Orthopedic postoperative rehabilitation training monitoring method based on artificial intelligence

Abstract

The invention belongs to the technical field of orthopaedic postoperative rehabilitation training, and provides an orthopaedic postoperative rehabilitation training monitoring method and system based on artificial intelligence, and the method comprises the steps: in a rehabilitation period, according to the set time DT (n, i) of each rehabilitation training action in a rehabilitation training action group selected by a patient, calculating the set time DT (n, i) of each rehabilitation training action; each rehabilitation training action is shot to obtain a shot picture of the rehabilitation training action, whether the shot picture is qualified or not is judged according to point position data in the shot picture and a picture abnormal value, the shot picture is marked according to a judgment result, and an unqualified picture is processed and analyzed to obtain a qualified action value HD; according to the qualified action value HD, the rehabilitation training actions are marked as qualified actions and unqualified actions, and whether the rehabilitation training actions performed by the patient are qualified or not is judged by processing and analyzing the pictures taken when the patient performs each rehabilitation training action in the rehabilitation training action group.

Description

Technical Field

[0001] The present invention belongs to the technical field of orthopedic postoperative rehabilitation training, and specifically is an orthopedic postoperative rehabilitation training monitoring method based on artificial intelligence. Background Art

[0002] With the continuous advancement of medical technology and the change of medical concepts, orthopedic treatment is no longer limited to wound healing and fracture reduction, but pays more attention to the recovery of patients' postoperative functions and the improvement of their quality of life. This concept has prompted orthopedic surgeons and rehabilitation therapists to incorporate rehabilitation training into the important part of orthopedic postoperative treatment. Clinical practice has proved that orthopedic postoperative rehabilitation training has a significant effect on promoting patient recovery, reducing complications, and improving the quality of life. Through rehabilitation training, patients can accelerate wound healing, promote fracture healing, restore joint function, and enhance muscle strength. These effects have been widely recognized and accepted, further promoting the application and promotion of orthopedic postoperative rehabilitation training.

[0003] A Chinese patent application with publication number CN117637104A discloses an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method, system and server, including: by collecting a number of rehabilitation process training data, a first machine learning model for evaluating the degeneration rate and a second machine learning model for the number of repairs are trained. The smart wearable device determines whether there is a training delay each time the monitored patient undergoes rehabilitation training. If there is a training delay, the first machine learning model and the second machine learning model are used to obtain the predicted degeneration rate and the predicted number of repairs, respectively, and the rehabilitation training schedule is modified. When the monitored patient undergoes training according to the modified rehabilitation training schedule, the smart wearable device makes a second modification to the rehabilitation training schedule; this avoids the problem of discomfort or even injury to the patient due to sudden rehabilitation training according to the rehabilitation training schedule after the training is delayed.

[0004] In the above-mentioned prior art, by collecting a number of rehabilitation process training data, a first machine learning model for evaluating the degeneration rate and a second machine learning model for the number of repairs are trained. The smart wearable device determines whether there is a training delay each time the monitored patient undergoes rehabilitation training. However, in the above-mentioned prior art, there is a lack of judgment on the eligibility of the rehabilitation training movements performed by the patient and the judgment of the rehabilitation training movement group to be recommended to the patient based on multiple groups of performance data in historical rehabilitation cycles. If only one group of rehabilitation training movements is analyzed, and there is a lack of analysis of multiple movements in multiple groups of rehabilitation training movements, it may lead to misjudgment of the patient's rehabilitation status, thereby affecting the patient's further rehabilitation training.

[0005] To this end, the present invention provides an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method. Summary of the invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is: an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method: comprising:

[0008] Step 1: The medical staff evaluates multiple rehabilitation training action groups that are suitable for the patient based on the patient's performance, and inputs the multiple rehabilitation training action groups that are suitable for the patient into the selection terminal. During the rehabilitation period, the patient selects a rehabilitation training action group from the selection terminal for rehabilitation training, wherein each rehabilitation training action group contains multiple rehabilitation training actions, and each rehabilitation training action is marked as (n, i), wherein n is 1, 2, 3..., n represents the rehabilitation training action group number, and i is 1, 2, 3..., i represents the rehabilitation training action number in the rehabilitation training group;

