescue is taken as the evaluation standard, and each evaluation constituent element category is not evaluated separately, so that the training personnel cannot clearly and definitely know whether their training operation meets all the evaluation constituent element categories. In this step, the operation data of each evaluation constituent element category is evaluated, the mastery degree of each evaluation constituent element category of the training personnel can be obtained, and the training personnel can quickly find the deficiencies and thus perform targeted training to improve the training efficiency.

[0048] S4, acquire the running state data of the simulation model of the historical round of cardiopulmonary resuscitation training operation of the training personnel, the operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation element category, and obtain the predicted evaluation of each evaluation element category of the training personnel in the current round based on the score prediction model.

[0049] In this step, the historical training data of the training personnel is collected to predict the mastery degree of the training personnel on each evaluation element category. In the process of cardiopulmonary resuscitation teaching and training, training places such as medical colleges usually provide multiple simulation models for training personnel to practice. The running state of the hardware and monitoring device of each simulation model is different, and the training personnel also cross-use them. Therefore, it may lead to different evaluation results when the same training personnel performs cardiopulmonary resuscitation training operation on different simulation models. In addition, the failure of a certain hardware or monitoring device on the simulation model, or the replacement of the hardware or monitoring device, may also lead to different evaluation results of the training personnel, resulting in a large difference in the evaluation results of the training personnel in the historical round of cardiopulmonary resuscitation training operation for mastering cardiopulmonary resuscitation operation skills, which cannot reflect the true level of the training personnel. Therefore, in order to objectively evaluate and predict the mastery degree of the training personnel on each evaluation element category of cardiopulmonary resuscitation, the historical training data collected in the step of building the score prediction model is processed, which specifically includes:

[0050] S41, acquire the running state data of the historical simulation model of the historical round of cardiopulmonary resuscitation training operation of the historical training personnel, the historical operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation element category, and build a data set T={Xi, i=1, 2,…n}, Xi={JM1, JM2, JM3}, wherein Xi is the data set of the historical round of cardiopulmonary resuscitation training operation of the i-th historical training personnel, n is the number of historical training personnel, J represents the round of cardiopulmonary resuscitation training operation, JM1, JM2, and JM3 represent the running state data of the J-round simulation model, the operation data of the cardiopulmonary resuscitation training operation, and the evaluation of each evaluation element category, respectively.

[0051] Reference Figure 5 , shows the use of simulation models by some historical training personnel, wherein training personnel 1 uses simulation model 1 in the first round and simulation model 1' in the J-th round. Simulation model 1' indicates that the monitoring device 1 is replaced by monitoring device 1' based on the unchanged hardware 1 of simulation model 1, which can be the replacement of all monitoring devices or the replacement of part of the monitoring devices; training personnel m uses the same simulation model 1 as training personnel 1 in the first round in the N-th round.

[0052] S42, process the historical evaluation of each evaluation component category, screen and mark the effective evaluation, including:

[0053] S421, extract the historical evaluation of the training personnel arranged in time sequence according to the evaluation component category, set the threshold, and calculate the variance of the adjacent historical evaluation in turn;

[0054] Generally, the cardiopulmonary resuscitation skill will be gradually improved with the multiple training operations of the training personnel. In the early stage of training, there may be large evaluation fluctuation, and in the later stage, it will tend to be stable, but some factors may also cause operation errors or invalidity. Therefore, by calculating the variance of the adjacent historical evaluation and setting the threshold, the evaluation results in the stable state and the abnormal evaluation results can be screened out;

[0055] S422, judge whether the variance meets the threshold range,

[0056] Yes, mark the historical evaluation as normal evaluation;

[0057] No, mark the historical evaluation as abnormal evaluation, and enter step S423;

[0058] S423, based on the corresponding relationship between the operation data and the evaluation component category, determine the monitoring device and hardware corresponding to the operation data of the abnormal evaluation according to the evaluation component category corresponding to the abnormal evaluation, obtain the running state data of the monitoring device and the hardware corresponding to the corresponding round of the abnormal evaluation, and obtain the health status of the monitoring device and the hardware;

