Parkinson's disease emotion accompanying robot interaction method and system

Abstract

The invention relates to the technical field of accompanying robots, and discloses a Parkinson's disease emotion accompanying robot interaction method and system, and the method comprises the following steps: S1, carrying out the interaction initialization and scene adaptation, completing the hardware self-inspection through a multi-mode precise collection module after an emotion accompanying robot is started, and carrying out the interaction initialization and scene adaptation; the method comprises the following steps: acquiring basic physiological signals, initial behavior actions and real-time environment information of a Parkinson's patient, and establishing an interactive initial data set; and S2, multi-modal information is accurately acquired, a targeted acquisition strategy is triggered based on the initial data set, and dynamic physiological data, real-time limb movement amplitude, gait characteristics and voice emotion clues of the patient are continuously supplemented. According to the invention, through multi-mode accurate acquisition and disease condition and emotion joint identification, core demands such as synchronous and accurate capture of disease condition states and emotion states of patients, targeted matching of interaction strategies and rehabilitation assistance, emotion comfort and the like are realized, and the problem that'indiscriminate interaction 'of a traditional accompanying robot is not matched with individual demands of the patients is solved.

Description

Technical Field

[0001] This invention relates to the field of companion robot technology, specifically to an interactive method and system for an emotional companion robot for Parkinson's disease. Background Technology

[0002] The Parkinson's Disease Emotional Companion Robot is an intelligent service robot developed to meet the physiological and psychological needs of Parkinson's patients. It integrates artificial intelligence, sensing technology, rehabilitation assistance, and emotional interaction functions. It is a subcategory of medical and health care robots, designed to provide Parkinson's patients with both physical and mental companionship and assistance services.

[0003] Currently, most traditional emotional companion robots are general-purpose designs, lacking specific adaptation to the physiological characteristics of Parkinson's patients such as tremor, rigidity, and bradykinesia. During interaction, excessive physical movements or fast-paced speech can easily cause discomfort to patients. Some rehabilitation assistive devices focus only on a single function and cannot simultaneously capture negative emotional states such as anxiety and loneliness in patients, resulting in a lack of emotional warmth in the interaction and low patient cooperation. To address this, we propose an interactive method and system for an emotional companion robot for Parkinson's disease. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an interactive method and system for an emotional companion robot for Parkinson's disease. It solves the problems that most existing emotional companion robots are general-purpose designs, lacking specific adaptation to the physiological characteristics of Parkinson's patients such as tremor, rigidity, and bradykinesia. During interaction, excessive limb movements or fast-paced speech can easily cause discomfort to patients. Some rehabilitation assistive devices focus only on a single function and cannot simultaneously capture negative emotional states such as anxiety and loneliness in patients, resulting in a lack of emotional warmth in the interaction and low patient cooperation.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for interactive use of an emotional companion robot for Parkinson's disease, comprising the following steps:

[0006] S1. Interaction initialization and scene adaptation: After the emotional companion robot is started, it completes hardware self-test through the multimodal precision acquisition module, collects the basic physiological signals, initial behavioral actions and real-time environmental information of Parkinson's patients, and establishes the interaction initial dataset.

[0007] S2. Accurate collection of multimodal information: Based on the initial dataset, a targeted collection strategy is triggered to continuously supplement the patient's dynamic physiological data, real-time limb movement range, gait characteristics and voice emotion cues.

[0008] The formula for dynamically adjusting the sampling frequency is:

[0009]

[0010] in, Based on the sampling frequency, To adjust the coefficient, The instability coefficient of the condition;

[0011] S3. Joint recognition of disease condition and emotion: It calls up the preset Parkinson's disease patient state feature database and uses a weighted fusion algorithm to cross-validate the collected data to accurately identify the patient's disease condition and emotional state.

[0012] The feature fusion formula is:

[0013]

[0014] in, For the first Weights of class data For the first Normalized eigenvalues ​​of class data;

[0015] S4. Personalized interaction needs analysis: Combining the patient's historical interaction data, disease progression records and current recognition results, analyze the patient's core needs and clarify the priority of those needs;

[0016] The formula for calculating demand priority is as follows:

[0017]

[0018] in, , These are the weighting coefficients. Assess the urgency of the illness. Rate the intensity of emotional needs;

[0019] S5. Dynamic generation of interaction strategies: Based on the priority of needs, the intelligent interaction strategy library is called, and the details of the plan are optimized by combining the patient's personalized tags to generate exclusive interaction scripts.

