Parkinson data analysis pen based on 6-axis IMU (Inertial Measurement Unit) and neural network
The Parkinson's disease data analysis pen based on a 6-axis IMU and a one-dimensional convolutional neural network solves the problems of complex actions and high costs in existing technologies, realizing low-cost and easy-to-promote Parkinson's disease data collection and analysis, and improving patient compliance and analysis accuracy.
Patent Information
- Application Number
- CN202511351047.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-22
- Publication Date
- 2026-02-17
AI Technical Summary
In existing technologies, the diagnosis and monitoring of Parkinson's disease rely on costly, complex standardized actions and sophisticated instruments, which are difficult to maintain consistency in different scenarios, have poor compliance with patients with cognitive impairment or the elderly, and suffer from artifacts and high power consumption.
A Parkinson's disease data analysis pen based on a 6-axis IMU and a one-dimensional convolutional neural network is used to collect data from patients' natural writing movements and perform real-time analysis using a pre-trained neural network, thereby reducing the barrier to entry and improving accuracy.
It enables low-cost, large-scale data collection and analysis of Parkinson's disease, improving patient compliance and the accuracy and real-time nature of analysis results.
Smart Images
Figure CN121533685A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of medical device technology, and in particular to a Parkinson's disease data analysis pen based on a 6-axis IMU and neural network. Background Technology
[0002] Parkinson's disease has become one of the fastest-growing neurodegenerative diseases worldwide, affecting over ten million people. The typical motor symptoms caused by this disease, such as tremor, rigidity, and bradykinesia, significantly limit patients' mobility, leading to a sharp decline in quality of life. Currently, clinical diagnosis and disease monitoring mainly rely on doctors' subjective observation of motor signs. Evaluation results are easily influenced by experience differences, lacking objective and quantifiable diagnostic criteria. Related tests require sophisticated instruments and specialized operation, and can only be performed in large medical centers.
[0003] Existing technologies integrate inertial measurement units (IMUs) into gloves or watches to collect data from patients performing actions such as arm extension or resting movements, and finger-to-nose tests. Algorithms then calculate severity scores on grading scales to assess symptoms like Parkinson's tremor and bradykinesia. However, this approach's high reliance on standardized movement paradigms leads to poor adherence by patients with cognitive impairment or the elderly, and the process is difficult to maintain consistently across different scenarios. Furthermore, fabric slippage, differences in wearing position, and asymmetry of symptoms between the left and right hands can introduce additional artifacts. High power consumption, frequent charging, and the cost and maintenance issues associated with customization also hinder the large-scale adoption of this method in clinical and home settings.
[0004] Therefore, there is an urgent need for a product that is highly accurate, low-cost, and easy to scale up, with the function of collecting and analyzing Parkinson's disease data. Summary of the Invention
[0005] This application provides a Parkinson's disease data analysis pen based on a 6-axis IMU and neural network, which can collect motion data of patients during natural writing actions and perform real-time data analysis at the pen tip through a one-dimensional convolutional neural network model.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a Parkinson's disease data analysis pen based on a 6-axis IMU and a neural network, comprising: The data acquisition module is used to acquire the patient's writing data through a 6-axis IMU built into the pen barrel, and send the writing data to the data analysis module. The writing data includes the tremor direction, tremor frequency and tremor amplitude data of the hand. The data analysis module includes an MCU built into the pen barrel, used to receive the writing data, calibrate the zero-point drift of the writing data, and input the calibrated data into a pre-trained one-dimensional convolutional neural network in the MCU to obtain classification data. The classification data is analyzed based on a preset threshold to obtain the patient's health status. The communication module is used to receive the classified data and send it to the terminal device or the cloud.
[0007] In a preferred example of this application, it can be further configured to include a model pre-training module for: The patient's historical writing data is acquired through the 6-axis IMU built into the pen barrel; The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score normalization; The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; The preprocessed data is used as the training set and input into a one-dimensional convolutional neural network. The Adam optimizer is used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network. The pre-trained one-dimensional convolutional neural network is integrated into the data analysis module.
