Patient pain prediction method and device based on wearable equipment

Abstract

The invention discloses a patient pain prediction method and device based on wearable equipment. The method comprises the steps that historical physiological data and activity data of a patient are collected through the wearable equipment; acquiring a pain assessment report of the patient through an electronic medical record system, wherein the pain assessment report comprises a pain intensity score, the frequency and duration of pain attack and a treatment and intervention scheme; establishing a pain multi-modal evaluation model according to the physiological data and the pain evaluation report; wherein the pain multi-modal evaluation model comprises a learning identification module, a 1-all strategy module, a time sequence prediction module and a multi-modal fusion evaluation module; inputting the current physiological data into the pain multi-mode evaluation model, and predicting the pain intensity and the pain mode of the patient. Through the model comprising rule learning, the 1-all strategy, time sequence prediction and multi-mode fusion evaluation, the prediction precision of the pain intensity and the pain mode of the patient is improved, and the diagnosis and treatment effect of the patient, especially the chronic pain patient, can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pain auxiliary diagnosis, and in particular to a patient pain prediction method and device based on a wearable device, and a computing device. Background Art

[0002] Currently, doctors usually ask patients about their pain conditions, including the time, frequency, and duration of pain, to assess the location and severity of pain. Doctors or patients often rely on regular follow-up visits or self-reporting by patients, which makes it impossible to continuously monitor the pain status and is not conducive to timely detection and treatment of new pain problems.

[0003] In recent years, a variety of wearable health monitoring devices (such as smart bracelets, watches, etc.) have appeared on the market, which can collect users' physiological parameters (such as heart rate, blood oxygen saturation, skin conductance response, etc.) and activity data (such as steps, distance, speed, etc.). However, their application in the field of medical health is still in the exploratory stage, especially in pain management. They are mostly limited to data collection and simple statistical analysis, and have not been deeply applied to the prediction and management of patient pain, especially for chronic pain, which requires long-term follow-up observation. In addition, they cannot provide personalized pain prediction solutions based on individual specific conditions.

[0004] To solve the above problems, the present invention proposes a patient pain prediction method based on wearable devices to improve the prediction accuracy of patients' pain intensity and pain pattern, thereby improving the diagnosis and treatment effect, especially for patients with chronic pain. Summary of the invention

[0005] In view of the above problems, a patient pain prediction method and device based on wearable devices and a computing device are proposed.

[0006] According to one aspect of the present invention, a method for predicting patient pain based on a wearable device is provided, comprising:

[0007] The patient's historical physiological data and activity data are collected through wearable devices; wherein the physiological data include heart rate, blood pressure and skin conductance response; the activity data include walking distance, activity time, movement trajectory and movement speed;

[0008] Obtaining the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes pain intensity score, frequency and duration of pain attacks, and treatment and intervention plan;

[0009] A pain multimodal assessment model is established according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module;

[0010] The current physiological data is input into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

[0011] In an optional manner, the pain intensity score includes the pain intensity score including an oral grading scale, a McGill pain questionnaire, and a behavioral pain scale;

[0012] The treatment and intervention program includes drug therapy, physical therapy and psychotherapy; wherein the drug therapy includes analgesics, sedatives and corresponding dosages; and the physical therapy includes hot compress, cold compress, massage and corresponding time.

[0013] In an optional manner, the learning recognition module learns according to the patient's multidimensional features and physiological indicator change rule set to obtain the patient's pain learning results; wherein the physiological indicator change rule set includes the patient's heart rate change rule, blood pressure change rule and skin conductance reaction rule; the multidimensional features include the patient's age, gender, disease category, weight, height and treatment time;

[0014] The 1-all strategy module performs fuzzy logic operation according to the patient pain learning result to obtain a first physiological state decision output result;

[0015] The time series prediction module predicts the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result; wherein the physiological data is divided into the first physiological data and the second physiological data;

[0016] The multimodal fusion evaluation module performs fusion evaluation based on the first physiological state decision output result and the second physiological state prediction result to obtain the patient's pain intensity and pain pattern.

[0017] In an optional manner, the calculation formula for the patient's pain learning result is:

[0018]

[0019] Among them, α is the weight of physiological index; β HR is the heart rate variability coefficient; β BP is the blood pressure variation coefficient; β GSR is the skin conductance response coefficient; ΔHR(t) is the heart rate change; γ is the multidimensional feature influence weight; δ A is the age coefficient; δ Gis the gender coefficient; δ D is the disease category coefficient; δ W is the body weight coefficient; δ H is the height coefficient; δ T is the treatment time coefficient; AIF is age; GIF is gender; DIF is disease category; WIF is weight; HIF is height; TIF is treatment time; η is the weight of physiological index change rate; ζ is the integral weight of multidimensional feature change rate; b is the bias term, which is used to adjust the basic pain level.

