Quantitative early diagnosis system for Parkinson's disease based on behavior feature recognition

Abstract

The invention discloses a quantitative early diagnosis system for Parkinson's disease based on behavior feature recognition, and belongs to the technical field of Parkinson's disease diagnosis, and the system comprises an intelligent collection device which is used for collecting motion data, voice data, writing data and face data of a patient, determining real-time behavior data of the patient based on the Internet of Things, and sending the real-time behavior data to the intelligent collection device; patient behavior real-time data based on the Internet of Things is transmitted to a given quantitative diagnosis platform based on a wireless network; and the quantitative diagnosis platform is used for preprocessing and analyzing the real-time data of the behaviors of the patient based on the Internet of Things, performing quantitative early diagnosis on the Parkinson's disease of the patient, and performing intelligent management and control treatment on the patient according to the diagnosis result of the patient. The method solves the problem that the diagnosis effect of the Parkinson's disease patient is poor because the Parkinson's disease quantitative early diagnosis cannot be carried out on the patient based on behavior feature recognition in the prior art. According to the invention, quantitative early diagnosis of Parkinson's disease can be carried out on the patient based on behavior feature recognition, and the diagnosis effect of the Parkinson's disease patient can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of Parkinson's diagnosis, and specifically to a quantitative early diagnosis system for Parkinson's based on behavioral feature recognition. Background Art

[0002] Parkinson's disease is a complex syndrome involving multiple organs, multiple systems, and multiple neurotransmitters, and is not a simple movement disorder disease. In patients with advanced Parkinson's disease, the progression of the disease itself, the side effects of drugs, movement complications, and many non-motor symptoms such as depression, sleep disorders, pain, and fatigue are intertwined, greatly increasing the treatment difficulty. Most patients with advanced Parkinson's disease are unable to take care of themselves and need home treatment. Therefore, early diagnosis is crucial for delaying the progression of the disease and improving the quality of life of patients.

[0003] Chinese Patent No. CN117253592A discloses a whole-course management model for Parkinson's disease, which consists of a management team composed of Parkinson's disease specialists in tertiary hospitals and community doctors specializing in Parkinson's disease. The Parkinson's disease specialists in tertiary hospitals train the community doctors specializing in Parkinson's disease to achieve homogenization in the diagnosis and treatment of Parkinson's disease; the community doctors specializing in Parkinson's disease are responsible for the implementation and follow-up of the treatment plans for all enrolled patients, and are responsible for the home treatment of patients with advanced Parkinson's disease who are unable to take care of themselves; information sharing is carried out between tertiary hospitals and communities through a Parkinson's disease management platform, and the communication between doctors and patients is mediated by WeChat. This management model is led by Parkinson's disease specialists, centered on community doctors specializing in Parkinson's disease, and patient-centered, providing multidisciplinary, whole-course, individualized, and online-offline integrated management for patients with early, middle, and advanced Parkinson's disease; however, this patent has the following defects: The existing technology cannot perform quantitative early diagnosis of Parkinson's disease on patients based on behavioral feature recognition, resulting in poor diagnostic effects for Parkinson's patients. Summary of the Invention

[0004] The purpose of the present invention is to provide a quantitative early diagnosis system for Parkinson's based on behavioral feature recognition, which can perform quantitative early diagnosis of Parkinson's disease on patients based on behavioral feature recognition, and can improve the diagnostic effects of Parkinson's patients, solving the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions: A quantitative early diagnosis system for Parkinson's based on behavioral feature recognition, the system includes: an intelligent acquisition device and a quantitative diagnosis platform. The intelligent acquisition device is configured to collect the behavioral features of a patient during the patient's behavioral activities, determine the real-time data of the patient's behavior, and transmit the collected real-time data of the patient's behavior to the quantitative diagnosis platform based on wireless transmission technology; Among them, it is comprehensively judged whether to trigger the writing photographing device to take a photo by the corresponding values of the writing tremor times and the writing pressure; The quantitative diagnosis platform is configured to construct a model for early quantitative diagnosis of Parkinson's based on the patient's behavioral characteristics, and analyze the real-time data of the patient's behavior and perform early quantitative diagnosis of Parkinson's based on the constructed model.

[0006] Preferably, the intelligent acquisition device includes: a motion acquisition unit, a voice acquisition unit, a writing acquisition unit, and a face acquisition unit; The motion acquisition unit is configured to monitor and collect the gait data, balance data, and finger tapping data of the patient in real time based on the Internet of Things technology, and obtain the patient's motion data; The voice acquisition unit is configured to monitor and collect the voice characteristics and voice content of the patient in real time based on the Internet of Things technology, and obtain the patient's voice data; The writing acquisition unit is configured to monitor and collect the writing dynamics data and writing graphic data of the patient in real time based on the Internet of Things technology, and obtain the patient's writing data; The face acquisition unit is configured to monitor and collect the facial motion data and facial expression data of the patient in real time based on the Internet of Things technology, and obtain the patient's face data; Among them, based on the collected patient's motion data, patient's voice data, patient's writing data, and patient's face data, the real-time data of the patient's behavior based on the Internet of Things is determined.