[0008] In a preferred example of this application, it may further be configured to include: An AC charging module includes a Type-C charging slot disposed on the side wall of the pen barrel and a dustproof silicone plug disposed in the Type-C charging slot, for supplying power to the data acquisition module, the data analysis module and the communication module; A battery charging module, located at the end of the pen, includes a swivelable conical battery compartment cover and a built-in battery, used to power the data acquisition module, the data analysis module, and the communication module.
[0009] Secondly, this application provides a Parkinson's disease data analysis method based on a 6-axis IMU and neural network, which collects the patient's historical writing data and real-time writing data using any of the Parkinson's disease data analysis pens based on a 6-axis IMU and neural network as described in any one of the claims, and trains a one-dimensional convolutional neural network using the historical writing data, including: The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score normalization; The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; The preprocessed data is used as the training set and input into a one-dimensional convolutional neural network. The Adam optimizer is then used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
[0010] In a preferred example of this application, it may further be configured to include: Real-time data collection and writing; The real-time writing data is input into the pre-trained one-dimensional convolutional neural network to obtain the PD probability; The patient's health status is obtained by analyzing the PD probability based on a preset threshold.
[0011] In a preferred example of this application, it may further be configured to include: The real-time writing data is sent to the cloud or terminal device via wireless communication; The health status is displayed in real time using the user interface of the terminal device.
[0012] Thirdly, this application provides a Parkinson's disease data analysis device based on a 6-axis IMU and a neural network, the device comprising: The data acquisition module is used to collect the patient's historical writing data; The data preprocessing module is used to slice and randomly sort the first data using a sliding window to obtain preprocessed data; and to eliminate individual baseline differences in the historical written data through z-score standardization to obtain the first data. The model training module is used to input the preprocessed data as a training set into a one-dimensional convolutional neural network, and to use the Adam optimizer to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
[0013] Fourthly, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any of the preceding claims.
[0014] Fifthly, this application provides a computer-readable storage medium storing a program, wherein when the program is executed by a processor, it implements the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any of the preceding claims.
[0015] Sixthly, this application provides a computer program product including computer instructions that, when executed by a processor, implement the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any of the preceding claims.
[0016] In summary, compared with the prior art, the beneficial effects of the technical solution provided in this application include at least the following: By acquiring data on the direction, frequency, and amplitude of hand tremors during the writing process using a built-in 6-axis IMU (Inertial Measurement Unit), and using a built-in pre-trained one-dimensional convolutional neural network model for real-time classification, high-accuracy real-time data analysis of Parkinson's disease was achieved.
[0017] The data analysis pen is designed with a common pen-shaped structure, which is not only simple to manufacture and inexpensive, but also allows users to write naturally as if using a regular pen, without having to perform complex standardized actions. This greatly reduces the barrier to use and improves patient compliance, especially for patients with cognitive impairment or the elderly.
[0018] By using a communication module built into the pen, the analyzed data is uploaded to mobile apps and other terminal devices via wireless communication methods such as Bluetooth. The analysis results are displayed instantly through the application interface, improving the intuitiveness and real-time nature of the analysis results. Attached Figure Description
[0019] Figure 1 A block diagram of a Parkinson's data analysis pen based on a 6-axis IMU and neural network is provided for one embodiment of this application.
[0020] Figure 2 This is a pen body structure diagram of a Parkinson's data analysis pen based on a 6-axis IMU and neural network, provided as an embodiment of this application.
[0021] Figure 3 A cross-sectional view of the pen body of a Parkinson's data analysis pen based on a 6-axis IMU and neural network, provided as an embodiment of this application.