[0020] In an optional manner, performing fuzzy logic operation according to the patient's pain learning result to obtain the first physiological state decision output result further includes:

[0021] According to the fuzzy set and the corresponding membership function, an aggregated fuzzy rule function is obtained, wherein the fuzzy set includes mild pain, moderate pain and severe pain;

[0022] Defuzzifying the aggregated fuzzy rule function to obtain the first physiological state decision output result;

[0023] The calculation formula of the fuzzy processing is:

[0024]

[0025] Among them, x is the change in pain intensity; u(x) is the membership function of the fuzzy set corresponding to the change in pain intensity.

[0026] In an optional manner, predicting the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result further includes:

[0027] Testing the stability of the first physiological data by using an ADF test method;

[0028] Constructing a SARIMA prediction model according to the first physiological data and the pain assessment report;

[0029] The optimal p, d, q, P, D, Q and s combination of the SARIMA prediction model is searched according to the AIC grid search algorithm; wherein p is the order of the autoregressive term AR; d is the order of the difference; q is the order of the moving term SM; P is the order of the autoregressive average term ARA; D is the average order of the time difference; Q is the order of the moving average term SMA; s is the repetition frequency;

[0030] The second physiological data is input into the SARIMA prediction model to obtain the second physiological state prediction result.

[0031] In an optional manner, the calculation formula for the fusion evaluation is:

[0032]

[0033] Where X = [w1, w2] T [y1,y2], y1 is the output result of the first physiological state decision; y2 is the prediction result of the second physiological state; w1,w2 are the weights of y1 and y2; θ1 is the steepness factor, θ2 is the offset term; τ is the calibration coefficient.

[0034] In an optional manner, the wearable device is a smart watch, and the smart watch is provided with a photoelectric heart rate sensor, a SpO2 sensor, a skin conductance response GSR sensor, and an activity tracking APP;

[0035] Among them, the photoelectric heart rate sensor is used to monitor continuous heart rate; the SpO2 sensor is used to estimate blood pressure; the skin conductance response GSR sensor is used to measure skin conductance response; the activity tracking APP is used to record walking distance, activity time, exercise trajectory and exercise speed.

[0036] According to another aspect of the present invention, a patient pain prediction device based on a wearable device is provided, comprising:

[0037] The first acquisition module is used to collect the patient's historical physiological data and activity data through a wearable device; wherein the physiological data includes heart rate, blood pressure and skin conductance response; the activity data includes walking distance, activity time, movement trajectory and movement speed;

[0038] A second acquisition module is used to obtain the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes the pain intensity score, the frequency and duration of pain attacks, and the treatment and intervention plan;

[0039] A model building module, used to establish a pain multimodal assessment model according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module;

[0040] An assessment module is used to input current physiological data into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

[0041] According to another aspect of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus.

[0042] According to the scheme provided by the present invention, the patient's historical physiological data and activity data are collected through wearable devices; wherein the physiological data include heart rate, blood pressure and skin conductance response; the activity data include walking distance, activity time, movement trajectory and movement speed; the patient's pain assessment report is obtained through the electronic medical record system, and the pain assessment report includes pain intensity score, frequency and duration of pain attacks, and treatment and intervention plan; a pain multimodal assessment model is established according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a timing prediction module and a multimodal fusion assessment module; the current physiological data is input into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern. The present invention not only collects the patient's physiological indicators through wearable devices, but also collects activity-related data, and more comprehensively evaluates the patient's condition. Combined with the pain assessment report in the electronic medical record, the accuracy of pain prediction is improved. Through a model that includes regular learning, 1-all strategy, time series prediction and multimodal fusion evaluation, the prediction accuracy of patients' pain intensity and pain patterns is further improved, thereby significantly improving the diagnosis and treatment effects, especially for patients with chronic pain.