[0007] Preferably, the writing acquisition unit includes an intelligent writing pen and a writing photographing device, wherein a wireless communication connection is established between the intelligent writing pen and the writing photographing device: The intelligent writing pen collects the writing tremor times and writing pressure of the patient within a unit time in real time; The writing tremor times and the writing pressure are respectively compared with a preset writing tremor times threshold and a writing pressure threshold; When the writing tremor times exceed the preset writing tremor times threshold, but the writing pressure does not exceed the writing pressure threshold, the writing photographing device is triggered by the intelligent writing pen to take a photo, and a first writing image is obtained; When the writing tremor times do not exceed the preset writing tremor times threshold, but the writing pressure exceeds the writing pressure threshold, it is comprehensively judged whether to trigger the writing photographing device to take a photo by the corresponding values of the writing tremor times and the writing pressure; When both the writing tremor times and the writing pressure do not exceed their corresponding writing tremor times threshold and writing pressure threshold, the writing photographing device is controlled to take a photo according to a preset image acquisition frequency; a second writing image is obtained; Set the evaluation weight value of the first writing image to be higher than that of the second writing image.

[0008] Preferably, it is comprehensively judged whether to trigger the writing photographing device to take a photo by the number of writing tremors and the corresponding value of writing pressure, including: When the writing pressure exceeds the writing pressure threshold, extract the number of writing tremors corresponding to the unit time when the number of writing tremors does not exceed the preset number of writing tremors threshold as the target tremor number data set; Generate a comprehensive writing state parameter by using the target tremor number data set and the writing pressure; wherein, the comprehensive writing state parameter is used to evaluate whether it is necessary to collect an image of the current patient's writing state; Wherein, the comprehensive writing state parameter is obtained by the following formula: Wherein, S represents the comprehensive writing state parameter; P represents the writing pressure corresponding to exceeding the writing pressure threshold; P y represents the writing pressure threshold; n represents the number of data included in the target tremor number data set, wherein the number of data is the same as the number of unit times when the number of writing tremors does not exceed the preset number of writing tremors threshold; C i represents the tremor number value corresponding to the i-th tremor data in the target tremor number data set; N represents the number of unit times experienced by the patient in writing; n0 represents the reference value of the preset number of unit times; C y represents the preset number of writing tremors threshold; Compare the comprehensive writing state parameter with the preset parameter threshold; When the comprehensive writing state parameter is lower than the preset parameter threshold, it is determined that there is no need to trigger the writing photographing device to take a photo; When the comprehensive writing state parameter is not lower than the preset parameter threshold, it is determined that it is necessary to trigger the writing photographing device to take a photo; Take the writing image obtained when the number of writing tremors does not exceed the preset number of writing tremors threshold, but the writing pressure exceeds the writing pressure threshold as the third writing image; Set the evaluation weight value of the third writing image to be higher than that of the second writing image, but lower than that of the first writing image.

[0009] Preferably, the gait data includes step length, step width, step frequency, walking speed, swing phase, stance phase, gait symmetry, gait variability, trunk swing, arm swing and turning gait; The balance data includes standing balance, sitting balance, center of gravity swing and postural stability; Finger tapping data includes tapping frequency, rhythm, intensity, and finger coordination; Voice features include fundamental frequency, intensity, duration, speech rate, pauses, intonation, pronunciation clarity, and speech fluency; Speech content includes vocabulary, grammar structure, semantic expression, and emotional expression; Writing dynamics data includes writing speed, acceleration, and pressure, stroke length, width, and angle, writing pauses, and tremors; Writing graphic data includes font size, shape, and spacing, handwriting neatness, and coherence; Facial movement data includes movements of the eyebrows, eyes, nose, and mouth; Facial expression data includes the frequency, amplitude, and duration of expression changes.

[0010] Preferably, the quantitative diagnosis platform includes: a behavior data preprocessing unit and an early diagnosis and evaluation unit; The behavior data preprocessing unit is configured to clean, transform, integrate, and extract features from the real-time behavior data of patients based on the Internet of Things, and determine the real-time behavior feature data of patients based on the Internet of Things; The early diagnosis and evaluation unit is configured to analyze the real-time behavior feature data of patients based on the Internet of Things, and perform quantitative early diagnosis of Parkinson's disease on patients. According to the diagnosis results of patients, intelligent management and treatment are carried out on patients.