[0022] Figure 4 A flowchart illustrating a Parkinson's disease data analysis method based on a 6-axis IMU and neural network, provided as an embodiment of this application; Figure 5 A block diagram of a Parkinson's disease data analysis device based on a 6-axis IMU and neural network, provided as an embodiment of this application. Detailed Implementation
[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0024] In one embodiment of this application, a Parkinson's disease data analysis pen based on a 6-axis IMU and neural network is provided. Please refer to [link to relevant documentation]. Figure 1 As shown, the pen includes: The data acquisition module Q100 is used to acquire the patient’s writing data through a 6-axis IMU built into the pen barrel and send the writing data to the data analysis module. The writing data includes the tremor direction, tremor frequency and tremor amplitude data of the hand. The data analysis module Q200 includes an MCU built into the pen barrel, which receives the writing data, calibrates the zero-point drift of the writing data, and inputs the calibrated data into a pre-trained one-dimensional convolutional neural network in the MCU to obtain classification data. The classification data is then analyzed based on a preset threshold to obtain the patient's health status. The communication module Q300 is used to receive the classified data and send it to the terminal device or the cloud.
[0025] In specific implementation, such as Figure 2 As shown, the physical structure of the Parkinson's disease data analysis pen includes a pen barrel 1, a pen tip 2, a pen refill 3, and a flexible circuit board 4. Specifically, the pen's internal structure is a three-section modular structure: the front end is a replaceable conventional pen refill and a metal pen tip; the middle section is a hollow pen barrel made of thin-walled aluminum alloy, with an embedded flexible circuit integrating an IMU and an MCU; the pen's outer diameter can be 11 mm, its weight is 15g, and its center of gravity is located 15 mm behind the grip, providing a writing feel no different from a traditional pen.
[0026] More specifically, a cross-sectional view of the pen is shown below. Figure 3 As shown, there is a flexible circuit board 4 containing an MCU and a 6-axis IMU, and a pen refill 3.
[0027] The data acquisition module collects writing data from the patient during the writing process using a 6-axis IMU built into the pen. The handwriting can be "SYSU". The 6-axis IMU collects three-dimensional acceleration and angular velocity data during writing, i.e., the direction, frequency, and amplitude of hand tremors. Then, the writing data is sent to the data analysis module, i.e., the MCU (microcontroller unit) in the flexible circuit, for data analysis. The MCU receives the writing data, calibrates the zero-point drift of the writing data, and inputs the calibrated data into a pre-trained one-dimensional convolutional neural network within the MCU to obtain classification data. Based on a preset threshold, the classification data is analyzed to determine the patient's health status.
[0028] The communication module uploads the patient's health status to a mobile application (APP) via Bluetooth, and displays the data analysis results through the application interface.
[0029] Furthermore, the pre-training steps of the one-dimensional convolutional neural network include: The patient's historical writing data is acquired through the 6-axis IMU built into the pen barrel; The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score normalization; The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; The preprocessed data is used as the training set and input into a one-dimensional convolutional neural network. The Adam optimizer is then used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
[0030] After training is completed, the pre-trained one-dimensional convolutional neural network is integrated into the data analysis module.
[0031] In this embodiment, the direction, frequency, and amplitude data of hand tremors during writing are collected by a built-in 6-axis IMU (Inertial Measurement Unit), and real-time classification is performed using a built-in pre-trained one-dimensional convolutional neural network model, achieving high-accuracy real-time data analysis for Parkinson's disease. The data analysis pen is designed with a standard pen-like structure, which is not only simple to manufacture and inexpensive, but also allows users to write naturally like with a regular pen, without needing to perform complex standardized action patterns, thus greatly reducing the barrier to entry and improving patient compliance, especially for patients with cognitive impairment or the elderly. Through a communication module built into the pen, the analyzed data is uploaded to mobile applications and other terminal devices via wireless communication methods such as Bluetooth. The analysis results are displayed instantly through the application interface, improving the intuitiveness and real-time nature of the analysis results.
[0032] In some embodiments, it also includes: An AC charging module includes a Type-C charging slot disposed on the side wall of the pen barrel and a dustproof silicone plug disposed in the Type-C charging slot, for supplying power to the data acquisition module, the data analysis module and the communication module; A battery charging module, located at the end of the pen, includes a swivelable conical battery compartment cover and a built-in battery, used to power the data acquisition module, the data analysis module, and the communication module.