[0020] The formula for calculating the suitability of the solution is:

[0021]

[0022] in, These are standard parameters for the strategy library. Personalized parameters for patients, The number of parameter dimensions;

[0023] S6. Interactive Execution and Real-time Feedback Acquisition: The robot executes operations according to a dedicated interactive script, and simultaneously collects patient feedback information in real time through a multimodal precision acquisition module to form an interactive feedback dataset.

[0024] S7. Interaction effect evaluation and strategy iteration: Based on the feedback dataset, an effect evaluation model is built to evaluate the interaction quality from multiple dimensions. When the preset threshold is not reached, the interaction strategy parameters are adjusted and the relevant data is updated to form a closed-loop optimization mechanism.

[0025] Preferably, the basic physiological signals in S1 include heart rate, muscle rigidity, and tremor frequency; the initial behavioral actions include static limb posture and facial expression; and the real-time environmental information includes the scene type, ambient noise level, and light intensity parameters.

[0026] Preferably, the targeted acquisition strategy described in S2 specifically includes: increasing the frequency of physiological signal acquisition for patients with unstable conditions; enhancing environmental noise filtering and behavioral recognition accuracy in noisy environments; achieving synchronous and accurate capture of the patient's condition and emotional state through multimodal precise acquisition and joint recognition of condition and emotion; and targeted matching of core needs such as rehabilitation assistance and emotional comfort with the interaction strategy, thus solving the problem of mismatch between the "indiscriminate interaction" of traditional companion robots and the personalized needs of patients.

[0027] Preferably, the Parkinson's disease patient state feature database in S3 includes feature parameters of different disease stages and emotional types. The physiological signal weights of the weighted fusion algorithm are dynamically allocated based on the severity of the patient's condition. The disease states include stable state, tremor attack state, myotonic discomfort state, and bradykinesia state. The emotional states include calm, anxiety, irritability, loneliness, and depression. The interaction scheme is optimized based on the patient's personalized tags. Gentle limb movements are adapted to the patient's range of motion. The voice content and rehabilitation guidance are tailored to the patient's acceptance ability. At the same time, the data accuracy in complex scenarios is ensured through targeted collection strategies, which greatly improves the patient's interaction cooperation and experience.

[0028] Preferably, the core needs mentioned in S4 include emotional comfort needs, rehabilitation assistance needs, environmental adjustment needs, or social companionship needs. The priority of the needs is determined based on the urgency of the patient's condition and the intensity of the emotional needs. The system automatically completes patient status recognition, need parsing, and interactive execution, reducing reliance on family members or caregivers. It is especially suitable for daily companionship and rehabilitation assistance scenarios during the stable period of the condition, effectively alleviating the time and energy pressure of human companionship and improving the convenience of care for Parkinson's patients.

[0029] Preferably, the multi-dimensional companionship scheme of the intelligent interaction strategy library in S5 includes voice comfort content, gentle physical interaction actions, step-by-step rehabilitation guidance prompts, and environmental parameter adaptation and adjustment. The range and frequency of the gentle interaction actions are preset based on the limb activity range threshold of Parkinson's patients. The patient's personalized tags include age, personality preferences, and disease severity. Through real-time feedback collection and multi-dimensional effect evaluation, the interaction strategy parameters are dynamically adjusted and patient data is updated, forming a closed-loop mechanism of "collection-identification-interaction-feedback-optimization" to ensure that the interaction scheme continuously adapts to the patient's disease progression and emotional changes, maintaining a stable companionship and assistance effect in the long term.

[0030] Preferably, the multi-dimensional assessment in S7 includes three dimensions: degree of emotional improvement, effect of relieving discomfort, and degree of cooperation in interaction. The preset threshold is determined based on the patient's historical best interaction effect, medical rehabilitation standards, and emotional improvement benchmark value. The intelligent interaction strategy library integrates emotionally oriented content and tiered rehabilitation guidance, which can not only relieve the patient's negative emotions through voice soothing and gentle physical interaction, but also provide rehabilitation training support under the premise of adapting to the condition, so as to achieve an organic combination of emotional companionship and rehabilitation assistance and fully meet the needs of patients.

[0031] Preferably, an interactive system for emotional companionship robots for Parkinson's disease includes: a multimodal precision acquisition module, a state joint recognition module, a personalized needs analysis module, an intelligent interaction strategy library, an interaction execution control module, a real-time feedback monitoring module, a closed-loop optimization management module, and a central coordination control module.

[0032] The multimodal precision acquisition module is used to collect dynamic physiological signals, behavioral characteristics and environmental dynamic information of Parkinson's patients in layers, and to perform noise reduction and standardization processing.

[0033] The state joint recognition module incorporates a Parkinson's disease patient state feature library and a weighted fusion algorithm model to simultaneously identify the patient's disease state and emotional state.