[0009] Step 2: In a rehabilitation cycle, according to the predetermined time DT(n, i) of each rehabilitation training action in the rehabilitation training action group selected by the patient, each rehabilitation training action is photographed to obtain photographs of the rehabilitation training actions, and according to the point data in the photographs, wherein the point data includes the action point positions, the photo abnormal values ​​are obtained by processing and analyzing, and whether the photographs are qualified is judged according to the photo abnormal values, and the photographs are marked according to the judgment results, and are marked as qualified photos and unqualified photos;

[0010] Step 3: In a rehabilitation cycle, unqualified photos are processed and analyzed to obtain qualified action values ​​HD, and rehabilitation training actions are marked according to the qualified action values ​​HD, marking them as qualified actions and unqualified actions;

[0011] Step 4: In a rehabilitation cycle, the number of unqualified movements in the rehabilitation training movement group is obtained, and the qualified movement group value G of the rehabilitation training movement group is obtained by processing and analyzing, and the corresponding rehabilitation training movement group is marked according to the qualified movement group value G, and marked as an unqualified movement group and a qualified movement group;

[0012] Step 5: Mark the rehabilitation training action group corresponding to the unqualified action group as the target analysis action group, obtain the performance data of the target analysis action group in multiple historical rehabilitation cycles, wherein the performance data includes the performance value of the number of unqualified times and the unqualified fluctuation value, obtain the recommended added value based on the performance data processing and analysis, and determine whether to recommend the target analysis action group corresponding to the recommended added value to the patient first based on the recommended added value.

[0013] As a further technical solution of the present invention, the method for obtaining the photo abnormal value BZ(n, i, j) is as follows:

[0014] The photo outlier value of the j-th photo is calculated by the formula: BZ(n,i,j)=s1×FPSB(n,i,j)+s2×PJLB(n,i,j), where s1 and s2 are preset proportional coefficients, and both s1 and s2 are greater than 0.

[0015] As a further technical solution of the present invention, the non-matching point bit ratio FPSB (n, i, j) and the average distance offset value PJLB (n, i, j) are obtained as follows:

[0016] Divide the given time DT(n, i) into j time nodes, obtain the point position in the j-th photo taken at the j-th time node, and compare it with the position of the target point. Specifically:

[0017] Mark the points in the j-th photo whose positions are different from the target points as non-salient points, and mark the points whose positions are the same as the target points as salient points;

[0018] Count the number of non-prominent points in the j-th photo taken at the j-th time node to obtain the number of non-matching points, and mark it as FPS (n, i, j);

[0019] Get the target point quantity value TC(n,i,j) through the formula: Calculate the non-matching point ratio FPSB (n, i, j) of the j-th photo;

[0020] Measure the distance between the position of the non-prominent point in the jth photo taken at the jth time node and the position corresponding to the target point, sum and average them to obtain the average distance offset value, and mark it as PJL(n, i, j), measure the length value of the taken photo, and ratio the average distance offset value PJL(n, i, j) with the length value of the taken photo to obtain the average distance offset ratio, and mark it as PJLB(n, i, j).

[0021] As a further technical solution of the present invention, a method for obtaining a marking result obtained by marking corresponding photos according to the abnormal values ​​of the photos is as follows:

[0022] Compare the photo outlier to the photo outlier threshold:

[0023] If the photo outlier value is greater than or equal to the photo outlier value threshold, the corresponding j-th photo is marked as an unqualified photo;

[0024] If the photo anomaly value is less than the photo anomaly threshold, the corresponding j-th photo is marked as a qualified photo.

[0025] As a further technical solution of the present invention, the unqualified action value HD is obtained in the following manner:

[0026] The unqualified action value HD of the rehabilitation training action in the rehabilitation training action group is calculated by the formula: HD=a1×BZB+a2×BCB, wherein a1 and a2 are both preset proportional coefficients, and a1 and a2 are both greater than 0.