[0059] Different monitoring devices have different functions and collect different operation data. In the embodiment, the corresponding relationship between the operation data and the evaluation component category is constructed in advance, so as to quickly correspond to the operation data corresponding to the evaluation component of the abnormal evaluation, so as to correspond to the monitoring device and / or hardware collecting the operation data, and further determine the running state of the monitoring device and / or hardware corresponding to the abnormal evaluation. By judging the health status of the current monitoring device and / or hardware of the abnormal evaluation, it can be judged whether the abnormal evaluation is an abnormal effective evaluation of the training personnel. Specifically, if all are healthy, it indicates that the abnormal evaluation is a failure or error of the cardiopulmonary resuscitation training operation of the training personnel, or a substandard or unstable situation, and if some monitoring devices and / or hardware are in a non-healthy state, it indicates that the abnormal evaluation is an invalid evaluation of the training personnel;

[0060] S424, judge whether the monitoring device and the hardware are in a healthy state,

[0061] Yes, mark the abnormal evaluation as unstable evaluation,

[0062] No, mark the abnormal evaluation as invalid evaluation;

[0063] Through this step, invalid evaluations caused by faults, wear and tear, etc. of the simulation model hardware structure and the monitoring device can be removed, and unstable evaluations can also be screened out, which can be used to guide and advise the training personnel and improve the training effect.

[0064] In some embodiments, after step S423, the health status of the monitoring device and the hardware is also verified by using the historical evaluations of other historical training personnel who used the same simulation model at the same time in the time sequence, specifically including:

[0065] S4231, extracting a data set of cardiopulmonary resuscitation training operations of other historical training personnel who used the same simulation model to perform cardiopulmonary resuscitation training operations at the same time in the time sequence as the training personnel;

[0066] S4232, repeating steps S421-S422;

[0067] S4233, determining whether the historical evaluation of the same evaluation element category of the other historical training personnel using the same simulation model at the same time is a normal evaluation,

[0068] Yes, it is determined that the hardware and the monitoring device on the simulation model are in a healthy state;

[0069] No, it is determined that at least one of the hardware and the monitoring device on the simulation model is in an unhealthy state.

[0070] S43, predicting the prediction evaluation of the current round of the training personnel based on the valid evaluation, specifically including:

[0071] S431, counting the normal evaluation, and calculating the interval range of the prediction evaluation according to the confidence interval logic of the normal distribution in statistics;

[0072] S432, calculating the probability of the occurrence of unstable evaluation in the prediction evaluation according to the frequency change trend of the occurrence of unstable evaluation in the valid evaluation.

[0073] S5, comparing the real-time evaluation with the prediction evaluation to evaluate the cardiopulmonary resuscitation training operation of the training personnel.

[0074] In this embodiment, by constructing a performance prediction model, referring to the data of the historical cardiopulmonary resuscitation training operation of the training personnel, removing the factors affecting the objective evaluation, predicting the cardiopulmonary resuscitation training operation mastery of the training personnel, and comparing the real-time evaluation of the training operation of the training personnel collected in real time with the prediction evaluation, the training personnel can be evaluated comprehensively and objectively.

