Transcranial focused ultrasound therapy-based depression patient emotion analysis platform

Abstract

The invention discloses a depression patient emotion analysis platform based on transcranial focused ultrasound treatment, and relates to the technical field of medical intelligent signal processing, and the depression patient emotion analysis platform comprises a data collection module which synchronously collects physiological signals, behavior data and subjective emotion feedback of a patient; the treatment parameter synchronization module obtains ultrasonic treatment core parameters; the emotional feature extraction module preprocesses the data and extracts high-dimensional emotional feature vectors; the intelligent emotion analysis module evaluates depression degree and treatment response through a fusion algorithm model; the feedback regulation and control module pushes parameter regulation suggestions, intervention prompts and emotion regulation guidance to construct a closed-loop regulation and control system; the data storage and management module stores data in a classified manner; the man-machine interaction module provides data display and operation functions, and meets the requirements of the medical care terminal and the patient terminal. According to the invention, multi-modal data and intelligent analysis are integrated, and the emotional state and treatment response of a patient are accurately evaluated; multidisciplinary collaboration and data safety guarantee promote standardized and precise development of depression treatment.

Description

Technical Field

[0001] This invention relates to the field of medical intelligent signal processing technology, and in particular to a mood analysis platform for patients with depression based on transcranial focused ultrasound therapy. Background Technology

[0002] Depression, a prevalent mental illness worldwide, severely impacts patients' physical and mental health and social functioning. Transcranial focused ultrasound (TUS) therapy, due to its non-invasive and highly targeted advantages, has become an important clinical treatment method. Emotional state, as a core indicator for assessing treatment effectiveness and adjusting treatment plans, requires precise monitoring and dynamic analysis to improve treatment efficacy. However, current emotional assessments in depression treatment largely rely on subjective observation by healthcare professionals and periodic self-rating scale feedback from patients. This approach suffers from problems such as assessment lag, strong subjectivity, and fragmented data, failing to capture subtle emotional fluctuations in real time during treatment and hindering the establishment of a correlation between emotional changes and treatment parameters. Consequently, adjustments to treatment plans lack timely and objective data support.

[0003] Existing emotion analysis technologies have significant limitations and are ill-suited to the clinical needs of transcranial focused ultrasound (TCU) therapy. Regarding data collection, they are often limited to single-dimensional data, relying solely on physiological signals or subjective scales, lacking simultaneous collection and integration of behavioral data and treatment parameters. This results in a one-sided emotional assessment that fails to fully reflect the patient's true emotional state. In terms of analytical models, they often employ single algorithms, failing to fully integrate the complementarity of multimodal data and lacking targeted optimization for the treatment scenario. This leads to insufficient accuracy in emotion assessment and difficulty in predicting treatment response and emotional improvement trends. Regarding feedback mechanisms, a complete "collection-analysis-feedback-intervention" closed loop has not been established. Analysis results are only used as a reference and cannot directly guide adjustments to treatment parameters or personalized interventions, resulting in a disconnect between treatment and emotion management.

[0004] Insufficient intelligence, collaboration, and data security further hinder clinical application. Existing systems lack self-learning capabilities, are unable to iteratively optimize models based on multi-center clinical data, and have a weak ability to adapt to the personalized needs of patients of different ages, disease stages, and severity. The lack of multidisciplinary medical team collaboration mechanisms makes it difficult for different roles to efficiently share data and collaboratively develop plans, affecting the timeliness and overall effectiveness of interventions. Inadequate medical data privacy protection measures pose risks of leakage and misuse of sensitive patient information, and the lack of cross-institutional data sharing mechanisms makes it difficult to integrate high-quality clinical resources to drive technological iteration. These problems hinder the improvement of the precision and individualization of transcranial focused ultrasound (TCU) treatment, restricting further optimization of the treatment effect for depression. Summary of the Invention

[0005] The present invention proposes an emotion analysis platform for patients with depression based on transcranial focused ultrasound therapy to solve the problems mentioned in the prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a mood analysis platform for patients with depression based on transcranial focused ultrasound therapy, comprising the following modules:

[0007] The data acquisition module is equipped with a high-precision electrocardiogram sensor, electroencephalogram acquisition electrodes, skin conductance sensor, high-definition facial image camera, high-fidelity voice microphone and digital self-assessment scale interaction unit. It simultaneously collects the patient's physiological signals, behavioral data and subjective emotional feedback during treatment. After acquisition, signal amplification and filtering noise reduction, the data is transmitted to the core processing module.