[0034] The personalized needs analysis module is used to analyze the patient's core needs and needs priority based on the identification results.

[0035] The intelligent interaction strategy library stores multi-dimensional, dynamically updatable companionship solutions;

[0036] The interactive execution control module is used to call the corresponding scheme and generate a dedicated interactive script to control the orderly execution of each execution unit;

[0037] The real-time feedback monitoring module is used to collect patient interaction feedback information and form a structured dataset;

[0038] The closed-loop optimization management module is used to evaluate the interaction effect and dynamically adjust the interaction strategy.

[0039] The central coordination and control module is used to coordinate the working sequence of each module and ensure that the interaction process is executed in a coherent and orderly manner.

[0040] Preferably, the multimodal precision acquisition module integrates a flexible skin-fitting physiological sensor, an anti-motion blur visual recognition unit, and an environmental perception component;

[0041] The dynamic physiological signals include heart rate, muscle rigidity, tremor frequency, and respiratory rate. The behavioral characteristics include gait parameters, limb movement amplitude, facial micro-expressions, and vocal emotional cues. The environmental dynamic information includes scene type, noise level, light intensity, temperature, and humidity. The physiological sensors adopt a flexible skin-fitting design, and the limb interaction actions are preset with safe amplitude and force. The environmental adjustment parameters meet the patient's comfort threshold. At the same time, the system has hardware self-testing and emergency handling functions to avoid causing physical discomfort or safety risks to the patient during interaction and to ensure long-term reliability.

[0042] Preferably, the personalized needs parsing module stores the patient's historical interaction data, disease progress records, and personalized tags, including age, personality preferences, rehabilitation goals, and contraindications;

[0043] The companionship solutions in the intelligent interaction strategy library correspond one-to-one with different disease states, emotional states, and needs. The movement amplitude and frequency of the gentle interaction action library are based on the preset threshold of the limb movement range of Parkinson's patients and are dynamically adjusted through environmental perception and collection strategies to adapt to different scenarios such as bedrooms, living rooms, and rehabilitation rooms. Moreover, the state feature library and the interaction strategy library support dynamic updates and can cover patients in different disease stages of early, middle, and late Parkinson's disease, and have wide applicability.

[0044] In summary, the technical effects and advantages of this invention are as follows:

[0045] 1. In this invention, by combining multimodal precise acquisition with the joint recognition of the patient's condition and emotions, the synchronous and precise capture of the patient's condition and emotional state is achieved. The interaction strategy is specifically matched to the core needs such as rehabilitation assistance and emotional comfort, which solves the problem of the mismatch between the "indiscriminate interaction" of traditional companion robots and the personalized needs of patients.

[0046] 2. In this invention, the interaction scheme is optimized based on the patient's personalized tags, the gentle limb movements are adapted to the patient's range of motion, the voice content and rehabilitation guidance are tailored to the patient's acceptance ability, and the data accuracy in complex scenarios is ensured through targeted collection strategies, which greatly improves the patient's interaction cooperation and experience.

[0047] 3. In this invention, by collecting feedback in real time and evaluating the effects in multiple dimensions, the interaction strategy parameters are dynamically adjusted and the patient data is updated, forming a closed-loop mechanism of "collection-identification-interaction-feedback-optimization". This ensures that the interaction plan continues to adapt to the patient's condition and emotional changes, and maintains a stable companionship and assistance effect in the long term.

[0048] 4. In this invention, the intelligent interactive strategy library integrates emotion-oriented content with tiered rehabilitation guidance. It can alleviate patients' negative emotions through voice soothing and gentle physical interaction, and provide rehabilitation training support under the premise of adapting to the condition. It achieves an organic combination of emotional companionship and rehabilitation assistance, and fully meets the needs of patients.

[0049] 5. In this invention, the physiological sensor adopts a flexible skin-fitting design, the limb interaction action is preset with a safe range and force, the environmental adjustment parameters meet the patient's comfort threshold, and the system has hardware self-test and emergency handling functions to avoid causing physical discomfort or safety risks to the patient during the interaction process and ensure the reliability of long-term use.

[0050] 6. In this invention, patient status recognition, needs analysis and interactive execution are completed automatically, reducing reliance on family members or caregivers. It is especially suitable for daily companionship and rehabilitation assistance during the stable period of the disease, effectively alleviating the time and energy pressure of human companionship and improving the convenience of care for Parkinson's patients.

