CN-121983319-A - Automatic scoring method and system for anorectal patient rehabilitation compliance

CN121983319ACN 121983319 ACN121983319 ACN 121983319ACN-121983319-A

Abstract

The invention belongs to the technical field of medical informatization, and particularly discloses an automatic scoring method and an automatic scoring system for rehabilitation compliance of anorectal patients, wherein the method comprises the steps of acquiring multidimensional rehabilitation behavior original data under active participation of the patients through a multi-element sensing acquisition terminal deployed in a patient rehabilitation environment, uploading the multidimensional rehabilitation behavior original data to an edge computing gateway, and cleaning and feature extracting the original data through a preset signal processing algorithm to construct a rehabilitation behavior space-time feature vector, wherein the multidimensional rehabilitation behavior original data comprises hip bath behavior parameters, pelvic floor muscle function exercise parameters, medication execution parameters and diet structure image data.

Inventors

TIAN JUN
YANG XIANGDONG
CHEN XIAOCHAO

Assignees

成都肛肠专科医院
成都思酷智能科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. An automatic scoring method for anorectal patient rehabilitation compliance, comprising: Acquiring multidimensional rehabilitation behavior original data of a patient under active participation by a multi-element sensing acquisition terminal deployed in a rehabilitation environment of the patient, wherein the multidimensional rehabilitation behavior original data comprises hip bath behavior parameters, pelvic floor muscle function exercise parameters, medication execution parameters and meal structure image data; Uploading the multidimensional rehabilitation behavior original data to an edge computing gateway, cleaning and extracting features of the original data through a preset signal processing algorithm, and constructing a rehabilitation behavior space-time feature vector; Inputting the rehabilitation behavior space-time feature vector into a pre-constructed compliance evaluation deep learning model, carrying out weight distribution on rehabilitation behavior features with different dimensions by the deep learning model based on a multi-head attention mechanism, and calculating to obtain initial compliance scores with all the single dimensions; Acquiring individual physiological state reference data of a patient, generating weight adjustment coefficients of each dimension according to the individual physiological state reference data by utilizing a fuzzy logic-based correction algorithm, dynamically weighting the initial compliance scores, and generating correction scores of each rehabilitation dimension; and carrying out weighted summation on the correction scores of all the dimensions according to a preset clinical rehabilitation path weight matrix to obtain a comprehensive score of rehabilitation compliance, and predicting the evolution trend of the rehabilitation compliance of the patient by combining a hidden Markov model.
2. The automatic scoring method for rehabilitation compliance of anorectal patients according to claim 1, wherein the obtaining of the hip bath behavior parameters comprises obtaining a pressure value, a liquid level height and a hip bath water temperature respectively by using a pressure sensor, a liquid level sensor and a thermistor sensor which are integrated at the bottom of the intelligent bidet; when the pressure value exceeds a preset patient weight threshold lower limit, the liquid level height is in a preset standard depth interval, and the hip bath water temperature is maintained between 38 ℃ and 42 ℃, the system judges an effective hip bath state and records single hip bath duration, hip bath frequency and hip bath temperature stability.
3. The automatic grading method for rehabilitation compliance of anorectal patients according to claim 1, wherein the acquiring pelvic floor muscle function exercise parameters comprises the steps of capturing an electric signal of the patient when performing Kegel exercise by using a surface myoelectric electrode through a wearable pelvic floor muscle electric acquisition device, and acquiring a contraction pressure signal by using a flexible pressure sensor arranged at the sphincter position; the pelvic floor muscle function exercise parameters comprise contraction peak value, duration time, contraction frequency and muscle fatigue index; The wearable pelvic floor myoelectricity acquisition equipment adopts a underpants type structure, a flexible fabric electrode is fixed at the position of the perineum corresponding to the pelvic floor muscle, and the electrode is filtered and amplified through a signal processing circuit to inhibit motion artifacts and power frequency interference.
4. The automatic grading method for anorectal patient rehabilitation compliance according to claim 1, wherein the acquiring of the meal structure image data comprises acquiring food comparison images before and after patient eating through an image acquisition module of a mobile terminal, and performing target detection and segmentation on food types in the images by using a convolutional neural network; for mixed dishes, the system provides a food splitting option on the interactive interface, and the patient is assisted to confirm the main food materials in the dishes; The pixel-level depth information is combined to calculate the volume change of the food, and the estimated intake of cellulose, protein and moisture is deduced from the pre-established food composition database.
5. The automatic scoring method for anorectal patient rehabilitation compliance according to claim 1, wherein the cleaning and feature extraction of the raw data comprises filtering high frequency noise in the pressure signal by a five-point cubic smoothing algorithm; extracting frequency domain characteristics of the pelvic floor electromyographic signals by utilizing short-time Fourier transformation; reducing the dimension of the feature vector by utilizing linear discriminant analysis, and extracting a feature subset with highest correlation with a clinical rehabilitation target, wherein the label data required by the linear discriminant analysis is defined by clinical specialists according to the compliance grade of patient rehabilitation outcome assessment; aiming at the meal image, adopting a multi-scale feature fusion strategy to extract the texture, color and geometric shape features of the food.
6. The method for automatically scoring the rehabilitation compliance of anorectal patients according to claim 1, wherein the compliance assessment deep learning model adopts a two-way long-short-term memory network as a feature extractor to capture the time sequence dependency of rehabilitation behaviors, and a multi-head attention mechanism layer is accessed after the two-way long-short-term memory network layer, and behavior sequences with key influences on postoperative complications prevention are automatically identified through the mechanism model, so that initial compliance scores of all dimensions are output.
7. The automatic scoring method for anorectal patient rehabilitation compliance according to claim 1, wherein the individual physiological state reference data comprise age, basic disease, operation mode, wound healing grading and postoperative pain scoring of the patient, the physiological indexes are used as front pieces by a fuzzy reasoning system, membership functions are established, weight adjustment coefficients of all dimensions are output, the weight adjustment coefficients range from 0.5 to 1.5, and the default value is 1.0.
8. The automatic anorectal patient recovery compliance scoring method of claim 1, wherein the recovery compliance composite score is obtained by weighted summation of corrected scores for each dimension, the weighted summation weights being weights in a preset clinical recovery path weight matrix; the hidden Markov model takes the compliance comprehensive score as an observation sequence, takes the rehabilitation state of the patient as a hidden state, and predicts the compliance evolution trend of the patient in the future set days.
9. An anorectal patient rehabilitation compliance automatic scoring system for performing the method of any of claims 1-8, the system comprising: The data acquisition layer consists of an intelligent hip bath monitoring terminal, pelvic floor myoelectricity acquisition wearing equipment, a mobile diet recording module and an intelligent medicine box and is used for acquiring full-dimension rehabilitation behavior data of a patient; The data transmission layer comprises a Bluetooth low-power module, a Wi-Fi module and an NB-IoT communication gateway and is used for transmitting acquired data to a back-end server; the logic processing layer is deployed on the cloud server and comprises a signal preprocessing module, a characteristic engineering module, a compliance evaluation model calculation unit and a clinical knowledge base storage unit; And the application interaction layer provides a medical care end management interface and a patient end feedback interface for displaying scoring results, early warning information and personalized rehabilitation advice.
10. The automatic anorectal patient recovery compliance scoring system of claim 9, further comprising a closed loop intervention engine, wherein when the recovery compliance composite score is continuously below a preset pre-alarm threshold, the system automatically invokes preset intervention logic to push recovery guidance information to the patient by applying an interaction layer and to send risk notification information synchronously to the healthcare end.