[0043] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0045] Figure 1 A schematic diagram showing a flow chart of a method for predicting pain in a patient based on a wearable device according to an embodiment of the present invention;

[0046] Figure 2 A schematic diagram showing a flow chart of data collection according to an embodiment of the present invention is shown;

[0047] Figure 3 A schematic diagram showing a process flow of a multimodal pain assessment model according to an embodiment of the present invention is shown;

[0048] Figure 4 A schematic diagram showing a framework of a patient pain prediction device based on a wearable device according to an embodiment of the present invention;

[0049] Figure 5 A schematic diagram of the structure of a computing device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0050] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.

[0051] Figure 1 FIG. 1 is a flow chart of a method for predicting pain in a patient based on a wearable device according to an embodiment of the present invention. Specifically, Figure 1 As shown, the following steps are included:

[0052] Step S101, collecting historical physiological data and activity data of the patient through a wearable device; wherein the physiological data includes heart rate, blood pressure and skin conductance response; and the activity data includes walking distance, activity time, movement trajectory and movement speed.

[0053] In this embodiment, the wearable device monitors the patient's heart rate, blood pressure and other physiological parameters and daily activities in real time, so that the doctor can understand the changes in the patient's physical condition in a timely manner. Compared with traditional intermittent measurements, the wearable device provides a continuous data stream, which can better observe the development trend of pain. In addition, since the wearable device is small and comfortable to wear, it will not bring additional psychological burden or discomfort to the patient.

[0054] like Figure 2As shown, the wearable device is a smart watch, which is provided with a photoelectric heart rate sensor, a SpO2 sensor, a skin conductance response GSR sensor and an activity tracking APP. Among them, the photoelectric heart rate sensor is used to monitor continuous heart rate; the SpO2 sensor is used to estimate blood pressure; the skin conductance response GSR sensor is used to measure skin conductance response; and the activity tracking APP is used to record walking distance, activity time, movement trajectory and movement speed. In this embodiment, the photoelectric heart rate sensor uses an LED light source to emit near-infrared light to irradiate the skin, and it calculates the heart rate by detecting the change in the amount of light reflected back by the receiving end. The SpO2 sensor indirectly estimates blood pressure by measuring the oxygenation level in the blood. The skin conductance response GSR sensor reflects the change in physiological state by measuring the change in skin surface resistance. When a person is in a state of tension, anxiety or excitement, the increased activity of sweat glands leads to increased skin conductivity, thereby causing the GSR value to rise. Under stress or stress, the human autonomic nervous system is activated, resulting in increased sweating, which in turn affects the GSR reading. Compared with a high-stress state, when people are in a relaxed state, the GSR value is usually lower because the sweat gland activity is weakened at this time. Activity tracking apps record users’ movement trajectory, speed, distance, and other information by combining GPS and other sensor data.

[0055] Step S102, obtaining the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes a pain intensity score, a frequency and duration of pain attacks, and a treatment and intervention plan.

[0056] In an optional manner, the pain intensity score includes the pain intensity score including an oral grading scale, a McGill pain questionnaire, and a behavioral pain scale;

[0057] The treatment and intervention program includes drug therapy, physical therapy and psychotherapy; wherein the drug therapy includes analgesics, sedatives and corresponding dosages; and the physical therapy includes hot compress, cold compress, massage and corresponding time.

[0058] In this embodiment, through a variety of assessment tools such as an oral grading score sheet, McGill pain questionnaire, and behavioral pain scale, the patient's pain perception can be fully understood from different angles, and combined with treatment and intervention programs such as drug therapy, physical therapy, and psychotherapy, the needs of different patients can be better met. Among them, the oral grading score sheet refers to a score sheet given by the doctor based on the description of the patient's own pain perception and verbal description according to the standard grading. The McGill pain questionnaire refers to a questionnaire filled out by the patient, describing the location, nature, intensity and other information of the pain in detail. The behavioral pain scale is suitable for patients who cannot clearly express their pain perception. The doctor assesses the degree of pain by observing their behavioral performance (such as facial expressions, body movements, etc.). Drug therapy refers to the doctor prescribing the corresponding analgesics and sedatives and the specific dosage indicated. Physical therapy includes physical therapies such as hot compresses and cold compresses, and stipulates the length of time for each treatment. Psychotherapy refers to providing psychological counseling to patients to help them cope with the psychological pressure caused by pain.

[0059] Step S103, establishing a pain multimodal assessment model according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module.