[0027] As a further technical solution of the present invention, the method for obtaining the unqualified photo quantity ratio BZB and the photo abnormality degree value BCB is as follows:

[0028] According to the patient's performance in the rehabilitation training action group, the number of unqualified photos in the j photos taken corresponding to the rehabilitation training action is counted and marked as BZS;

[0029] By formula: The ratio of the number of unqualified photos of rehabilitation training actions in the rehabilitation training action group BZB is calculated;

[0030] In the unqualified photos corresponding to the rehabilitation training movements, the photo anomaly value of the unqualified photos is subtracted from the photo anomaly threshold to obtain the photo anomaly difference, the photo anomaly differences corresponding to all the unqualified photos are summed and averaged to obtain the photo anomaly difference mean, the photo anomaly difference mean is ratioed with the photo anomaly threshold to obtain the photo anomaly degree value corresponding to the rehabilitation training movements, and it is marked as BCB.

[0031] As a further technical solution of the present invention, the qualified action group value G is obtained in the following manner:

[0032] Count the number of unqualified movements in the rehabilitation training movement group and mark them as GPP, obtain the total number of rehabilitation training movements in the rehabilitation training movement group and mark it as i;

[0033] By formula: Calculate and obtain the qualified action group value G of the rehabilitation training action group.

[0034] As a further technical solution of the present invention, the marking result obtained by marking the corresponding rehabilitation training action group is obtained in the following manner:

[0035] Compare the eligible action set value to the eligible action set threshold:

[0036] If the qualified action group value is greater than the qualified action group threshold, the corresponding rehabilitation training action group is marked as an unqualified action group;

[0037] If the qualified action group value is less than or equal to the qualified action group threshold, the corresponding rehabilitation training action group is marked as a qualified action group.

[0038] As a further technical solution of the present invention, the method for obtaining the recommended added value C is:

[0039] The recommended added value C for each rehabilitation cycle of the patient is calculated by the formula: C=b1×BCBX+b2×BPJB+b3×BBD, where b1, b2, and b3 are all preset proportional coefficients, and b1, b2, and b3 are all greater than 0.

[0040] As a further technical solution of the present invention, the method of obtaining the unqualified number performance value BCBX, the unqualified action average value ratio BPJB and the unqualified fluctuation value BBD is as follows:

[0041] Counting the number of times the target analysis action group is selected in multiple rehabilitation cycles and marking it as the number of selections, counting the number of times the target analysis action group is marked as an unqualified action group in the number of selections and marking it as the number of unqualified times;

[0042] The number of unqualified times is compared with the number of selected times to obtain the unqualified times performance value of the target analysis action group, and it is marked as BCBX;

[0043] The unqualified action value when the target analysis action group is marked as an unqualified action group is subtracted from the unqualified action threshold to obtain the unqualified action value difference, all the unqualified action value differences are summed up and averaged to obtain the average unqualified action value difference, the average unqualified action value difference is ratioed with the unqualified action threshold to obtain the unqualified action average value ratio, and it is marked as BPJB;

[0044] All unqualified action values ​​obtained when the target analysis action group is marked as an unqualified action group are integrated to obtain an unqualified action value data group, and the variance value of the unqualified action value data group is obtained and marked as an unqualified fluctuation value, and marked as BBD.

[0045] The beneficial effects of the present invention are as follows:

[0046] 1. The medical staff evaluates multiple rehabilitation training action groups that are suitable for the patient based on the patient's performance, and inputs the multiple rehabilitation training action groups that are suitable for the patient into the selection terminal. During the rehabilitation cycle, the patient selects a rehabilitation training action group from the selection terminal for rehabilitation training, wherein each rehabilitation training action group contains multiple rehabilitation training actions, and each rehabilitation training action is marked as (n, i), wherein n is 1, 2, 3..., n represents the rehabilitation training action group number, and i is 1, 2, 3..., i represents the rehabilitation training action number in the rehabilitation training group. During a rehabilitation cycle, each rehabilitation training action is performed according to the predetermined time DT(n, i) of each rehabilitation training action in the rehabilitation training action group selected by the patient. Shoot to obtain photos of rehabilitation training movements, and according to the point data in the photos, wherein the point data includes the action point position, process and analyze to obtain photo abnormal values, and judge whether the photos are qualified according to the photo abnormal values, and mark the photos according to the judgment results, and mark them as qualified photos and unqualified photos. Within a rehabilitation cycle, process and analyze the unqualified photos to obtain qualified action values ​​HD, and mark the rehabilitation training movements according to the qualified action values ​​HD, and mark them as qualified movements and unqualified movements. The present invention judges whether the rehabilitation training movements performed by the patients are qualified by processing and analyzing the photos taken when the patients perform each rehabilitation training movement in the rehabilitation training movement group.