[0033] In practice, a Type-C charging slot is milled into the side wall of the pen barrel, and it is equipped with a dustproof silicone plug. The tail end has a screw-on cone-shaped battery compartment cover, with a built-in battery powering the pen.
[0034] In this embodiment, the pen is self-powered and offers multiple optional charging methods, improving ease of use.
[0035] This application also provides a Parkinson's disease data analysis method based on a 6-axis IMU and neural network. Using the aforementioned Parkinson's disease data analysis pen based on a 6-axis IMU and neural network, historical and real-time writing data of the patient are collected, and a one-dimensional convolutional neural network is trained using the historical writing data. (See also...) Figure 4 As shown, it includes: S100: Eliminate individual baseline differences in the historical writing data through z-score standardization to obtain the first data; S200: The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; S300: The preprocessed data is used as the training set to input into a one-dimensional convolutional neural network, and the Adam optimizer is used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
[0036] In practice, an end-to-end one-dimensional convolutional neural network training paradigm is used to train the one-dimensional convolutional neural network. The Parkinson's data analysis pen is used to collect historical training data, which is then preprocessed and used as the training set.
[0037] During data collection, independent samples were first divided by "person," including 3 clinically diagnosed PD subjects and 13 healthy subjects. All subjects used the same diagnostic pen in the same environment to complete the writing task of the designated trajectory "SYSU" 30 times each. Subsequently, data preprocessing was performed. Z-score standardization was used to eliminate individual baseline differences in the historical writing data to obtain the first data. The first data was then sliced and randomly shuffled using a sliding window (10 s window length, 1 s step size) to obtain preprocessed data. The preprocessed data was input into a one-dimensional convolutional neural network, and the Adam optimizer was used for iterative convergence on the training set. After training, independent validation was performed on 1 PD subject and 3 healthy subjects who did not participate in the training. After verifying the accuracy of the one-dimensional convolutional neural network, it was integrated into the MCU of the pen to obtain the pre-trained one-dimensional convolutional neural network.
[0038] In this embodiment, the accuracy of patient handwriting data classification is improved by pre-training a one-dimensional convolutional neural network.
[0039] In some embodiments, it also includes: Real-time data collection and writing; The real-time writing data is input into the pre-trained one-dimensional convolutional neural network to obtain the PD probability; The patient's health status is obtained by analyzing the PD probability based on a preset threshold.
[0040] In practice, the output of the one-dimensional convolutional neural network is the PD probability. Based on the PD probability and a preset threshold, the patient's health status is determined. The preset threshold is 0.5; a value greater than 0.5 indicates PD, otherwise the patient is considered healthy.
[0041] In this embodiment, the patient's health status is judged by a preset threshold, resulting in a more intuitive and accurate analysis.
[0042] In some embodiments, it also includes: The real-time writing data is sent to the cloud or terminal device via wireless communication; The health status is displayed in real time using the user interface of the terminal device.
[0043] In practice, after collecting the real-time writing data, it also supports uploading the data to the cloud for analysis, receiving the analysis results from the cloud, and sending them to the user interface of the terminal device to display the health status in real time.
[0044] In this embodiment, the efficiency of writing data analysis can be improved when local computing resources are insufficient.
[0045] This application also provides a Parkinson's disease data analysis device based on a 6-axis IMU and a neural network. Please refer to [link to relevant documentation]. Figure 5 As shown, the device includes: The data acquisition module Z100 is used to collect the patient's historical writing data; The data preprocessing module Z200 is used to slice and randomly sort the first data using a sliding window to obtain preprocessed data; and to eliminate individual baseline differences in the historical written data through z-score standardization to obtain the first data. The model training module Z300 is used to input the preprocessed data as a training set into a one-dimensional convolutional neural network, and to use the Adam optimizer to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
[0046] The functional implementation of each module in the above-mentioned Parkinson's data analysis device based on a 6-axis IMU and a neural network corresponds to the steps in the above-mentioned Parkinson's data analysis method embodiment based on a 6-axis IMU and a neural network. Their functions and implementation processes will not be described in detail here.
[0047] This application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any of the above embodiments.