[0060] In this embodiment, the learning and recognition module determines the association between changes in specific physiological parameters and pain based on medical theories and clinical lesions, and can quickly identify obvious precursors to pain, providing a basis for immediate intervention. For example, "when the heart rate exceeds a certain threshold and the skin conductance response also increases, the patient is about to experience a pain attack." The 1-all strategy module uses different analysis strategies for classification and prediction for different types and characteristics of pain (such as suddenness, persistence, etc.). Specific algorithms or models can be set according to different types of pain. For example, for persistent pain, more attention is paid to the long-term trend analysis of pain; while for sudden pain, the focus is on the rapid changes in physiological parameters in a short period of time. The time series prediction module uses time series analysis technology to predict the possibility of pain occurring in the future. For example, by analyzing the time pattern in historical data, the periodic or trend characteristics of pain attacks are identified, and future pain conditions are predicted. It can help doctors and patients prepare in advance, take preventive measures, and reduce the impact of pain on quality of life. The multimodal fusion assessment module integrates multiple data sources (physiological parameters + activity data + electronic medical records) and uses multimodal data fusion methods to more comprehensively assess the patient's pain condition, avoid the deviation caused by a single data source, and further improve the prediction accuracy.

[0061] In an optional manner, the learning recognition module performs reasoning based on the patient's multidimensional features and physiological indicator change rule set to obtain the patient's pain learning result; wherein the physiological indicator change rule set includes the patient's heart rate change rule, blood pressure change rule and skin conductance response rule; the multidimensional features include the patient's age, gender, disease category, weight, height and treatment time;

[0062] The 1-all strategy module performs fuzzy logic operation according to the patient pain learning result to obtain a first physiological state decision output result;

[0063] The time series prediction module predicts the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result; wherein the physiological data is divided into the first physiological data and the second physiological data;

[0064] The multimodal fusion evaluation module performs fusion evaluation based on the first physiological state decision output result and the second physiological state prediction result to obtain the patient's pain intensity and pain pattern.

[0065] In this embodiment, the set of physiological indicator change rules is learned according to the patient's heart rate change rules, blood pressure change rules and skin conductance reaction rules to obtain the patient's pain learning results. Multidimensional features include factors such as the patient's age, gender, disease category, weight, height and treatment time. The 1-all strategy module performs fuzzy logic operations based on the patient's pain learning results to obtain the first physiological state decision output result. Fuzzy logic operations can handle uncertainty and ambiguity, and are more in line with actual conditions. The time series prediction module divides the physiological data into first physiological data and second physiological data, the former is used to train the model, and the latter is used for prediction. The second physiological data is predicted based on the first physiological data and the pain assessment report to obtain the second physiological state prediction result. The multimodal fusion evaluation module integrates the first physiological state decision output result and the second physiological state prediction result for fusion evaluation, and finally obtains the patient's pain intensity and pain pattern.

[0066] For example, the learning recognition module learns based on the patient's physiological indicator change pattern set (heart rate, blood pressure, skin conductance response), and finds that both heart rate and skin conductance response have increased, and blood pressure has also fluctuated slightly. Combined with the patient's multidimensional characteristics (age, gender, disease category, etc.), the patient's pain learning results are obtained: the patient's current physiological state indicates that pain is about to occur.

[0067] The 1-all strategy module performs fuzzy logic operations based on the pain learning results and obtains the first physiological state decision output result: the patient experiences a pain attack in the next few hours.

[0068] The time series prediction module analyzes the patient's recent first physiological data (historical physiological data) and pain assessment report, predicts the current second physiological data (real-time physiological data), and obtains the second physiological state prediction result: the patient has a high probability of pain onset in the next few hours.

[0069] The multimodal fusion evaluation module integrates the first physiological state decision output result (the result of fuzzy logic operation) and the second physiological state prediction result (the result of time series prediction) for fusion evaluation, and finally obtains the patient's pain intensity and pain pattern: it predicts that the patient will experience moderate pain in the next few hours, and the pain pattern is sudden.

[0070] The smartwatch sends warning information to the patient, suggesting that they take a proper rest or take relief measures. At the same time, the relevant information is synchronized to the doctor so that the doctor can follow up in time and make corresponding medical interventions. For example, the doctor may suggest that the patient adjust the dosage of the medication or arrange further examinations. Therefore, it can help patients predict the possibility of pain attacks in advance and take preventive measures, and also provide doctors with a more scientific basis for diagnosis and treatment decisions.