[0047] 2. Within a rehabilitation cycle, the number of unqualified actions in the rehabilitation training action group is obtained, and the qualified action group value G of the rehabilitation training action group is obtained through processing and analysis. The corresponding rehabilitation training action group is marked according to the qualified action group value G, and is marked as an unqualified action group and a qualified action group. The present invention divides the rehabilitation training action group into an unqualified action group and a qualified action group by marking the rehabilitation training action group within a rehabilitation cycle.

[0048] 3. Mark the rehabilitation training action group corresponding to the unqualified action group as the target analysis action group, obtain the performance data of the target analysis action group in multiple historical rehabilitation cycles, wherein the performance data includes the unqualified number performance value and the unqualified fluctuation value, obtain the recommended added value based on the performance data processing and analysis, and judge whether to give priority to recommending the target analysis action group corresponding to the recommended added value to the patient based on the recommended added value. The present invention judges whether to add the rehabilitation training action group corresponding to the unqualified action group to the next rehabilitation cycle of the patient by processing and analyzing the unqualified action groups in multiple historical rehabilitation cycles, which helps to adjust the rehabilitation training action group for the next cycle, thereby helping the patient to recover quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The present invention will be further described below in conjunction with the accompanying drawings.

[0050] Figure 1It is a flowchart of the steps of an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method described in an embodiment of the present invention.

[0051] Figure 2 It is a flowchart of the steps for obtaining the recommended added value of an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method described in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0053] Example 1

[0054] like Figure 1 The embodiment of the present invention shows an artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method, comprising:

[0055] Step 1: The medical staff evaluates multiple rehabilitation training action groups that are suitable for the patient based on the patient's performance, and inputs the multiple rehabilitation training action groups that are suitable for the patient into the selection terminal. During the rehabilitation period, the patient selects a rehabilitation training action group from the selection terminal for rehabilitation training, wherein each rehabilitation training action group contains multiple rehabilitation training actions, and each rehabilitation training action is marked as (n, i), wherein n is 1, 2, 3..., n represents the rehabilitation training action group number, and i is 1, 2, 3..., i represents the rehabilitation training action number in the rehabilitation training group;

[0056] Step 2: In a rehabilitation cycle, according to the predetermined time DT(n, i) of each rehabilitation training action in the rehabilitation training action group selected by the patient, each rehabilitation training action is photographed to obtain photographs of the rehabilitation training actions, and according to the point data in the photographs, wherein the point data includes the action point positions, the photo abnormal values ​​are obtained by processing and analyzing, and whether the photographs are qualified is judged according to the photo abnormal values, and the photographs are marked according to the judgment results, and are marked as qualified photos and unqualified photos;

[0057] The predetermined time DT(n, i) means: marking the rehabilitation action in the rehabilitation training action group selected by the patient as the predetermined action, marking the corresponding point in the predetermined action as the target point, and marking the set time for the patient to maintain the predetermined action as the predetermined time DT(n, i), wherein n represents the rehabilitation training action group number, and i represents the rehabilitation training action number in the rehabilitation training group;

[0058] Specifically, the non-matching point number ratio FPSB(n, i, j) and the average distance offset value PJLB(n, i, j) are obtained as follows:

[0059] Divide the given time DT(n, i) into j time nodes, obtain the point position in the j-th photo taken at the j-th time node, and compare it with the position of the target point. Specifically:

[0060] Mark the points in the j-th photo whose positions are different from the target points as non-salient points, and mark the points whose positions are the same as the target points as salient points;

[0061] Count the number of non-prominent points in the j-th photo taken at the j-th time node to obtain the number of non-matching points, and mark it as FPS (n, i, j);

[0062] Get the target point quantity value TC(n,i,j) through the formula: Calculate the non-matching point ratio FPSB (n, i, j) of the j-th photo;

[0063] Measure the distance between the position of the non-prominent point in the j-th photo taken at the j-th time node and the position corresponding to the target point, and sum and average them to obtain the average distance offset value, which is marked as PJL(n, i, j). Measure the length value of the photograph taken, and compare the average distance offset value PJL(n, i, j) with the length value of the photograph taken to obtain the average distance offset ratio, which is marked as PJLB(n, i, j).