[0011] Preferably, the behavior data preprocessing unit includes: a behavior data cleaning module and a behavior data transformation module; The behavior data cleaning module is configured to clean the real-time behavior data of patients based on the Internet of Things; Among them, the real-time behavior data of patients based on the Internet of Things is checked based on the Python library, and duplicate values, missing values, and outliers in the real-time behavior data of patients based on the Internet of Things are identified, and the duplicate values, missing values, and outliers in the real-time behavior data of patients based on the Internet of Things are deleted; The behavior data transformation module is configured to transform the real-time behavior data of patients based on the Internet of Things; Among them, the real-time behavior data of patients based on the Internet of Things is normalized based on Z-Score standardization, and the real-time behavior data of patients based on the Internet of Things is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, removing the dimensional differences between the real-time behavior data of patients based on the Internet of Things.

[0012] Preferably, the behavior data preprocessing unit further includes: a behavior data integration module and a behavior data extraction module; The behavior data integration module is configured to integrate the real-time behavior data of patients based on the Internet of Things; Among them, based on the Python library, the real-time data of patients' behaviors based on the Internet of Things is integrated, and the real-time data of patients' behaviors based on the Internet of Things is integrated into a unified data view, and the integrated real-time data of patients' behaviors based on the Internet of Things is verified to determine whether there is a lack of the integrated real-time data of patients' behaviors based on the Internet of Things; The behavior data extraction module is configured to extract the real-time data of patients' behaviors based on the Internet of Things; Among them, features related to Parkinson's disease are extracted from the real-time data of patients' behaviors based on the Internet of Things, and the extracted feature vectors are weighted and fused to determine the behavior feature data of patients based on the Internet of Things.

[0013] Preferably, the early diagnosis and evaluation unit includes: a diagnosis model training module, and the diagnosis model training module is configured to train a quantitative early diagnosis model for Parkinson's disease; According to the quantitative early diagnosis requirements of Parkinson's disease based on behavior feature recognition, the historical data of patients' behaviors based on the Internet of Things is collected, and the collected historical data of patients' behaviors based on the Internet of Things is divided into a training set and a test set; Based on deep learning technology, the training set is used to train the deep learning model, so that the deep learning model autonomously learns the quantitative early diagnosis process of Parkinson's disease based on behavior feature recognition, and determines the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition; Based on the test set, the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition is tested, and the performance of the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition is evaluated based on indicators such as accuracy, recall rate, and F1 value, and it is judged whether the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition can achieve the expected effect; According to the test evaluation results, the parameters and structure of the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition are adjusted, and through continuous iterative optimization, the optimal quantitative early diagnosis model of Parkinson's disease is determined.

[0014] Preferably, the early diagnosis and evaluation unit further includes: a diagnosis evaluation control module; the diagnosis evaluation control module is configured to perform quantitative early diagnosis and evaluation control on patients; Obtain the optimal quantitative early diagnosis model of Parkinson's disease, and deploy the optimal quantitative early diagnosis model of Parkinson's disease in the actual quantitative early diagnosis environment of Parkinson's disease based on behavior feature recognition; Input the patient behavior characteristic data based on the Internet of Things into the optimal Parkinson's quantitative early diagnosis model, analyze the patient behavior characteristic data based on the Internet of Things by the optimal Parkinson's quantitative early diagnosis model, and conduct Parkinson's quantitative early diagnosis on the patient to determine the patient's diagnosis result; Among them, based on the patient's diagnosis result, analyze the patient behavior characteristic data to determine the patient's condition assessment report, conduct intelligent management and control treatment on the patient based on the patient's condition assessment report, and conduct real-time monitoring on the treatment effect, and dynamically adjust the patient's management and control treatment plan according to the monitoring situation.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The present invention determines the real-time patient behavior data based on the Internet of Things by collecting the patient's motion data, patient voice data, patient writing data and patient facial data, determines the patient behavior characteristic data based on the Internet of Things by cleaning, transforming, integrating and feature extracting the real-time patient behavior data based on the Internet of Things, trains the Parkinson's quantitative early diagnosis model based on deep learning technology, analyzes the patient behavior characteristic data based on the Internet of Things by the Parkinson's quantitative early diagnosis model, conducts Parkinson's quantitative early diagnosis on the patient to determine the patient's diagnosis result, analyzes the patient behavior characteristic data based on the patient's diagnosis result to determine the patient's condition assessment report, conducts intelligent management and control treatment on the patient based on the patient's condition assessment report, and conducts real-time monitoring on the treatment effect, and dynamically adjusts the patient's management and control treatment plan according to the monitoring situation. It can conduct Parkinson's quantitative early diagnosis on the patient based on behavior feature recognition and improve the diagnosis effect of Parkinson's patients. Description of the Drawings

[0016] Figure 1 It is the module structure diagram of the Parkinson's quantitative early diagnosis system based on behavior feature recognition of the present invention. Detailed Embodiments

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0018] To solve the problem that the existing technology cannot conduct Parkinson's quantitative early diagnosis on patients based on behavior feature recognition, resulting in poor diagnosis effect for Parkinson's patients, please refer to Figure 1 , the following technical solutions are provided in this embodiment: The Parkinson's quantitative early diagnosis system based on behavior feature recognition includes: an intelligent acquisition device and a quantitative diagnosis platform.