[0071] In an optional manner, the calculation formula for the patient's pain learning result is:

[0072]

[0073] Among them, α is the weight of physiological index; β HR is the heart rate variability coefficient; β BP is the blood pressure variation coefficient; β GSR is the skin conductance response coefficient; ΔHR(t) is the heart rate change; γ is the multidimensional feature influence weight; δ A is the age coefficient; δ G is the gender coefficient; δ D is the disease category coefficient; δ W is the body weight coefficient; δ H is the height coefficient; δ T is the treatment time coefficient; AIF is age; GIF is gender; DIF is disease category; WIF is weight; HIF is height; TIF is treatment time; η is the weight of physiological index change rate; ζ is the integral weight of multidimensional feature change rate; b is the bias term, which is used to adjust the basic pain level.

[0074] In this embodiment, by introducing the change rate of physiological indicators (such as the change rate of heart rate, blood pressure, and skin conductance response) and the integral of the change rate of multidimensional features (such as the integral of the change rate of age, weight, and treatment time), the dynamic changes of the patient's state can be monitored in real time, thereby more accurately reflecting the fluctuation of pain perception. Through the above-mentioned coefficients, adjustments can be made according to actual conditions to meet the pain assessment needs of different patients and different disease states. In addition, based on physiological and medical principles, through quantitative indicators and coefficients, pain perception is converted into measurable and comparable data, thereby improving the scientific nature of pain assessment.

[0075] In an optional manner, performing fuzzy logic operation according to the patient's pain learning result to obtain the first physiological state decision output result further includes:

[0076] According to the fuzzy set and the corresponding membership function, an aggregated fuzzy rule function is obtained, wherein the fuzzy set includes mild pain, moderate pain and severe pain;

[0077] Defuzzifying the aggregated fuzzy rule function to obtain the first physiological state decision output result;

[0078] The calculation formula of the fuzzy processing is:

[0079]

[0080] Among them, x is the change in pain intensity; u(x) is the membership function of the fuzzy set corresponding to the change in pain intensity.

[0081] In this embodiment, fuzzy logic operations can be used to handle the uncertainty and ambiguity in pain assessment, and multiple rule learning results can be combined to obtain a more accurate and comprehensive first physiological state decision output result, which is more precise than traditional binary logic (yes / no) or single threshold judgment.

[0082] Among them, the membership functions of mild pain, moderate pain and severe pain are:

[0083]

[0084]

[0085] Among them, a, b, c, and d are thresholds defined by the change in pain intensity.

[0086] For example, the three regularity learning results correspond to different degrees of mild pain, moderate pain, and severe pain. These results can be mapped to the corresponding fuzzy sets, and the membership of each fuzzy set can be calculated. Then, according to the aggregated fuzzy rule function (weighted summation or maximum value operation), the fuzzified pain state representation is obtained. Finally, the above defuzzification formula is used to calculate the specific pain intensity value y, for example, y = 5 (indicating moderate pain).

[0087] In an optional manner, predicting the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result further includes:

[0088] Testing the stability of the first physiological data by using an ADF test method;

[0089] Constructing a SARIMA prediction model according to the first physiological data and the pain assessment report;

[0090] The optimal p, d, q, P, D, Q and s combination of the SARIMA prediction model is searched according to the AIC grid search algorithm; wherein p is the order of the autoregressive term AR; d is the order of the difference; q is the order of the moving term SM; P is the order of the autoregressive average term ARA; D is the average order of the time difference; Q is the order of the moving average term SMA; s is the repetition frequency;

[0091] The second physiological data is input into the SARIMA prediction model to obtain the second physiological state prediction result.

[0092] In this embodiment, the ADF (Augmented Dickey-Fuller) test method is used to ensure that the first physiological data is stable before building the prediction model, and pseudo regression can be avoided. The SARIMA (Seasonal Autoregressive Integrated Moving Average) model combines the ability of the ARIMA model to process non-seasonal time series and the ability of seasonal difference to process seasonal time series, and is more suitable for the prediction of physiological data with obvious seasonal or periodic changes. The AIC (Akaike Information Criterion) grid search algorithm is used to search for the optimal parameter combination (p, d, q, P, D, Q, s) of the SARIMA model, which can automatically find the model parameters that best suit the data, avoid the subjectivity of artificially selected parameters, and improve the prediction accuracy of the model. The prediction model is constructed by combining the first physiological data and the pain assessment report, which not only takes into account the inherent laws of the physiological data, but also incorporates external factors of pain assessment, and the prediction results are more comprehensive and accurate.