[0064] It should be noted that each point in the photo corresponds to a target point;

[0065] The photo outlier value of the j-th photo is calculated by the formula: BZ(n,i,j)=s1×FPSB(n,i,j)+s2×PJLB(n,i,j), where s1 and s2 are both preset proportional coefficients, and both s1 and s2 are greater than 0;

[0066] Compare the photo outlier to the photo outlier threshold:

[0067] If the photo outlier value is greater than or equal to the photo outlier value threshold, the corresponding j-th photo is marked as an unqualified photo;

[0068] If the photo abnormality value is less than the photo abnormality threshold, the corresponding j-th photo is marked as a qualified photo;

[0069] Step 3: In a rehabilitation cycle, unqualified photos are processed and analyzed to obtain qualified action values ​​HD, and rehabilitation training actions are marked according to the qualified action values ​​HD, marking them as qualified actions and unqualified actions;

[0070] 031: When the patient is doing the rehabilitation training action in the rehabilitation training action group, count the number of unqualified photos in the j photos taken corresponding to the rehabilitation training action and mark them as BZS;

[0071] By formula: The ratio of the number of unqualified photos of rehabilitation training actions in the rehabilitation training action group BZB is calculated;

[0072] 032: In the unqualified photos corresponding to the rehabilitation training action, the photo abnormality value of the unqualified photo is subtracted from the photo abnormality threshold to obtain the photo abnormality difference, the photo abnormality differences corresponding to all the unqualified photos are summed and averaged to obtain the photo abnormality difference mean, the photo abnormality difference mean is ratioed with the photo abnormality threshold to obtain the photo abnormality degree value corresponding to the rehabilitation training action, and it is marked as BCB;

[0073] 033: The unqualified action value HD of the rehabilitation training action in the rehabilitation training action group is calculated by the formula: HD = a1 × BZB + a2 × BCB, wherein a1 and a2 are both preset proportional coefficients, and a1 and a2 are both greater than 0;

[0074] Compare the failure action value to the failure action threshold:

[0075] If the unqualified action value is less than the unqualified action threshold, the patient has a high degree of restoration of the given action, and the rehabilitation training action corresponding to the unqualified action value is marked as a qualified action;

[0076] If the unqualified action value is greater than or equal to the unqualified action threshold, the patient's restoration degree of the given action is low, and the rehabilitation training action corresponding to the unqualified action value is marked as an unqualified action;

[0077] It should be noted that qualified movements indicate that when the patient performs the given movements, the degree of restoration of the given movements is high, which further indicates that the patient's physical function when performing the given movements is good; unqualified movements indicate that when the patient performs the given movements, the degree of restoration of the given movements is low, which further indicates that the patient's physical function when performing the given movements is poor.

[0078] The technical solution of the embodiment of the present invention is: the medical staff evaluates multiple rehabilitation training action groups that are suitable for the patient according to the patient's performance, inputs the multiple rehabilitation training action groups that are suitable for the patient into the selection terminal, and within the rehabilitation cycle, the patient selects a rehabilitation training action group from the selection terminal for rehabilitation training, wherein each rehabilitation training action group contains multiple rehabilitation training actions, and each rehabilitation training action is marked as (n, i), wherein n is 1, 2, 3..., n represents the rehabilitation training action group number, i is 1, 2, 3..., i represents the rehabilitation training action number in the rehabilitation training group, and within a rehabilitation cycle, according to the predetermined time DT(n, i) of each rehabilitation training action in the rehabilitation training action group selected by the patient, each rehabilitation training action is performed. The exercise movements are photographed to obtain photographs of the rehabilitation training movements. According to the point data in the photographs, wherein the point data include the movement point positions, the photo abnormality values ​​are obtained through processing and analysis. Whether the photographs are qualified is judged according to the photo abnormality values. The photographs are marked according to the judgment results, and they are marked as qualified photos and unqualified photos. Within a rehabilitation cycle, the unqualified photos are processed and analyzed to obtain qualified action values ​​HD. The rehabilitation training movements are marked according to the qualified action values ​​HD, and they are marked as qualified actions and unqualified actions. The present invention determines whether the rehabilitation training movements performed by the patients are qualified by processing and analyzing the photographs taken when the patients perform each rehabilitation training movement in the rehabilitation training movement group.