[0008] The treatment parameter synchronization module adopts a dual-link communication protocol of industrial Ethernet and 5G to establish a stable connection with the transcranial focused ultrasound treatment equipment and acquire core parameters in real time.

[0009] The emotion feature extraction module performs systematic preprocessing on the raw data and integrates multi-source data to extract high-dimensional structured emotion feature vectors.

[0010] The intelligent emotion analysis module is equipped with a fusion algorithm model of random forest, convolutional neural network and long short-term memory network. It takes emotion feature vector and synchronous treatment parameters as input, dynamically evaluates the patient's condition, and generates quantitative emotion scores and qualitative analysis reports.

[0011] The feedback and control module receives intelligent analysis results, sends parameter adjustment suggestions to the treatment equipment, pushes assessment reports and intervention prompts to the medical staff, and sends emotion regulation guidance to the patient.

[0012] The data storage and management module adopts a dual architecture of local encrypted storage and cloud distributed server, which stores various types of data in categories and supports multi-dimensional query, export and traceability;

[0013] The human-computer interaction module is configured with an interactive interface between a professional workstation for medical staff and a mobile terminal for patients. It displays relevant data and information in real time and supports adjustments by medical staff, data submission by patients, and viewing of guidance content.

[0014] Furthermore, it also includes an emotion state assessment module. This module integrates multi-dimensional features output from the emotion feature extraction module and constructs a comprehensive assessment model through a multi-factor weighted fusion algorithm, using a formula... Calculate the patient's real-time comprehensive emotional score, among which Assess the overall emotional state. These are the weighting coefficients for physiological signal characteristics. Standardized scores for physiological signals, These are the weighting coefficients for behavioral data features. Standardize the scores for behavioral data. These are the weighting coefficients for subjective scale features. Standardized scores for subjective scales. For treatment response feature weighting coefficients, Standardized score for treatment response.

[0015] Furthermore, it also includes a treatment response prediction and parameter optimization module. This module, based on historical assessment data and treatment parameter records from the intelligent emotion analysis module, predicts the patient's subsequent treatment response effect through a time-series trend analysis algorithm and a formula. Calculate the optimal adjustment value for ultrasound therapy intensity, where This is the intensity adjustment amount for ultrasound therapy. This is the treatment response sensitivity coefficient. The difference in the overall emotional score between two adjacent treatment cycles is... , This refers to the duration of the current treatment cycle. Based on the current treatment pulse repetition frequency, personalized treatment parameter optimization suggestions are generated according to the adjustment amount.

[0016] Furthermore, it also includes a remote monitoring and anomaly early warning module. This module uses 5G ultra-wideband communication technology to complete real-time data transmission between medical staff and patients in the treatment scenario. It sets three levels of early warning thresholds for emotional fluctuations, physiological abnormalities, and treatment responses. When it detects that the patient's emotional score fluctuation exceeds the set range, abnormal peaks occur in physiological signals, or the emotional score improvement is insufficient for two consecutive treatment cycles, the level of early warning mechanism is immediately triggered. It simultaneously records the time of the abnormality, relevant data, and early warning level, supporting medical staff to remotely view real-time data, issue emergency intervention instructions, and adjust treatment plans.