[0093] For example, a set of daily heart rate data is used as the first physiological data, and the corresponding daily pain assessment report. First, the heart rate data is tested for ADF. If the data is not stable, the first-order difference can be processed to make the data stable. Next, the SARIMA model is constructed, and the seasonal parameter s=7 is set (because the data is daily, and one week is a cycle). Then, the AIC grid search algorithm is used to search for the optimal combination of p, d, q, P, D, and Q parameters. Among them, p determines how many past non-seasonal observations (i.e., observations after removing seasonal effects) are used in the model to predict the current value, which is used to capture the short-term autocorrelation in the time series. d is the number of differences required to make the non-seasonal part stable. Difference is a mathematical transformation that can eliminate the trend and seasonal components in the time series (in the case of non-seasonal differences), making the remaining part more suitable for autoregressive modeling. q determines how many past forecast errors are used in the model to correct the current forecast value. The moving average term helps capture random fluctuations in the time series and reduce noise in the forecast. P is similar to p, but it determines how many past seasonal observations (i.e., observations separated by s time units) are used in the model to predict the current value for the seasonal component to capture the seasonal autocorrelation in the time series. D is the number of differences required to make the seasonal part stable. Seasonal differences are used to eliminate seasonal trends and convert seasonal components into stable series, making it easier to model. Q is similar to q, which determines how many past seasonal forecast errors are used in the model to correct the current seasonal forecast value. S is the frequency of seasonal pattern repetition, that is, the time interval between seasonal changes. It is a parameter used to determine the interval between seasonal observations in seasonal differences and seasonal autoregression / moving average terms. In this embodiment, the SARIMA model can better capture and predict time series data with complex seasonal patterns by combining non-seasonal and seasonal autoregression, differences, moving average terms, and seasonal cycle lengths. Assume that the optimal parameter combination finally obtained is p=2, d=1, q=1, P=1, D=0, Q=1, s=7. Finally, this set of parameters is used to train the SARIMA model, and the heart rate data to be predicted is input into the model to obtain the heart rate prediction results for the next few days, that is, the second physiological state prediction results.

[0094] In an optional manner, the calculation formula for the fusion evaluation is:

[0095]

[0096] Where X = [w1, w2] T [y1,y2], y1 is the output result of the first physiological state decision; y2 is the prediction result of the second physiological state; w1,w2 are the weights of y1 and y2; θ1 is the steepness factor, θ2 is the offset term; τ is the calibration coefficient.

[0097] In this embodiment, the steepness factor θ1 plays a role in adjusting the slope of the curve in the above function. By adjusting the value of θ1, the speed of the output E of the function changing when the value of X changes can be controlled. A larger θ1 value will cause the function to quickly transition from a low value to a high value when X approaches θ2, thereby increasing the sensitivity of the evaluation result to input changes. At the same time, the steepness factor indirectly affects the position and shape of the decision boundary by adjusting the shape of the above function, so that the decision boundary can more reasonably divide the evaluation results of different categories, thereby improving the evaluation accuracy of the model.

[0098] Step S104, inputting the current physiological data into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

[0099] In this embodiment, not only the pain intensity is predicted (divided into mild, moderate, and severe levels), but also the pain pattern (such as sudden, continuous, rhythmic, and fluctuating) is further distinguished. It can provide doctors with more detailed pain information, which helps to develop more personalized pain treatment plans. By inputting current physiological data into the model in real time, instant prediction and monitoring of pain can be achieved. This is especially important for pain management scenarios that require rapid response (such as postoperative pain management, acute pain treatment, etc.), thereby improving the patient's quality of life and satisfaction.

[0100] According to the scheme provided by the present invention, the patient's historical physiological data and activity data are collected through wearable devices; wherein the physiological data include heart rate, blood pressure and skin conductance response; the activity data include walking distance, activity time, movement trajectory and movement speed; the patient's pain assessment report is obtained through the electronic medical record system, and the pain assessment report includes pain intensity score, frequency and duration of pain attacks, and treatment and intervention plan; a pain multimodal assessment model is established according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a timing prediction module and a multimodal fusion assessment module; the current physiological data is input into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern. The present invention not only collects the patient's physiological indicators through wearable devices, but also collects activity-related data, and more comprehensively evaluates the patient's condition. Combined with the pain assessment report in the electronic medical record, the accuracy of pain prediction is improved. Through a model that includes regular learning, 1-all strategy, time series prediction and multimodal fusion evaluation, the prediction accuracy of patients' pain intensity and pain patterns is further improved, thereby significantly improving the diagnosis and treatment effects, especially for patients with chronic pain.