Claims

1. A method for evaluating cardiopulmonary resuscitation training operation based on artificial intelligence, characterized by, The method comprises the following steps: S1, obtaining the running state data of the simulation model for the cardiopulmonary resuscitation training operation of the current round of the training personnel, including the running state data of the hardware for simulating the human body structure and the running state data of the monitoring device for collecting the operation data of the cardiopulmonary resuscitation training operation; S2, obtaining the operation data of the cardiopulmonary resuscitation training operation of the current round of the training personnel collected by the monitoring device; S3, evaluating the operation data according to the evaluation constituent element categories of the cardiopulmonary resuscitation operation rescue standard and the evaluation standard, to obtain the real-time evaluation of each evaluation constituent element category of the training personnel; S4, obtaining the running state data of the simulation model for the cardiopulmonary resuscitation training operation of the historical round of the training personnel, the operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation constituent element category, and obtaining the predicted evaluation of each evaluation constituent element category of the current round of the training personnel based on a performance prediction model, wherein the performance prediction model comprises the following steps: S41, obtaining the running state data of the historical simulation model for the cardiopulmonary resuscitation training operation of the historical round of the historical training personnel, the historical operation data of the cardiopulmonary resuscitation training operation, and the historical evaluation of each evaluation constituent element category, and constructing a data set T={Xi, i=1, 2, …n}, Xi={JM1, JM2, JM3}, wherein Xi is the data set of the historical round of the cardiopulmonary resuscitation training operation of the i-th historical training personnel, n is the number of historical training personnel, J represents the round of the cardiopulmonary resuscitation training operation, JM1, JM2, and JM3 represent the running state data of the J-round simulation model, the operation data of the cardiopulmonary resuscitation training operation, and the evaluation of each evaluation constituent element category, respectively; S42, processing the historical evaluation of each evaluation constituent element category, screening the effective evaluation, and marking, comprising: S421, extracting the historical evaluation arranged in time sequence according to the evaluation constituent element category, setting a threshold, and calculating the variance of adjacent historical evaluations in sequence; S422, judging whether the variance meets the threshold range, yes, marking the historical evaluation as normal evaluation; no, marking the historical evaluation as abnormal evaluation, and entering step S423; S423, determining the monitoring device and the hardware corresponding to the operation data of the abnormal evaluation according to the evaluation constituent element category corresponding to the abnormal evaluation based on the corresponding relationship between the operation data and the evaluation constituent element category, obtaining the running state data of the monitoring device and the hardware of the round corresponding to the abnormal evaluation, and obtaining the health status of the monitoring device and the hardware; S424, judging whether the monitoring device and the hardware are in the healthy state, yes, marking the abnormal evaluation as unstable evaluation, no, marking the abnormal evaluation as invalid evaluation; S43, predicting the predicted evaluation of the current round of the training personnel based on the effective evaluation; S5, comparing the real-time evaluation with the predicted evaluation, and evaluating the cardiopulmonary resuscitation training operation of the training personnel. 2.The AI-based CPR training operation evaluation method of claim 1, wherein, The monitoring device comprises a pressure sensor, a distance measuring sensor, and a flow monitoring device. 3.The AI-based CPR training operation evaluation method of claim 1, wherein, The hardware comprises an elastic module for simulating the chest of a human body. 4.The AI-based CPR training operation evaluation method of claim 1, wherein The evaluation component categories include pressing position, pressing depth, chest recoil, ventilation volume, ventilation duration, airway opening, interruption frequency, pressing duration ratio, and pressing frequency to ventilation frequency ratio. 5.The artificial intelligence-based CPR training operation evaluation method of claim 1, wherein, After step S423, the method further includes verifying the health status of the monitoring device and the hardware by checking the historical evaluations of other historical trainees using the same simulation model at the same time, specifically including: S4231, extracting a data set of cardiopulmonary resuscitation training operation of other historical trainees using the same simulation model at the same time as the trainee; S4232, repeating steps S421-S422; S4233, determining whether the historical evaluations of the same evaluation component categories of other historical trainees using the same simulation model at the same time are normal evaluations, Yes, determining that the hardware and the monitoring device on the simulation model are in a healthy state; No, determining that at least one of the hardware and the monitoring device on the simulation model is in an unhealthy state. 6.The AI-based CPR training operation evaluation method of claim 1, wherein, Step S43, predicting the prediction evaluation of the current round of the trainee based on the valid evaluation includes: S431, counting normal evaluations, calculating the interval range of the prediction evaluation according to the confidence interval logic of the normal distribution in statistics; S432, calculating the probability of unstable evaluation in the prediction evaluation according to the frequency change trend of unstable evaluation in the valid evaluation.

Images