[0019] Among them, the intelligent acquisition device is used to collect the patient's movement data, voice data, writing data, and facial data, determine the real-time patient behavior data based on the Internet of Things, and transmit the collected real-time patient behavior data based on the Internet of Things to the quantitative diagnosis platform through wireless transmission technology; In this embodiment, the intelligent acquisition device includes: A motion acquisition unit, which is used to monitor and collect the gait data, balance data, and finger tapping data of the patient in real time based on the Internet of Things technology to obtain the patient's movement data; It should be noted that the gait data includes step length, step width, step frequency, walking speed, swing phase, support phase, gait symmetry, gait variability, trunk swing, arm swing, and turning gait.

[0020] It should be noted that the balance data includes standing balance, sitting balance, center of gravity swing, and postural stability.

[0021] It should be noted that the finger tapping data includes tapping frequency, rhythm, strength, and finger coordination.

[0022] A voice acquisition unit, which is used to monitor and collect the voice characteristics and voice content of the patient in real time based on the Internet of Things technology to obtain the patient's voice data; It should be noted that the voice characteristics include fundamental frequency, sound intensity, sound length, speech rate, pause, intonation, pronunciation clarity, and speech fluency.

[0023] It should be noted that the voice content includes vocabulary, grammar structure, semantic expression, and emotional expression; A writing acquisition unit, which is used to monitor and collect the writing dynamics data and writing graphic data of the patient in real time based on the Internet of Things technology to obtain the patient's writing data; It should be noted that the writing dynamics data includes writing speed, acceleration, and pressure, stroke length, width, and angle, writing pauses, and tremors.

[0024] It should be noted that the writing graphic data includes font size, shape, and spacing, writing neatness, and coherence; A facial acquisition unit, which is used to monitor and collect the facial movement data and facial expression data of the patient in real time based on the Internet of Things technology to obtain the patient's facial data; It should be noted that the facial movement data includes the movements of the eyebrows, eyes, nose, and mouth.

[0025] It should be noted that the facial expression data includes the frequency, amplitude, and duration of expression changes.

[0026] Among them, based on the patient's movement data, voice data, writing data, and facial data, real-time patient behavior data based on the Internet of Things is determined.

[0027] Specifically, by collecting the patient's movement data, voice data, writing data, and facial data, determining real-time patient behavior data based on the Internet of Things can provide data support for the quantitative early diagnosis of Parkinson's disease in patients.

[0028] Among them, the quantitative diagnosis platform is used to preprocess and analyze the real-time patient behavior data based on the Internet of Things, and perform quantitative early diagnosis of Parkinson's disease in patients. According to the patient's diagnosis results, intelligent management and treatment of the patient are carried out.

[0029] Specifically, the writing acquisition unit includes an intelligent writing pen and a writing photographing device. Among them, a wireless communication connection is established between the intelligent writing pen and the writing photographing device: The intelligent writing pen real-time collects the number of writing tremors and writing pressure of the patient within a unit time; The number of writing tremors and the writing pressure are respectively compared with a preset number-of-writing-tremors threshold and a writing-pressure threshold; When the number of writing tremors exceeds the preset number-of-writing-tremors threshold, but the writing pressure does not exceed the writing-pressure threshold, the intelligent writing pen triggers the writing photographing device to take a photo to obtain a first writing image; When the number of writing tremors does not exceed the preset number-of-writing-tremors threshold, but the writing pressure exceeds the writing-pressure threshold, a comprehensive judgment is made based on the corresponding values of the number of writing tremors and the writing pressure to determine whether to trigger the writing photographing device to take a photo; When both the number of writing tremors and the writing pressure do not exceed their corresponding number-of-writing-tremors threshold and writing-pressure threshold, the writing photographing device is controlled to take a photo according to a preset image acquisition frequency; a second writing image is obtained; The evaluation weight value of the first writing image is set to be higher than the evaluation weight value corresponding to the second writing image.