[0101] Figure 4 The schematic diagram of the structure of the patient pain prediction device based on the wearable device according to the embodiment of the present invention is shown. The patient pain prediction device based on the wearable device comprises:

[0102] The first acquisition module is used to collect the patient's historical physiological data and activity data through a wearable device; wherein the physiological data includes heart rate, blood pressure and skin conductance response; the activity data includes walking distance, activity time, movement trajectory and movement speed;

[0103] A second acquisition module is used to obtain the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes the pain intensity score, the frequency and duration of pain attacks, and the treatment and intervention plan;

[0104] A model building module, used to establish a pain multimodal assessment model according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module;

[0105] An assessment module is used to input current physiological data into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

[0106] Figure 5 The schematic diagram of the structure of the computing device embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computing device.

[0107] like Figure 5 As shown, the computing device may include: a processor (processor) 502 , a communications interface (Communications Interface) 504 , a memory (memory) 506 , and a communication bus 508 .

[0108] The processor 502, the communication interface 504, and the memory 506 communicate with each other via the communication bus 508. The communication interface 504 is used to communicate with other devices such as a client or other server network elements. The processor 502 is used to execute the program 510, which can specifically execute the relevant steps in the above-mentioned embodiment of the patient pain prediction method based on the wearable device.

[0109] Specifically, the program 510 may include program codes, which include computer operation instructions.

[0110] The processor 502 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0111] The memory 506 is used to store the program 510. The memory 506 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0112] According to the scheme provided by the present invention, the patient's historical physiological data and activity data are collected through wearable devices; wherein the physiological data include heart rate, blood pressure and skin conductance response; the activity data include walking distance, activity time, movement trajectory and movement speed; the patient's pain assessment report is obtained through the electronic medical record system, and the pain assessment report includes pain intensity score, frequency and duration of pain attacks, and treatment and intervention plan; a pain multimodal assessment model is established according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a timing prediction module and a multimodal fusion assessment module; the current physiological data is input into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern. The present invention not only collects the patient's physiological indicators through wearable devices, but also collects activity-related data, and more comprehensively evaluates the patient's condition. Combined with the pain assessment report in the electronic medical record, the accuracy of pain prediction is improved. Through a model that includes regular learning, 1-all strategy, time series prediction and multimodal fusion evaluation, the prediction accuracy of patients' pain intensity and pain patterns is further improved, thereby significantly improving the diagnosis and treatment effects, especially for patients with chronic pain.

[0113] Those skilled in the art will appreciate that the modules in the devices in the embodiments can be adaptively changed and set in one or more devices different from the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and in addition they can be divided into multiple submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner can be combined in any combination. Unless otherwise explicitly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) can be replaced by an alternative feature that provides the same, equivalent or similar purpose. In addition, those skilled in the art will appreciate that although some embodiments herein include certain features included in other embodiments but not other features, the combination of features of different embodiments means being within the scope of the present invention and forming different embodiments. For example, in the following claims, any one of the claimed embodiments can be used in any combination. The present invention can be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. The steps in the above embodiments should not be understood as limiting the execution order unless otherwise specified.

Claims

1. A patient pain prediction method based on wearable devices, characterized in that: include: The patient's historical physiological data and activity data are collected through wearable devices; wherein the physiological data include heart rate, blood pressure and skin conductance response; the activity data include walking distance, activity time, movement trajectory and movement speed; Obtaining the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes pain intensity score, frequency and duration of pain attacks, and treatment and intervention plan; A pain multimodal assessment model is established according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module; The current physiological data is input into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

2. The patient pain prediction method based on wearable device according to claim 1, characterized in that: The pain intensity score includes the oral grading scale, the McGill pain questionnaire and the behavioral pain scale; The treatment and intervention program includes drug therapy, physical therapy and psychotherapy; wherein the drug therapy includes analgesics, sedatives and corresponding dosages; and the physical therapy includes hot compress, cold compress, massage and corresponding time.

3. The patient pain prediction method based on wearable device according to claim 1 or 2, characterized in that: The learning recognition module learns according to the patient's multidimensional features and physiological indicator change rule set to obtain the patient's pain learning results; wherein the physiological indicator change rule set includes the patient's heart rate change rule, blood pressure change rule and skin conductance reaction rule; the multidimensional features include the patient's age, gender, disease category, weight, height and treatment time; The 1-all strategy module performs fuzzy logic operation according to the patient pain learning result to obtain a first physiological state decision output result; The time series prediction module predicts the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result; wherein the physiological data is divided into the first physiological data and the second physiological data; The multimodal fusion evaluation module performs fusion evaluation based on the first physiological state decision output result and the second physiological state prediction result to obtain the patient's pain intensity and pain pattern.