[0051] 7. In this invention, the environmental perception and acquisition strategy is dynamically adjusted to adapt to different scenarios such as bedrooms, living rooms, and rehabilitation rooms. The state feature library and interaction strategy library support dynamic updates, which can cover patients in different stages of Parkinson's disease, including early, middle and late stages, and have wide applicability. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the interactive method of an emotional companion robot for Parkinson's disease according to the present invention.

[0053] Figure 2 This is a schematic diagram of an interactive robot system for emotional companionship in Parkinson's disease according to the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] refer to Figures 1-2 The method for interacting with an emotional companion robot for Parkinson's disease, as shown, includes the following steps:

[0056] S1. Interaction initialization and scene adaptation: After the emotional companion robot is started, it completes hardware self-test through the multimodal precision acquisition module, collects the basic physiological signals, initial behavioral actions and real-time environmental information of Parkinson's patients, and establishes the interaction initial dataset.

[0057] S2. Accurate collection of multimodal information: Based on the initial dataset, a targeted collection strategy is triggered to continuously supplement the patient's dynamic physiological data, real-time limb movement range, gait characteristics and voice emotion cues.

[0058] The formula for dynamically adjusting the sampling frequency is:

[0059]

[0060] in, Based on the sampling frequency, To adjust the coefficient, The instability coefficient of the condition;

[0061] S3. Joint recognition of disease condition and emotion: It calls up the preset Parkinson's disease patient state feature database and uses a weighted fusion algorithm to cross-validate the collected data to accurately identify the patient's disease condition and emotional state.

[0062] The feature fusion formula is:

[0063]

[0064] in, For the first Weights of class data For the first Normalized eigenvalues ​​of class data;

[0065] S4. Personalized interaction needs analysis: Combining the patient's historical interaction data, disease progression records and current recognition results, analyze the patient's core needs and clarify the priority of those needs;

[0066] The formula for calculating demand priority is as follows:

[0067]

[0068] in, , These are the weighting coefficients. Assess the urgency of the illness. Rate the intensity of emotional needs;

[0069] S5. Dynamic generation of interaction strategies: Based on the priority of needs, the intelligent interaction strategy library is called, and the details of the plan are optimized by combining the patient's personalized tags to generate exclusive interaction scripts.

[0070] The formula for calculating the suitability of the solution is:

[0071]

[0072] in, These are standard parameters for the strategy library. Personalized parameters for patients, The number of parameter dimensions;

[0073] S6. Interactive Execution and Real-time Feedback Acquisition: The robot executes operations according to a dedicated interactive script, and simultaneously collects patient feedback information in real time through a multimodal precision acquisition module to form an interactive feedback dataset.

[0074] S7. Interaction effect evaluation and strategy iteration: Based on the feedback dataset, an effect evaluation model is built to evaluate the interaction quality from multiple dimensions. When the preset threshold is not reached, the interaction strategy parameters are adjusted and the relevant data is updated to form a closed-loop optimization mechanism.

[0075] One of them is an interactive system for an emotional companion robot for Parkinson's disease, which includes: a multimodal precision acquisition module, a state joint recognition module, a personalized needs analysis module, an intelligent interaction strategy library, an interaction execution control module, a real-time feedback monitoring module, a closed-loop optimization management module, and a central coordination control module.

[0076] The multimodal precision acquisition module is used to collect dynamic physiological signals, behavioral characteristics and environmental dynamic information of Parkinson's patients in layers, and to perform noise reduction and standardization processing.

[0077] The state joint recognition module has a built-in Parkinson's disease patient state feature library and weighted fusion algorithm model, which is used to simultaneously identify the patient's disease state and emotional state;

[0078] The personalized needs analysis module is used to analyze the patient's core needs and needs priority based on the identification results;

[0079] The intelligent interaction strategy library stores multi-dimensional, dynamically updatable companionship solutions;

[0080] The interactive execution control module is used to call the corresponding scheme and generate a dedicated interactive script to control the orderly execution of each execution unit;

[0081] The real-time feedback monitoring module is used to collect patient interaction feedback information and form a structured dataset;

[0082] The closed-loop optimization management module is used to evaluate the interaction effect and dynamically adjust the interaction strategy;

[0083] The central coordination and control module is used to coordinate the working sequence of each module and ensure that the interaction process is executed in a coherent and orderly manner.

[0084] I. Overall Implementation Overview

[0085] This embodiment discloses an interactive method and system for an emotional companion robot for Parkinson's disease. It aims to achieve condition-adaptive emotional companionship for Parkinson's patients through multimodal perception, accurate recognition, personalized interaction, and closed-loop optimization, balancing emotional comfort and rehabilitation assistance needs, and improving the relevance and effectiveness of the interaction. The implementation details of the method and system are described in detail below, using specific application scenarios.