[0079] Example 2

[0080] Step 4: In a rehabilitation cycle, the number of unqualified movements in the rehabilitation training movement group is obtained, and the qualified movement group value G of the rehabilitation training movement group is obtained by processing and analyzing, and the corresponding rehabilitation training movement group is marked according to the qualified movement group value G, and marked as an unqualified movement group and a qualified movement group;

[0081] Specific:

[0082] 041: Count the number of unqualified movements in the rehabilitation training movement group and mark them as GPP, obtain the total number of rehabilitation training movements in the rehabilitation training movement group and mark them as i;

[0083] By formula: Calculate and obtain the qualified action group value G of the rehabilitation training action group;

[0084] Compare the eligible action set value to the eligible action set threshold:

[0085] If the qualified action group value is greater than the qualified action group threshold, the corresponding rehabilitation training action group is marked as an unqualified action group;

[0086] If the qualified action group value is less than or equal to the qualified action group threshold, the corresponding rehabilitation training action group is marked as a qualified action group;

[0087] The technical solution of the embodiment of the present invention is: within a rehabilitation cycle, the number of unqualified actions in the rehabilitation training action group is obtained, and the qualified action group value G of the rehabilitation training action group is obtained by processing and analyzing, and the corresponding rehabilitation training action group is marked according to the qualified action group value G, and is marked as an unqualified action group and a qualified action group. The present invention divides the rehabilitation training action group into an unqualified action group and a qualified action group by marking the rehabilitation training action group within a rehabilitation cycle.

[0088] Example 2

[0089] Step 5: Mark the rehabilitation training action group corresponding to the unqualified action group as the target analysis action group, obtain the performance data of the target analysis action group in multiple historical rehabilitation cycles, wherein the performance data includes the unqualified number performance value and the unqualified fluctuation value, obtain the recommended added value based on the performance data processing and analysis, and determine whether to recommend the target analysis action group corresponding to the recommended added value to the patient first based on the recommended added value;

[0090] 051: Count the number of times the target analysis action group is selected in multiple rehabilitation cycles, and mark it as the number of selections; count the number of times the target analysis action group is marked as an unqualified action group in the number of selections, and mark it as the number of unqualified times;

[0091] The number of unqualified times is compared with the number of selected times to obtain the unqualified times performance value of the target analysis action group, and it is marked as BCBX;

[0092] 052: When the target analysis action group is marked as an unqualified action group, the unqualified action value is subtracted from the unqualified action threshold to obtain the unqualified action value difference, all the unqualified action value differences are summed up and averaged to obtain the average unqualified action value difference, the average unqualified action value difference is ratioed with the unqualified action threshold to obtain the unqualified action average value ratio, and it is marked as BPJB;

[0093] 053: Integrate all unqualified action values ​​obtained when the target analysis action group is marked as an unqualified action group to obtain an unqualified action value data group, obtain the variance value of the unqualified action value data group, and mark it as an unqualified fluctuation value, and mark it as BBD;

[0094] 054: The recommended added value C for each rehabilitation cycle of the patient is calculated by the formula: C = b1 × BCBX + b2 × BPJB + b3 × BBD, where b1, b2, and b3 are all preset proportional coefficients, and b1, b2, and b3 are all greater than 0;

[0095] Compare the recommended addition value to the recommended addition threshold:

[0096] If the recommended addition value is greater than the recommended addition threshold, the corresponding rehabilitation training action group is added to the patient's next rehabilitation cycle;

[0097] If the recommended addition value is less than or equal to the recommended addition threshold, the corresponding rehabilitation training action group will not be added to the patient's next rehabilitation cycle;

[0098] It should be noted that the recommended added value is calculated by the performance value of the number of unqualified times, the ratio of the average value of unqualified actions and the unqualified fluctuation value. The performance value of the number of unqualified times reflects the number of times the selected rehabilitation training action group is an unqualified action group in multiple historical rehabilitation cycles. The average value of unqualified actions reflects the proportion of unqualified actions in the selected rehabilitation training action group as an unqualified action group in multiple historical rehabilitation cycles. The unqualified action value data group reflects the degree of discreteness of the number of unqualified action groups selected as unqualified action groups in multiple historical rehabilitation cycles. The lower the degree of discreteness, the more stable the number of unqualified actions in the unqualified action group.