4. The patient pain prediction method based on wearable device according to claim 3, characterized in that: The calculation formula of the patient's pain learning result is: Among them, α is the weight of physiological index; β HR is the heart rate variability coefficient; β BP is the blood pressure variation coefficient; β GSR is the skin conductance response coefficient; ΔHR(t) is the heart rate change; γ is the multidimensional feature influence weight; δ A is the age coefficient; δ G is the gender coefficient; δ D is the disease category coefficient; δ W is the body weight coefficient; δ H is the height coefficient; δ T is the treatment time coefficient; AIF is age; GIF is gender; DIF is disease category; WIF is weight; HIF is height; TIF is treatment time; η is the weight of physiological index change rate; ζ is the integral weight of multidimensional feature change rate; b is the bias term, which is used to adjust the basic pain level.

5. The patient pain prediction method based on wearable device according to claim 3, characterized in that: The step of performing fuzzy logic operation according to the patient's pain learning result to obtain the first physiological state decision output result further includes: According to the fuzzy set and the corresponding membership function, an aggregated fuzzy rule function is obtained, wherein the fuzzy set includes mild pain, moderate pain and severe pain; Defuzzifying the aggregated fuzzy rule function to obtain the first physiological state decision output result; The calculation formula of the fuzzy processing is: Among them, x is the change in pain intensity; u(x) is the membership function of the fuzzy set corresponding to the change in pain intensity.

6. The patient pain prediction method based on wearable device according to claim 3, characterized in that: The predicting of the second physiological data according to the first physiological data and the pain assessment report to obtain a second physiological state prediction result further includes: Testing the stability of the first physiological data by using an ADF test method; Constructing a SARIMA prediction model according to the first physiological data and the pain assessment report; The optimal p, d, q, P, D, Q and s combination of the SARIMA prediction model is searched according to the AIC grid search algorithm; wherein p is the order of the autoregressive term AR; d is the order of the difference; q is the order of the moving term SM; P is the order of the autoregressive average term ARA; D is the average order of the time difference; Q is the order of the moving average term SMA; s is the repetition frequency; The second physiological data is input into the SARIMA prediction model to obtain the second physiological state prediction result.

7. The patient pain prediction method based on wearable device according to claim 3, characterized in that: The calculation formula of the fusion evaluation is: Where X = [w1, w2] T [y1,y2], y1 is the output result of the first physiological state decision; y2 is the prediction result of the second physiological state; w1,w2 are the weights of y1 and y2; θ1 is the steepness factor, θ2 is the offset term; τ is the calibration coefficient.

8. The method for predicting pain in patients based on a wearable device according to any one of claims 1 to 7, characterized in that: The wearable device is a smart watch, which is provided with a photoelectric heart rate sensor, a SpO2 sensor, a skin conductance response GSR sensor and an activity tracking APP; Among them, the photoelectric heart rate sensor is used to monitor continuous heart rate; the SpO2 sensor is used to estimate blood pressure; the skin conductance response GSR sensor is used to measure skin conductance response; the activity tracking APP is used to record walking distance, activity time, exercise trajectory and exercise speed.

9. A patient pain prediction device based on a wearable device, characterized in that: include: The first acquisition module is used to collect the patient's historical physiological data and activity data through a wearable device; wherein the physiological data includes heart rate, blood pressure and skin conductance response; the activity data includes walking distance, activity time, movement trajectory and movement speed; A second acquisition module is used to obtain the patient's pain assessment report through the electronic medical record system, wherein the pain assessment report includes the pain intensity score, the frequency and duration of pain attacks, and the treatment and intervention plan; A model building module, used to establish a pain multimodal assessment model according to the physiological data and the pain assessment report; wherein the pain multimodal assessment model includes a learning recognition module, a 1-all strategy module, a time series prediction module and a multimodal fusion assessment module; An assessment module is used to input current physiological data into the pain multimodal assessment model to predict the patient's pain intensity and pain pattern; wherein the pain intensity includes mild level, moderate level and severe level; the pain pattern includes sudden pain pattern, continuous pain pattern, rhythmic pain pattern and fluctuating pain pattern.

10. A computing device comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned patient pain prediction method based on wearable device.

Images