[0086] II. Specific Implementation Steps of the Interaction Method

[0087] (I) S1: Interaction Initialization and Scene Adaptation

[0088] The emotional companion robot starts automatically after being powered on. The multimodal precision acquisition module first performs a hardware self-test, including signal connectivity detection and calibration of the flexible skin-fitting physiological sensor, the anti-motion blur visual recognition unit, and the environmental perception component, to ensure that each acquisition component is working properly.

[0089] After the self-check is completed, the robot collects the patient's basic physiological signals through physiological sensors, specifically heart rate through wristband sensors, muscle rigidity through contact pressure sensors, and limb tremor frequency through visual recognition units; it also collects the patient's initial behavioral movements through visual recognition units, including limb posture and facial expressions in a static state; and it collects real-time environmental information through environmental perception components, including the type of scene, ambient noise decibels, and light intensity parameters.

[0090] All collected data is transmitted to the robot's main control unit via the data bus, and stored according to the categories of "physiological signals - behavioral actions - environmental information" to establish an initial interactive dataset, providing a data foundation for triggering subsequent targeted data collection strategies.

[0091] (ii) S2: Accurate acquisition of multimodal information

[0092] The main control unit performs a preliminary analysis of the initial dataset from the interaction and triggers corresponding targeted data collection strategies:

[0093] If the patient's tremor frequency is ≥3 times / second and muscle rigidity is ≥3 in the initial data, the patient is judged to be an unstable patient, and the physiological signal acquisition frequency is increased to 5Hz, and the dynamic physiological data is updated once every 2 seconds.

[0094] If the ambient noise level is ≥60dB, it is determined to be a noisy environment. The ambient noise filtering algorithm is activated, and the frame rate of the visual recognition unit for recognizing actions is increased to reduce the impact of noise and motion blur on recognition accuracy.

[0095] During the data collection process, dynamic physiological data, real-time limb movement range, gait characteristics, and vocal emotion cues of the patients were continuously supplemented, and the collected data were subjected to real-time noise reduction processing to ensure data accuracy.

[0096] (III) S3: Joint Recognition of Illness and Emotion

[0097] The system calls upon a pre-defined database of Parkinson's disease patient status characteristics. This database is generated based on clinical case data and machine learning training. It includes physiological and behavioral characteristic parameters for different disease stages such as early, middle, and late stages, as well as voice and facial expression characteristic parameters for different emotional types such as calmness, anxiety, irritability, loneliness, and depression.

[0098] The collected data were cross-validated using a weighted fusion algorithm. The weights of physiological signals in the algorithm were dynamically allocated based on the severity of the patient's condition: the weight of physiological signals was set to 0.3 for mild cases, 0.5 for moderate cases, and 0.7 for severe cases, ensuring that condition-related data dominated the identification process.

[0099] Through feature matching and cross-validation, the system accurately identifies the patient's condition and emotional state, and the identification results are transmitted to the personalized needs analysis module in real time.

[0100] (iv) S4: Analysis of Personalized Interaction Needs

[0101] The personalized needs analysis module retrieves stored patient historical interaction data and disease progress records, and combines them with the current condition and emotional state recognition results to analyze the patient's core needs.

[0102] Core needs include emotional support, rehabilitation assistance, environmental adjustment, and social companionship. The priority of these needs is determined based on the urgency of the patient's condition and the intensity of their emotional needs.

[0103] If the patient is experiencing a tremor attack or muscle rigidity, the condition is the most urgent and rehabilitation support needs should be prioritized.

[0104] If the patient's emotional state is anxious, irritable, or depressed, and the condition is stable, the need for emotional comfort should be prioritized.

[0105] If the environmental information indicates that the light is too strong or the noise is too loud, and the patient is not in a state of emergency or strong emotional fluctuation, the need for environmental adjustment should be prioritized.

[0106] (v) S5: Dynamic generation of interaction strategies

[0107] The system invokes an intelligent interaction strategy library based on priority of needs. This library stores multi-dimensional, dynamically updatable companionship solutions, including:

[0108] Emotion-oriented voice content (such as comforting, encouraging, and companionship phrases, adapted to different emotional states);

[0109] Gentle physical interactions (such as patting the shoulder, shaking hands, etc., with an amplitude of ≤30cm and a frequency of ≤1 time / second, based on the preset threshold of limb range of motion for Parkinson's patients).

[0110] Step-by-step rehabilitation guidance (from simple to complex limb training guidance, such as finger flexion and extension, arm swing, etc., adapted to different severity of the condition);

[0111] Environmental parameter adaptation and adjustment (such as instructions to adjust light intensity to 300-500 lux and control environmental noise to ≤40dB).