[0030] The technical effects of the above technical solution are as follows: Real-time collection of the number of writing tremors and writing pressure of the patient by the intelligent writing pen can effectively improve the collection rate of obtaining the writing state information of the patient and the timeliness of data acquisition. At the same time, judging whether to trigger the writing photographing device to take a photo according to the preset number-of-writing-tremors threshold and writing-pressure threshold can effectively improve the accuracy of judging abnormal situations during the patient's writing process and the necessity of triggering, and prevent the problem that the writing image under important and key writing features cannot be successfully captured when taking images at a fixed shooting frequency.

[0031] When either the number of writing tremors or the writing pressure exceeds a threshold, the system will adopt different photographing strategies. In particular, when the number of writing tremors exceeds the threshold but the writing pressure does not, the system will immediately trigger a photograph; while when neither exceeds the threshold, photographing will be performed according to the preset image acquisition frequency. This photographing trigger strategy can effectively improve the pertinence of photographing, thereby improving the accuracy of subsequent Parkinson's assessment. On the other hand, the system sets the evaluation weight value of the first writing image (i.e., the photograph triggered under specific conditions) to be higher than that of the second writing image (i.e., the image taken at the preset frequency). This weight assignment method can focus more on key image data during the analysis process, thus improving the accuracy and effectiveness of the assessment.

[0032] Specifically, by comprehensively judging the corresponding values of the number of writing tremors and the writing pressure to determine whether to trigger the writing photographing device to take a photograph, including: When the writing pressure exceeds the writing pressure threshold, extract the number of writing tremors corresponding to the unit time when the number of writing tremors does not exceed the preset number of writing tremors threshold as the target tremor number dataset; Generate a comprehensive writing state parameter using the target tremor number dataset and the writing pressure; wherein, the comprehensive writing state parameter is used to evaluate whether the writing state of the current patient requires image acquisition. Among them, the comprehensive writing state parameter is obtained through the following formula: Among them, S represents the comprehensive writing state parameter; P represents the writing pressure corresponding to exceeding the writing pressure threshold; P y represents the writing pressure threshold; n represents the number of data included in the target tremor number dataset, where the number of data is the same as the number of unit times when the number of writing tremors does not exceed the preset number of writing tremors threshold; C i represents the tremor number value corresponding to the i-th tremor data in the target tremor number dataset; N represents the number of unit times the patient has experienced writing; n0 represents the reference value of the preset number of unit times; C y represents the preset number of writing tremors threshold; Compare the comprehensive writing state parameter with the preset parameter threshold; When the comprehensive writing state parameter is lower than the preset parameter threshold, it is determined that there is no need to trigger the writing photographing device to take a photograph; When the comprehensive writing state parameter is not lower than the preset parameter threshold, it is determined that it is necessary to trigger the writing photographing device to take a photograph; Take the writing image obtained when the number of writing tremors does not exceed the preset number of writing tremors threshold, but the writing pressure exceeds the writing pressure threshold as the third writing image; Set the evaluation weight value of the third written image to be higher than that of the second written image but lower than that of the first written image.

[0033] The technical effects of the above technical solution are as follows: By introducing the comprehensive writing state parameter S, the above technical solution can more precisely evaluate the writing state of the patient. By combining the number of writing tremors and writing pressure for comprehensive evaluation, the accuracy of evaluating the patient's actual writing situation can be effectively improved. At the same time, in the above technical solution, by using the number of tremors and pressure changes within different unit times during the patient's writing process and obtaining the comprehensive writing state parameter by extracting the target tremor number dataset, the followability between the comprehensive writing state parameter and the subtle changes in the patient's writing state can be effectively improved, thereby effectively improving the capture sensitivity and accuracy of the subtle changes in the patient's writing state. At the same time, in the above technical solution, weight assignments are made to the written images obtained under different conditions. The evaluation weight value of the third written image is higher than that of the second written image but lower than that of the first written image. This weight assignment method not only considers the importance differences of the images under different conditions but also maintains the rationality and coherence of the weight assignment, can effectively improve the distinguishability of the importance of the first written image, the second written image, and the third written image, and further improve the accuracy of subsequent early Parkinson's judgment.

[0034] In this embodiment, the quantitative diagnosis platform includes: a behavior data preprocessing unit and an early diagnosis evaluation unit.