[0017] Furthermore, it also includes a model self-learning optimization module, which continuously collects multi-center clinical treatment data, emotion analysis results, and treatment effect feedback to build a large-scale labeled dataset. Through incremental learning algorithms, it iteratively trains the fusion algorithm model of the intelligent emotion analysis module, dynamically optimizing the model's feature extraction weights, classification decision boundaries, and evaluation parameters. At the same time, it generates personalized analysis sub-models based on differences in patient age, disease course, disease severity, treatment targets, and basic health conditions.

[0018] Furthermore, it also includes a multimodal data fusion enhancement module. This module uses an attention mechanism and feature-level fusion algorithm to deeply fuse four types of data: physiological signals, behavioral data, subjective scale data, and treatment parameters. It dynamically allocates feature weights to highlight information that is strongly correlated with emotional state, extracts specific EEG features from the region associated with EEG signals and ultrasound treatment targets, and combines the emotional consistency verification of facial expressions and voice intonation to generate a more recognizable high-dimensional emotional feature vector.

[0019] Furthermore, it also includes a medical-nursing collaborative interaction module. This module supports hierarchical permission management for multidisciplinary medical teams, defining the operation permissions and data access scope for different roles. Attending physicians can view complete emotion analysis data, make suggestions to adjust treatment parameters, and sign intervention plans. Nurses can receive early warning information and record clinical observation results. Rehabilitation therapists can obtain emotion improvement trends and develop personalized emotion regulation training plans. Psychotherapists can view details of emotion fluctuations. The operation logs of all roles are recorded in real time and cannot be tampered with. It supports online communication and opinion collaboration within the team, generating a unified patient treatment and emotion management plan.

[0020] Furthermore, it also includes a patient emotion regulation guidance module. Based on the assessment results of the intelligent emotion analysis module and the treatment stage, this module automatically matches personalized emotion regulation plans. For low mood, it pushes breathing relaxation training, mindfulness meditation guided audio, and positive psychological suggestion graphic tutorials. For mood fluctuation, it provides cognitive behavioral regulation techniques, stress management suggestions, and emotion diary templates. For good treatment response, it pushes rehabilitation progress incentive content, interest activity recommendations, and social interaction guidance. The plan push is reasonably arranged in combination with the patient's daily routine and treatment time, and the patient's regulation plan is recorded at the same time.

[0021] Furthermore, it also includes a data security and privacy protection module. This module uses end-to-end encryption algorithms to encrypt the transmission and storage of sensitive information throughout the process, uses blockchain technology to trace and prevent tampering of key medical data, adopts a dual verification mechanism of biometrics and dynamic passwords, strictly follows relevant standards for medical data privacy protection, desensitizes patient identity information, sets up data access audit logs, and comprehensively records data operations.

[0022] Furthermore, it also includes a multi-center data sharing and model iteration module. This module adopts standardized data formats and interface protocols to support the interconnection and secure sharing of platform data from different medical institutions. It establishes a multi-center joint research database, performs data aggregation and analysis after removing patient privacy information, and uses federated learning algorithms to train the intelligent emotion analysis model across institutions by combining multi-center data without sharing original sensitive data. It integrates clinical experience and data resources from different regions and different levels of diagnosis and treatment.

[0023] Compared with existing technologies, the beneficial effects of this invention are:

[0024] This invention presents a mood analysis platform for patients with depression based on transcranial focused ultrasound therapy, achieving a multi-dimensional, intelligent, and closed-loop breakthrough in mood analysis during depression treatment, with significant core advantages. The multimodal data acquisition system breaks through the limitations of traditional single-dimensional methods, simultaneously integrating physiological signals, behavioral data, subjective feedback, and treatment parameters. Through systematic preprocessing and deep fusion, it comprehensively captures patient-related mood information, providing a solid data foundation for accurate analysis and addressing the pain points of traditional assessments being one-sided and fragmented.