[0048] This application also provides a computer-readable storage medium storing a program. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The working process, details, and technical effects of the computer-readable storage medium provided in this embodiment can be found in the above embodiment regarding a Parkinson's disease data analysis method based on a 6-axis IMU and neural network, and will not be repeated here.
[0049] The application also provides a computer program product, including computer instructions that, when executed by a processor, implement the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any of the above embodiments.
[0050] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0051] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A Parkinson's disease data analysis pen based on a 6-axis IMU and neural network, characterized in that, include: The data acquisition module is used to acquire the patient's writing data through a 6-axis IMU built into the pen barrel, and send the writing data to the data analysis module. The writing data includes the tremor direction, tremor frequency and tremor amplitude data of the hand. The data analysis module includes an MCU built into the pen barrel, used to receive the writing data, calibrate the zero-point drift of the writing data, and input the calibrated data into a pre-trained one-dimensional convolutional neural network in the MCU to obtain classification data. The classification data is analyzed based on a preset threshold to obtain the patient's health status. The communication module is used to receive the classified data and send it to the terminal device or the cloud.
2. The Parkinson's disease data analysis pen with a 6-axis IMU and neural network according to claim 1, characterized in that, It also includes a model pre-training module for: The patient's historical writing data is acquired through the 6-axis IMU built into the pen barrel; The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score standardization; The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; The preprocessed data is used as the training set and input into a one-dimensional convolutional neural network. The Adam optimizer is used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network. The pre-trained one-dimensional convolutional neural network is integrated into the data analysis module.
3. The Parkinson's disease data analysis pen with a 6-axis IMU and neural network according to claim 2, characterized in that, Also includes: An AC charging module includes a Type-C charging slot disposed on the side wall of the pen barrel and a dustproof silicone plug disposed in the Type-C charging slot, for supplying power to the data acquisition module, the data analysis module and the communication module; A battery charging module, located at the end of the pen, includes a swivelable conical battery compartment cover and a built-in battery, used to power the data acquisition module, the data analysis module, and the communication module.
4. A Parkinson's disease data analysis method based on a 6-axis IMU and a neural network, characterized in that, Using the Parkinson's data analysis pen based on a 6-axis IMU and neural network as described in any one of claims 1-3, historical writing data and real-time writing data of the patient are collected, and a one-dimensional convolutional neural network is trained using the historical writing data, including: The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score normalization; The first data is sliced and randomly sorted using a sliding window to obtain preprocessed data; The preprocessed data is used as the training set and input into a one-dimensional convolutional neural network. The Adam optimizer is then used to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
5. The Parkinson's disease data analysis method based on a 6-axis IMU and neural network according to claim 4, characterized in that, Also includes: Real-time data collection and writing; The real-time writing data is input into the pre-trained one-dimensional convolutional neural network to obtain the PD probability; The patient's health status is obtained by analyzing the PD probability based on a preset threshold.
6. The Parkinson's disease data analysis method based on a 6-axis IMU and neural network according to claim 5, characterized in that, Also includes: The real-time writing data is sent to the cloud or terminal device via wireless communication; The health status is displayed in real time using the user interface of the terminal device.
7. A Parkinson's disease data analysis device based on a 6-axis IMU and a neural network, characterized in that, include: The data acquisition module is used to collect the patient's historical writing data; The data preprocessing module is used to slice and randomly sort the first data using a sliding window to obtain preprocessed data; The first data is obtained by eliminating individual baseline differences in the historical writing data through z-score normalization; The model training module is used to input the preprocessed data as a training set into a one-dimensional convolutional neural network, and to use the Adam optimizer to iteratively converge on the training set to obtain a pre-trained one-dimensional convolutional neural network.
8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any one of claims 4 to 6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program, wherein when the program is executed by a processor, it implements the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any one of claims 4 to 6.
10. A computer program product comprising computer instructions, characterized in that, When executed by a processor, the computer instructions implement the steps of the Parkinson's data analysis method based on a 6-axis IMU and neural network as described in any one of claims 4 to 6.