[0035] Among them, the behavior data preprocessing unit is used to clean, transform, integrate, and extract features from the real-time patient behavior data based on the Internet of Things to determine the real-time patient behavior feature data based on the Internet of Things. It should be noted that the behavior data preprocessing unit includes: A behavior data cleaning module, which is used to clean the real-time patient behavior data based on the Internet of Things; Among them, based on the Python library, check the real-time patient behavior data based on the Internet of Things, identify duplicate values, missing values, and outliers in the real-time patient behavior data based on the Internet of Things, and delete the duplicate values, missing values, and outliers in the real-time patient behavior data based on the Internet of Things; A behavior data transformation module, which is used to transform the real-time patient behavior data based on the Internet of Things; Among them, based on Z-Score standardization, normalize the real-time patient behavior data based on the Internet of Things, convert the real-time patient behavior data based on the Internet of Things into a standard normal distribution with a mean of 0 and a standard deviation of 1, and remove the dimensional differences between the real-time patient behavior data based on the Internet of Things; A behavior data integration module, which is used to integrate the real-time patient behavior data based on the Internet of Things; Among them, based on the Python library, the real-time data of patients' behaviors based on the Internet of Things is integrated, and the real-time data of patients' behaviors based on the Internet of Things is integrated into a unified data view, and the integrated real-time data of patients' behaviors based on the Internet of Things is verified to determine whether the integrated real-time data of patients' behaviors based on the Internet of Things is missing; The behavior data extraction module is used to extract the real-time data of patients' behaviors based on the Internet of Things; Among them, the features related to Parkinson's disease are extracted from the real-time data of patients' behaviors based on the Internet of Things, and the extracted feature vectors are weighted and fused to determine the patient behavior feature data based on the Internet of Things.

[0036] Specifically, by cleaning, transforming, integrating and feature extraction of the real-time data of patients' behaviors based on the Internet of Things, the patient behavior feature data based on the Internet of Things is determined, where the patient behavior feature data includes gait parameters, voice features, writing dynamics features, facial expression features, etc.

[0037] Among them, the early diagnosis and evaluation unit is used to analyze the patient behavior feature data based on the Internet of Things, and perform quantitative early diagnosis of Parkinson's disease on the patient, and perform intelligent management and treatment on the patient according to the patient's diagnosis result.

[0038] It should be noted that the early diagnosis and evaluation unit includes: The diagnostic model training module is used to train the quantitative early diagnosis model of Parkinson's disease; According to the quantitative early diagnosis requirements of Parkinson's disease based on behavior feature recognition, the historical data of patients' behaviors based on the Internet of Things is collected, and the collected historical data of patients' behaviors based on the Internet of Things is divided, and the historical data of patients' behaviors based on the Internet of Things is divided into a training set and a test set; Based on deep learning technology, the training set is used to train the deep learning model, so that the deep learning model autonomously learns the quantitative early diagnosis process of Parkinson's disease based on behavior feature recognition, and determines the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition; Based on the test set, the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition is tested, and the performance of the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition is evaluated based on the accuracy rate, recall rate and F1 value indicators, and it is judged whether the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition can achieve the expected effect; According to the test evaluation results, the parameters and structure of the quantitative early diagnosis model of Parkinson's disease based on behavior feature recognition are adjusted, and through continuous iterative optimization, the optimal quantitative early diagnosis model of Parkinson's disease is determined; The diagnostic evaluation and management module is used to perform quantitative early diagnosis and evaluation and management of Parkinson's disease on the patient; Obtain the optimal early quantitative diagnosis model for Parkinson's disease, and deploy the optimal early quantitative diagnosis model for Parkinson's disease in an actual early quantitative diagnosis environment for Parkinson's disease based on behavioral feature recognition; Input the patient's behavioral feature data based on the Internet of Things into the optimal early quantitative diagnosis model for Parkinson's disease, analyze the patient's behavioral feature data based on the Internet of Things using the optimal early quantitative diagnosis model for Parkinson's disease, and conduct an early quantitative diagnosis of Parkinson's disease for the patient to determine the patient's diagnosis result; Among them, based on the patient's diagnosis result, analyze the patient's behavioral feature data to determine the patient's condition assessment report, conduct intelligent management and treatment for the patient based on the patient's condition assessment report, and monitor the treatment effect in real time, and dynamically adjust the patient's management and treatment plan according to the monitoring situation.

[0039] Therefore, analyze the patient's behavioral feature data based on the Internet of Things using the early quantitative diagnosis model for Parkinson's disease, conduct an early quantitative diagnosis of Parkinson's disease for the patient to determine the patient's diagnosis result, based on the patient's diagnosis result, analyze the patient's behavioral feature data to determine the patient's condition assessment report, conduct intelligent management and treatment for the patient based on the patient's condition assessment report, and monitor the treatment effect in real time, and dynamically adjust the patient's management and treatment plan according to the monitoring situation. It is possible to conduct an early quantitative diagnosis of Parkinson's disease for the patient based on behavioral feature recognition, and improve the diagnosis effect of Parkinson's patients.