[0112] The treatment plan is optimized by incorporating individual patient profiles (age, personality traits, and disease severity): for example, slower speech and higher volume are used for elderly patients; more interactive language is added for outgoing patients; and rehabilitation guidance actions are simplified for patients with severe conditions. Ultimately, a customized interaction script is generated, clearly defining the content, sequence, duration, and execution parameters of the interactions.

[0113] (vi) S6: Interactive execution and real-time feedback collection

[0114] The interactive execution control module controls each execution unit to perform operations in an orderly manner based on a dedicated interactive script:

[0115] The voice output unit plays the optimized voice content, with the volume controlled at 60-70dB and the speaking speed adjusted according to the patient's age (60-80 words / minute for elderly patients).

[0116] The limb movement execution component completes gentle limb interaction movements, and the force of the movement is fed back in real time by a pressure sensor to ensure that the force is ≤5N and to avoid causing discomfort to the patient.

[0117] If environmental adjustment is required, control commands are sent to associated smart devices (such as lights, air conditioners, and audio equipment) through the environmental adjustment control interface to adjust environmental parameters.

[0118] During the interactive process, the multimodal precision acquisition module synchronously collects the patient's feedback information in real time:

[0119] Physiological signal changes (whether the heart rate tends to stabilize, whether the tremor frequency decreases, and whether the degree of muscle rigidity is relieved);

[0120] Behavioral response (limb coordination, such as whether the rehabilitation guidance is followed to complete the movement, coordination score 0-10; facial expression changes, such as whether the expression changes from agitation to calmness).

[0121] The voice feedback content (patient's responses, requests, etc.) and emotional tendencies (emotional state re-identified through voice features) are categorized and organized to form a structured interactive feedback dataset.

[0122] (vii) S7: Interaction effect evaluation and strategy iteration

[0123] The closed-loop optimization management module has a built-in performance evaluation model that assesses interaction quality from three dimensions based on the interaction feedback dataset:

[0124] Degree of emotional improvement: Compare the patient's emotional state before and after the interaction. If the patient's emotions change from negative emotions (anxiety, irritability, depression) to calm or positive emotions, it is considered effective.

[0125] Symptom relief effect: Compare the patient's physiological signals before and after the interaction. If the tremor frequency decreases by ≥20% and the muscle rigidity decreases by ≥1 grade, it is considered effective.

[0126] Interaction cooperation level: If the patient's limb cooperation score is ≥6 points, it is considered effective.

[0127] The preset thresholds of the assessment model are determined based on a combination of the patient's historical best interaction results, medical rehabilitation standards, and emotional improvement benchmarks. For example, the preset threshold for the degree of emotional improvement is "the rate of relief of negative emotions after interaction is ≥50%", the preset threshold for the relief of discomfort is "the tremor frequency is reduced by ≥20% or the degree of muscle rigidity is reduced by ≥1 grade", and the preset threshold for the degree of cooperation in interaction is "the cooperation score is ≥6 points".

[0128] If the evaluation results do not reach the preset threshold, the closed-loop optimization management module automatically adjusts the corresponding scheme parameters in the intelligent interaction strategy library: for example, if the emotional comfort effect is not good, it changes the type of voice content, adjusts the speech rate, or adds physical interaction actions; if the rehabilitation assistance effect is not good, it simplifies the rehabilitation guidance actions and extends the guidance time. At the same time, it updates the patient's personalized tags and historical interaction data to provide an optimization basis for the generation of the next interaction strategy, forming a closed-loop optimization mechanism of "collection-recognition-interaction-feedback-optimization".

[0129] III. Specific Implementation Details of the Interactive System

[0130] (I) Hardware and software configuration of each module of the system

[0131] Multimodal precision acquisition module: integrates a flexible skin-fitting physiological sensor (wristband design, adapted to the patient's limb tremor state), an anti-motion blur visual recognition unit (high-definition camera + image preprocessing chip), and environmental perception components (microphone, photosensor, temperature and humidity sensor); on the software level, it is equipped with data noise reduction algorithm, adaptive noise cancellation technology, and image enhancement algorithm to ensure the accuracy and stability of the acquired data.

[0132] State Joint Recognition Module: It adopts an STM32 main control chip and has a built-in weighted fusion algorithm model trained in Python and a state feature library of Parkinson's disease patients. The feature library supports updates via the cloud to ensure the timeliness of the recognition model.

[0133] Personalized needs analysis module: Equipped with a 16GB storage chip, it stores patients' historical interaction data, disease progression records, and personalized tags, and achieves accurate matching between recognition results and needs analysis through data association algorithms.