[0099] The technical solution of the embodiment of the present invention is: marking the rehabilitation training action group corresponding to the unqualified action group as the target analysis action group, obtaining the performance data of the target analysis action group in multiple historical rehabilitation cycles, wherein the performance data includes the unqualified number performance value and the unqualified fluctuation value, obtaining the recommended added value based on the performance data processing and analysis, and judging whether to give priority to recommending the target analysis action group corresponding to the recommended added value to the patient based on the recommended added value. The present invention judges whether to add the rehabilitation training action group corresponding to the unqualified action group to the next rehabilitation cycle of the patient by processing and analyzing the unqualified action groups in multiple historical rehabilitation cycles, which helps to adjust the rehabilitation training action group of the next cycle, and further can help the patient recover quickly.

[0100] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method, characterized in that: include: Step 1: The medical staff evaluates multiple rehabilitation training action groups that are suitable for the patient based on the patient's performance, and inputs the multiple rehabilitation training action groups that are suitable for the patient into the selection terminal. During the rehabilitation period, the patient selects a rehabilitation training action group from the selection terminal for rehabilitation training, wherein each rehabilitation training action group contains multiple rehabilitation training actions, and each rehabilitation training action is marked as (n, i), wherein n is 1, 2, 3..., n represents the rehabilitation training action group number, and i is 1, 2, 3..., i represents the rehabilitation training action number in the rehabilitation training group; Step 2: In a rehabilitation cycle, according to the predetermined time DT(n, i) of each rehabilitation training action in the rehabilitation training action group selected by the patient, each rehabilitation training action is photographed to obtain photographs of the rehabilitation training actions, and according to the point data in the photographs, wherein the point data includes the action point positions, the photo abnormal values ​​are obtained by processing and analyzing, and whether the photographs are qualified is judged according to the photo abnormal values, and the photographs are marked according to the judgment results, and are marked as qualified photos and unqualified photos; Step 3: In a rehabilitation cycle, unqualified photos are processed and analyzed to obtain qualified action values ​​HD, and rehabilitation training actions are marked according to the qualified action values ​​HD, marking them as qualified actions and unqualified actions; Step 4: In a rehabilitation cycle, the number of unqualified movements in the rehabilitation training movement group is obtained, and the qualified movement group value G of the rehabilitation training movement group is obtained by processing and analyzing, and the corresponding rehabilitation training movement group is marked according to the qualified movement group value G, and marked as an unqualified movement group and a qualified movement group; Step 5: Mark the rehabilitation training action group corresponding to the unqualified action group as the target analysis action group, obtain the performance data of the target analysis action group in multiple historical rehabilitation cycles, wherein the performance data includes the performance value of the number of unqualified times and the unqualified fluctuation value, obtain the recommended added value based on the performance data processing and analysis, and determine whether to recommend the target analysis action group corresponding to the recommended added value to the patient first based on the recommended added value.

2. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 1, characterized in that: The photo abnormal value BZ(n, i, j) is obtained as follows: The photo outlier value of the j-th photo is calculated by the formula: BZ(n,i,j)=s1×FPSB(n,i,j)+s2×PJLB(n,i,j), where s1 and s2 are preset proportional coefficients, and both s1 and s2 are greater than 0.

3. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 2 is characterized by: The non-matching point bit ratio FPSB (n, i, j) and the average distance offset value PJLB (n, i, j) are obtained as follows: Divide the given time DT(n, i) into j time nodes, obtain the point position in the j-th photo taken at the j-th time node, and compare it with the position of the target point. Specifically: Mark the points in the j-th photo whose positions are different from the target points as non-salient points, and mark the points whose positions are the same as the target points as salient points; Count the number of non-prominent points in the j-th photo taken at the j-th time node to obtain the number of non-matching points, and mark it as FPS (n, i, j); Get the target point quantity value TC(n,i,j) through the formula: Calculate the non-matching point ratio FPSB (n, i, j) of the j-th photo; Measure the distance between the position of the non-prominent point in the jth photo taken at the jth time node and the position corresponding to the target point, sum and average them to obtain the average distance offset value, and mark it as PJL(n, i, j), measure the length value of the taken photo, and ratio the average distance offset value PJL(n, i, j) with the length value of the taken photo to obtain the average distance offset ratio, and mark it as PJLB(n, i, j).

4. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 1, characterized in that: The method for obtaining the marking result by marking the corresponding photos according to the abnormal values ​​of the photos is as follows: Compare the photo outlier to the photo outlier threshold: If the photo outlier value is greater than or equal to the photo outlier value threshold, the corresponding j-th photo is marked as an unqualified photo; If the photo anomaly value is less than the photo anomaly threshold, the corresponding j-th photo is marked as a qualified photo.

5. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 1, characterized in that: The unqualified action value HD is obtained in the following manner: The unqualified action value HD of the rehabilitation training action in the rehabilitation training action group is calculated by the formula: HD=a1×BZB+a2×BCB, wherein a1 and a2 are both preset proportional coefficients, and a1 and a2 are both greater than 0.

6. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 5, characterized in that: The method for obtaining the ratio of the number of unqualified photos BZB and the abnormality value of the photos BCB is as follows: According to the patient's performance in the rehabilitation training action group, the number of unqualified photos in the j photos taken corresponding to the rehabilitation training action is counted and marked as BZS; By formula: The ratio of the number of unqualified photos of rehabilitation training actions in the rehabilitation training action group BZB is calculated; In the unqualified photos corresponding to the rehabilitation training movements, the photo anomaly value of the unqualified photos is subtracted from the photo anomaly threshold to obtain the photo anomaly difference, the photo anomaly differences corresponding to all the unqualified photos are summed and averaged to obtain the photo anomaly difference mean, the photo anomaly difference mean is ratioed with the photo anomaly threshold to obtain the photo anomaly degree value corresponding to the rehabilitation training movements, and it is marked as BCB.

7. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 1, characterized in that: The qualified action group value G is obtained in the following manner: Count the number of unqualified movements in the rehabilitation training movement group and mark them as GPP, obtain the total number of rehabilitation training movements in the rehabilitation training movement group and mark it as i; By formula: Calculate and obtain the qualified action group value G of the rehabilitation training action group.

8. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 7, characterized in that: The marking result obtained by marking the corresponding rehabilitation training action group is obtained as follows: Compare the eligible action set value to the eligible action set threshold: If the qualified action group value is greater than the qualified action group threshold, the corresponding rehabilitation training action group is marked as an unqualified action group; If the qualified action group value is less than or equal to the qualified action group threshold, the corresponding rehabilitation training action group is marked as a qualified action group.

9. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 1, characterized in that: The recommended added value C is obtained as follows: The recommended added value C for each rehabilitation cycle of the patient is calculated by C=b1×BCBX+b2×BPJB+b3×BBD, where b1, b2, and b3 are all preset proportional coefficients, and b1, b2, and b3 are all greater than 0.

10. The artificial intelligence-based orthopedic postoperative rehabilitation training monitoring method according to claim 9, characterized in that: The method for obtaining the unqualified number performance value BCBX, the unqualified action average value ratio BPJB and the unqualified fluctuation value BBD is as follows: Counting the number of times the target analysis action group is selected in multiple rehabilitation cycles and marking it as the number of selections, counting the number of times the target analysis action group is marked as an unqualified action group in the number of selections and marking it as the number of unqualified times; The number of unqualified times is compared with the number of selected times to obtain the unqualified times performance value of the target analysis action group, and it is marked as BCBX; The unqualified action value when the target analysis action group is marked as an unqualified action group is subtracted from the unqualified action threshold to obtain the unqualified action value difference, all the unqualified action value differences are summed up and averaged to obtain the average unqualified action value difference, the average unqualified action value difference is ratioed with the unqualified action threshold to obtain the unqualified action average value ratio, and it is marked as BPJB; All unqualified action values ​​obtained when the target analysis action group is marked as an unqualified action group are integrated to obtain an unqualified action value data group, and the variance value of the unqualified action value data group is obtained and marked as an unqualified fluctuation value, and marked as BBD.

Images