[0040] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0041] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A quantitative early diagnosis system for Parkinson's disease based on behavioral feature recognition, characterized by: include: An intelligent collection device and a quantitative diagnosis platform, wherein the intelligent collection device is configured to collect patient behavior characteristics during the patient's behavior activities, determine the patient's real-time behavior data, and transmit the collected patient behavior real-time data to a given quantitative diagnosis platform based on wireless transmission technology; Wherein, whether to trigger the writing camera to take a photo is comprehensively judged based on the corresponding values ​​of the number of writing tremors and the writing pressure; The quantitative diagnosis platform is configured to construct a model for quantitative early diagnosis of Parkinson's disease based on the patient's behavioral characteristics, and to analyze the patient's real-time behavioral data and perform quantitative early diagnosis of Parkinson's disease based on the constructed model.

2. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 1, characterized in that: The intelligent acquisition device includes: a motion acquisition unit, a voice acquisition unit, a writing acquisition unit and a face acquisition unit; The motion collection unit is configured to monitor and collect the patient's gait data, balance data and finger tapping data in real time based on the Internet of Things technology to obtain the patient's motion data; The voice collection unit is configured to monitor and collect the patient's voice characteristics and voice content in real time based on the Internet of Things technology to obtain the patient's voice data; The writing collection unit is configured to monitor and collect the patient's writing dynamics data and writing graphic data in real time based on the Internet of Things technology to obtain the patient's writing data; The facial acquisition unit is configured to monitor and acquire the patient's facial action data and facial expression data in real time based on the Internet of Things technology to obtain the patient's facial data; Among them, the real-time data of patient behavior based on the Internet of Things is determined according to the collected patient movement data, patient voice data, patient writing data and patient facial data.

3. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 2, characterized in that: The writing collection unit includes an intelligent writing pen and a writing and photographing device, wherein a wireless communication connection is established between the intelligent writing pen and the writing and photographing device: The smart writing pen collects the number of writing tremors and writing pressure of the patient per unit time in real time; Comparing the writing tremor number and writing pressure with a preset writing tremor number threshold and writing pressure threshold respectively; When the number of writing tremors exceeds a preset writing tremor number threshold, but the writing pressure does not exceed the writing pressure threshold, the writing photographing device is triggered by the smart writing pen to take a photograph to obtain a first writing image; When the number of writing tremors does not exceed the preset writing tremor number threshold, but the writing pressure exceeds the writing pressure threshold, a comprehensive judgment is made based on the corresponding values ​​of the writing tremor number and the writing pressure to determine whether to trigger the writing camera to take a photo; When the writing tremor number and the writing pressure do not exceed their corresponding writing tremor number threshold and writing pressure threshold, the writing photographing device is controlled to take pictures according to a preset image acquisition frequency; and a second writing image is acquired; The evaluation weight value of the first writing image is set to be higher than the evaluation weight value corresponding to the second writing image.

4. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 3, characterized in that: Comprehensively judging whether to trigger the writing camera to take photos based on the number of writing tremors and the corresponding values ​​of writing pressure, including: When the writing pressure exceeds the writing pressure threshold, extracting the number of writing tremors corresponding to the unit time in which the number of writing tremors does not exceed the preset writing tremor number threshold as the target tremor number data set; Generate a comprehensive writing state parameter using the target tremor frequency data set and writing pressure; wherein the comprehensive writing state parameter is used to evaluate whether the current writing state of the patient meets the necessity of image acquisition; comparing the comprehensive writing state parameter with a preset parameter threshold; When the comprehensive writing state parameter is lower than a preset parameter threshold, it is determined that there is no need to trigger the writing photographing device to take a photograph; When the comprehensive writing state parameter is not lower than a preset parameter threshold, it is determined that the writing photographing device needs to be triggered to take a photograph; taking a writing image acquired when the number of writing tremors does not exceed a preset writing tremor number threshold, but the writing pressure exceeds a writing pressure threshold as a third writing image; The evaluation weight value of the third writing image is set to be higher than the evaluation weight value corresponding to the second writing image, but lower than the evaluation weight value corresponding to the first writing image.

5. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 2, characterized in that: The gait data includes step length, step width, step frequency, step speed, swing phase, stance phase, gait symmetry, gait variability, trunk swing, arm swing and turning gait; Balance data include standing balance, sitting balance, center of gravity swing and postural stability; Finger tapping data include tapping frequency, rhythm, strength, and finger coordination; Speech characteristics include fundamental frequency, intensity, duration, speech rate, pauses, intonation, articulation clarity, and speech fluency; The speech content includes vocabulary, grammatical structure, semantic expression and emotional expression; Writing dynamics data include writing speed, acceleration and pressure, stroke length, width and angle, writing pauses and tremors; Writing graphic data include font size, shape and spacing, handwriting neatness and coherence; Facial motion data includes the movements of eyebrows, eyes, nose and mouth; Facial expression data include the frequency, amplitude and duration of expression changes.

6. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 5, characterized in that: The quantitative diagnosis platform includes: a behavior data preprocessing unit and an early diagnosis evaluation unit; The behavior data preprocessing unit is configured to clean, transform, integrate and extract features of the real-time patient behavior data based on the Internet of Things, and determine the patient behavior feature data based on the Internet of Things; The early diagnosis and evaluation unit is configured to analyze the patient behavior characteristic data based on the Internet of Things, and perform quantitative early diagnosis of Parkinson's disease on the patient, and carry out intelligent management and treatment of the patient according to the diagnosis result of the patient.

7. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 6, characterized in that: The behavior data preprocessing unit includes: a behavior data cleaning module and a behavior data conversion module; The behavior data cleaning module is configured to clean the real-time patient behavior data based on the Internet of Things; Among them, the real-time data of patient behavior based on the Internet of Things is checked based on the Python library, duplicate values, missing values ​​and abnormal values ​​in the real-time data of patient behavior based on the Internet of Things are identified, and the duplicate values, missing values ​​and abnormal values ​​in the real-time data of patient behavior based on the Internet of Things are deleted; The behavior data transformation module is configured to transform the real-time patient behavior data based on the Internet of Things; Among them, the real-time patient behavior data based on the Internet of Things is normalized based on Z-Score standardization, and the real-time patient behavior data based on the Internet of Things is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1, thereby removing the dimensional differences between the real-time patient behavior data based on the Internet of Things.

8. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 7, characterized in that: The behavior data preprocessing unit further includes: a behavior data integration module and a behavior data extraction module; The behavior data integration module is configured to integrate the real-time patient behavior data based on the Internet of Things; Among them, the real-time patient behavior data based on the Internet of Things is integrated based on the Python library, the real-time patient behavior data based on the Internet of Things is integrated into a unified data view, and the integrated real-time patient behavior data based on the Internet of Things is verified to determine whether the integrated real-time patient behavior data based on the Internet of Things is missing; The behavior data extraction module is configured to extract real-time patient behavior data based on the Internet of Things; Among them, features related to Parkinson's disease are extracted from the real-time patient behavior data based on the Internet of Things, and the extracted feature vectors are weighted fused to determine the patient behavior feature data based on the Internet of Things.

9. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition as claimed in claim 8, characterized in that: The early diagnosis and evaluation unit includes: a diagnosis model training module, wherein the diagnosis model training module is configured to train a quantitative early diagnosis model for Parkinson's disease; According to the demand for quantitative early diagnosis of Parkinson's disease based on behavioral feature recognition, the patient behavior history data based on the Internet of Things is collected and divided into a training set and a test set. Based on deep learning technology, the deep learning model is trained with a training set, so that the deep learning model can autonomously learn the quantitative early diagnosis process of Parkinson's disease based on behavioral feature recognition, and determine the quantitative early diagnosis model of Parkinson's disease based on behavioral feature recognition; Based on the test set, the Parkinson's disease quantitative early diagnosis model based on behavioral feature recognition is tested, and the performance of the Parkinson's disease quantitative early diagnosis model based on behavioral feature recognition is evaluated based on the accuracy, recall rate and F1 value indicators to determine whether the Parkinson's disease quantitative early diagnosis model based on behavioral feature recognition can achieve the expected effect; According to the test evaluation results, the parameters and structure of the Parkinson's disease quantitative early diagnosis model based on behavioral feature recognition are adjusted. After continuous iterative optimization, the optimal Parkinson's disease quantitative early diagnosis model is determined.

10. The Parkinson's disease quantitative early diagnosis system based on behavioral feature recognition according to claim 9, characterized in that: The early diagnosis and evaluation unit further includes: a diagnosis and evaluation control module; the diagnosis and evaluation control module is configured to perform quantitative early diagnosis and evaluation control of Parkinson's disease on the patient; Obtain the optimal Parkinson's disease quantitative early diagnosis model, and deploy it in the actual Parkinson's disease quantitative early diagnosis environment based on behavioral feature recognition; Inputting the patient behavior characteristic data based on the Internet of Things into the optimal Parkinson's quantitative early diagnosis model, analyzing the patient behavior characteristic data based on the Internet of Things based on the optimal Parkinson's quantitative early diagnosis model, and performing Parkinson's quantitative early diagnosis on the patient to determine the patient's diagnosis result; Among them, based on the patient's diagnosis results, the patient's behavioral characteristic data is analyzed to determine the patient's condition assessment report, and the patient is given intelligent management and treatment based on the patient's condition assessment report. The treatment effect is monitored in real time, and the patient's management and treatment plan is dynamically adjusted according to the monitoring situation.

Images