Description

Automatic scoring method and system for anorectal patient rehabilitation compliance Technical Field The invention relates to the technical field of medical informatization, in particular to an automatic scoring method and system for rehabilitation compliance of anorectal patients. Background Anorectal postoperative rehabilitation management is critical to the final prognosis of the patient. With the development of medical informatization, various postoperative management methods and systems are developed, but the prior art still has a plurality of defects. The first prior art (CN 118116599B) discloses a rehabilitation management method and system applied after anorectal operation, by obtaining medical history data and preoperative body data of a patient, calculating analgesic drug dosage, diet demand and body evaluation index by using mathematical formulas, and generating a drug management table, a diet instruction manual and a exercise planning scheme. However, the method mainly relies on medical records and static measurements, lacks real-time monitoring of daily rehabilitation behaviors (such as hip bath and pelvic floor muscle exercise) of patients, and cannot objectively quantify the execution degree of the patient on medical advice, namely, rehabilitation compliance. The evaluation basis is still data input by medical staff, but not real behavior data of patients, and subjective deviation is difficult to avoid. In the second prior art (CN 119400344B), a construction method and a system based on an anorectal surgery postoperative care rehabilitation assessment model are provided, physiological function data are acquired through means of wound morphology image analysis, anorectal pressure measurement, gait analysis and the like, and the model is constructed to predict the rehabilitation state. The technology adopts a sensor and an image processing technology, but the acquisition process is finished by professional equipment or in a hospital, continuous monitoring in a home environment cannot be realized, and the evaluation object is the physiological function recovery condition instead of the execution behavior of a patient on a rehabilitation doctor advice, so that the technology cannot be directly used for compliance scoring. The third prior art (CN 120824025A) discloses an anorectal postoperative intelligent follow-up management method and system based on a mobile terminal, wherein a comprehensive risk value is obtained by quantitative weighting and summing of clinical indexes, and a follow-up interval is dynamically adjusted according to iterative optimization weights of real-time recovery data. The method realizes the individual adjustment of the follow-up period, but the basis of weight adjustment is to recover data (such as wound reduction percentage) so as to optimize follow-up frequency instead of carrying out fairness correction on patient compliance, and the data source still depends on manual uploading (such as wound photos and VAS scores) of a patient end, so that automatic monitoring of recovery behaviors (such as hip bath time, pelvic floor muscle exercise times and medication condition) is not realized, and the data objectivity is insufficient. In summary, the prior art does not solve the following technical problems of how to realize automatic, objective and continuous monitoring of multidimensional rehabilitation behaviors (hip bath, pelvic floor muscle exercise, medication and diet) of anorectal patients in a home environment, how to automatically quantify the execution degree (rehabilitation compliance) of the patients to medical advice based on monitoring data, and how to perform fairness correction on compliance scores by combining individual physiological differences, thereby realizing accurate early warning and closed-loop intervention. The present invention is directed to the above-mentioned technical problem. Disclosure of Invention The invention provides an anorectal patient rehabilitation compliance automatic scoring method and system, and aims to solve the technical problems that in the prior art, the anorectal postoperative rehabilitation compliance assessment depends on subjective statement of a patient to cause data distortion, monitoring discretization cannot be continuous, individual difference is ignored to cause scoring to be not public, and real-time early warning intervention capability is lacked. In order to solve the technical problems, the invention adopts the following technical scheme: A anorectal patient rehabilitation compliance automatic scoring method comprises the steps of obtaining multi-dimensional rehabilitation behavior original data of a patient under active participation through a multi-element sensing acquisition terminal deployed in a patient rehabilitation environment, wherein the multi-dimensional rehabilitation behavior original data comprise hip bath behavior parameters, pelvic floor muscle function exercise parameters, medication execution par