[0134] Intelligent Interaction Strategy Library: Built on a cloud server, it stores multi-dimensional companionship plans, supports dynamic updates of plan content based on clinical data and user feedback, and reserves a local caching module to ensure normal access to core plans even when offline.

[0135] Interactive execution control module: integrates a voice output unit (high-fidelity speaker), a limb motion execution component (servo-driven robotic arm equipped with a pressure sensor), and an environmental control interface (Wi-Fi / Bluetooth communication module), supporting multi-device linkage control.

[0136] Real-time feedback monitoring module: Shares hardware resources with the multimodal precision acquisition module, extracts relevant feedback data through data filtering algorithms, forms a structured dataset, and ensures the relevance of feedback information.

[0137] Closed-loop optimization management module: Equipped with an ARM Cortex-A9 processor, it runs performance evaluation models and parameter adjustment algorithms, supporting both real-time optimization and offline iteration modes.

[0138] Central Coordination and Control Module: Employs an FPGA chip, responsible for coordinating the working timing of each module, handling data transmission and command interaction between modules, and has a built-in fault self-checking program (periodically checks the operating status of each module) and emergency handling mechanism (automatically switches to a backup plan when a module fails to ensure that basic backup functions are normal).

[0139] (II) System Collaboration Process

[0140] After the system starts up, the central coordination and control module first triggers the multimodal precision acquisition module to perform hardware self-checks and initial acquisition, generating an initial interactive dataset. Subsequently, the coordination state joint recognition module completes the joint recognition of illness and emotion, and transmits the recognition results to the personalized needs analysis module. After the personalized needs analysis module analyzes the core needs and priorities, the central coordination and control module calls the intelligent interaction strategy library and the interaction execution control module to generate and execute a dedicated interaction script. During the interaction, the real-time feedback monitoring module collects feedback information synchronously, and the closed-loop optimization management module evaluates the interaction effect and optimizes the strategy based on the feedback data. The central coordination and control module controls the timing and data flow of each link throughout the process to ensure that the entire interaction process is coherent and orderly executed.

[0141] IV. Implementation Results Description

[0142] The interactive method and system described in this embodiment can accurately identify the condition and emotional state of Parkinson's patients, generate personalized companionship plans, and achieve the following effects in practical applications:

[0143] Emotional improvement rate ≥60%: Through targeted verbal comfort and physical interaction, patients' negative emotions such as anxiety, irritability, and loneliness are effectively alleviated;

[0144] Symptom relief rate ≥50%: Through rehabilitation guidance and environmental adaptation, patients can be helped to relieve symptoms such as tremor attacks and muscle rigidity.

[0145] Interaction compliance rate ≥70%: Solutions optimized based on personalized tags are more in line with patient needs, improving patients' willingness to interact and their compliance rate;

[0146] Closed-loop optimization improves effectiveness by ≥30%: Through continuous feedback and strategy iteration, the companionship program is constantly adapted to changes in the patient's condition and needs, maintaining good interaction effects in the long term.

[0147] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for interacting with a Parkinson's disease emotional companion robot, characterized in that, Includes the following steps: S1. Interaction initialization and scene adaptation: After the emotional companion robot is started, it completes hardware self-test through the multimodal precision acquisition module, collects the basic physiological signals, initial behavioral actions and real-time environmental information of Parkinson's patients, and establishes the interaction initial dataset. S2. Accurate collection of multimodal information: Based on the initial dataset, a targeted collection strategy is triggered to continuously supplement the patient's dynamic physiological data, real-time limb movement range, gait characteristics and voice emotion cues. S3. Joint recognition of disease condition and emotion: It calls up the preset Parkinson's disease patient state feature database and uses a weighted fusion algorithm to cross-validate the collected data to accurately identify the patient's disease condition and emotional state. S4. Personalized interaction needs analysis: Combining the patient's historical interaction data, disease progression records and current recognition results, analyze the patient's core needs and clarify the priority of those needs; S5. Dynamic generation of interaction strategies: Based on the priority of needs, the intelligent interaction strategy library is called, and the details of the plan are optimized by combining the patient's personalized tags to generate exclusive interaction scripts. S6. Interactive Execution and Real-time Feedback Acquisition: The robot executes operations according to a dedicated interactive script, and simultaneously collects patient feedback information in real time through a multimodal precision acquisition module to form an interactive feedback dataset. S7. Interaction effect evaluation and strategy iteration: Based on the feedback dataset, an effect evaluation model is built to evaluate the interaction quality from multiple dimensions. When the preset threshold is not reached, the interaction strategy parameters are adjusted and the relevant data is updated to form a closed-loop optimization mechanism.

2. The method of claim 1, wherein the method is a method of interacting with a Parkinson's disease emotional companion robot. The basic physiological signals mentioned in S1 include heart rate, muscle rigidity, and tremor frequency; the initial behavioral actions include static limb posture and facial expression; and the real-time environmental information includes the scene type, ambient noise level, and light intensity parameters. 3.The method of claim 1, wherein the method further comprises: determining a user’s emotional state based on the user’s voice and the user’s facial expression; and providing a response to the user based on the determined emotional state. The targeted acquisition strategy described in S2 specifically includes: increasing the frequency of physiological signal acquisition for patients with unstable conditions, and enhancing environmental noise filtering and behavioral action recognition accuracy in noisy environments.

4. The method of claim 1, wherein the method is a method of interacting with a Parkinson's disease emotional companion robot. The Parkinson's disease patient state feature database described in S3 contains feature parameters for different disease stages and emotional types. The physiological signal weights of the weighted fusion algorithm are dynamically allocated based on the severity of the patient's condition. The disease states include stable state, tremor attack state, myotonic discomfort state, and bradykinesia state. The emotional states include calm, anxiety, irritability, loneliness, and depression.

5. The method of claim 1, wherein the method is a method of interacting with a Parkinson's disease emotional companion robot, the method comprising: The core needs described in S4 include the need for emotional comfort, rehabilitation assistance, environmental adjustment, or social companionship. The priority of these needs is determined based on the urgency of the patient's condition and the intensity of their emotional needs.

6. The method of claim 1, wherein the method is a method of interacting with a Parkinson's disease emotional companion robot. The multi-dimensional companionship solution of the intelligent interaction strategy library described in S5 includes voice comfort content, gentle physical interaction actions, step-by-step rehabilitation guidance prompts, and environmental parameter adaptation and adjustment. The range and frequency of the gentle interaction action library are preset based on the limb activity range threshold of Parkinson's patients. The patient's personalized tags include age, personality preferences, and disease severity.

7. The method of claim 1, wherein the method is a method of interacting with a Parkinson's disease emotional companion robot. The multi-dimensional assessment described in S7 includes three dimensions: degree of emotional improvement, effect of relieving discomfort, and degree of cooperation in interaction. The preset threshold is determined based on the patient's historical best interaction effect, medical rehabilitation standards, and emotional improvement benchmark value. 8.A Parkinson's disease emotional companion robot interaction system, characterized in that, include: The system includes a multimodal precision acquisition module, a state joint identification module, a personalized demand analysis module, an intelligent interaction strategy library, an interaction execution control module, a real-time feedback monitoring module, a closed-loop optimization management module, and a central coordination and control module. The multimodal precision acquisition module is used to collect dynamic physiological signals, behavioral characteristics and environmental dynamic information of Parkinson's patients in layers, and to perform noise reduction and standardization processing. The state joint recognition module incorporates a Parkinson's disease patient state feature library and a weighted fusion algorithm model to simultaneously identify the patient's disease state and emotional state. The personalized needs analysis module is used to analyze the patient's core needs and needs priority based on the identification results. The intelligent interaction strategy library stores multi-dimensional, dynamically updatable companionship solutions; The interactive execution control module is used to call the corresponding scheme and generate a dedicated interactive script to control the orderly execution of each execution unit; The real-time feedback monitoring module is used to collect patient interaction feedback information and form a structured dataset; The closed-loop optimization management module is used to evaluate the interaction effect and dynamically adjust the interaction strategy. The central coordination and control module is used to coordinate the working sequence of each module and ensure that the interaction process is executed in a coherent and orderly manner.

9. The emotional companion robot interaction system for Parkinson's disease according to claim 8, wherein: The multimodal precision acquisition module integrates a flexible skin-fitting physiological sensor, an anti-motion blur visual recognition unit, and an environmental perception component; The dynamic physiological signals include heart rate, muscle rigidity, tremor frequency, and respiratory rate; the behavioral characteristics include gait parameters, limb movement amplitude, facial micro-expressions, and vocal emotional cues; and the environmental dynamic information includes scene type, noise level, light intensity, and temperature and humidity.

10. The Parkinson's disease emotional companion robot interaction system according to claim 8, characterized in that: The personalized needs analysis module stores the patient's historical interaction data, disease progress records, and personalized tags, which include age, personality preferences, rehabilitation goals, and contraindications. The companionship plans in the intelligent interaction strategy library correspond one-to-one with different disease states, emotional states, and needs. The range and frequency of the gentle interactive action library are preset based on the limb movement range threshold of Parkinson's patients.

Images