[0025] The intelligent analysis capabilities have been significantly enhanced. By integrating multiple advanced algorithms to build a model, it can not only dynamically assess patients' depression levels and mood fluctuations, but also accurately predict treatment response effects, generating quantitative scores and qualitative reports. The model's self-learning optimization mechanism continuously integrates multi-center data, dynamically adjusts parameters and decision boundaries, and adapts to the personalized characteristics of different patients, significantly improving the accuracy of mood assessment and treatment prediction, and providing an objective and scientific basis for adjusting treatment plans.

[0026] A closed-loop control and collaborative interaction system deeply integrates treatment and emotion management. Analysis results are directly transformed into suggestions for adjusting treatment parameters, clinical intervention prompts, and guidance for patient emotion regulation, constructing a fully interconnected intervention system to ensure timely and precise intervention. The medical-nursing collaboration module supports hierarchical management and efficient collaboration within multidisciplinary teams, with traceable operation logs enhancing the overall comprehensiveness and professionalism of treatment plans. Personalized guidance plans for patients are tailored to their treatment stage and emotional state, increasing patient compliance and self-management abilities.

[0027] Data security and cross-institutional sharing capabilities ensure the clinical application value. End-to-end encryption, blockchain traceability, and dual verification are among the multiple measures implemented to comprehensively prevent data leakage and misuse risks, strictly adhering to privacy protection standards. Multi-center data sharing and federated learning mechanisms integrate high-quality resources while protecting privacy, continuously optimizing model generalization capabilities and clinical applicability. This promotes the standardization, precision, and personalization of transcranial focused ultrasound (TUS) treatment for depression, providing strong technical support for improving overall treatment outcomes and patient prognosis. Attached Figure Description

[0028] Figure 1 This is a schematic block diagram of the emotion analysis platform for patients with depression based on transcranial focused ultrasound therapy proposed in this invention;

[0029] Figure 2 A line graph showing the changes in the comprehensive emotional score during the treatment period;

[0030] Figure 3 Bar charts evaluating the accuracy of different sentiment analysis modules;

[0031] Figure 4 Pie chart showing the contribution of features to multimodal data. Detailed Implementation

[0032] 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.

[0033] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0034] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. Furthermore, the terms "installed," "connected," and "linked" should be interpreted broadly; for example, they may refer to a fixed connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The invention will now be described in further detail with reference to the accompanying drawings.

[0035] Reference Figures 1 to 4 A mood analysis platform for patients with depression based on transcranial focused ultrasound therapy, comprising the following modules:

[0036] The data acquisition module is equipped with a high-precision ECG sensor, EEG acquisition electrodes, skin conductance sensor, high-definition facial image camera, high-fidelity voice microphone, and digital self-assessment scale interaction unit. It simultaneously collects physiological signals, behavioral data, and subjective emotional feedback from patients during treatment. Physiological signals include heart rate variability, EEG αβθ band energy, and skin conductance fluctuations. Behavioral data covers the displacement of 68 key facial feature points, speech fundamental frequency, speech rate, and limb movement amplitude. Subjective emotional feedback is submitted in real time through a standardized scale. All data is collected, amplified, filtered, and noise-reduced at a frequency of once per second before being transmitted to the core processing module.

[0037] The treatment parameter synchronization module adopts a dual-link communication protocol of industrial Ethernet and 5G to establish a stable connection with the transcranial focused ultrasound therapy equipment. It acquires core parameters such as treatment target coordinates, ultrasound intensity, pulse repetition frequency, treatment duration, cumulative number of treatments, and ultrasound beam focusing depth in real time. It adds millisecond-level timestamps to the collected data to achieve precise spatiotemporal correlation between emotional data and treatment parameters, and the analysis results are correlated and reliable.

[0038] The emotion feature extraction module systematically preprocesses the raw data, removes motion artifacts of physiological signals through Kalman filtering, extracts dynamic features of facial expressions using deep learning image algorithms, analyzes features such as tone changes, energy fluctuations, and pause frequency using speech emotion recognition technology, performs dimensional splitting and standardization transformation on digital scale data, and integrates multi-source data to extract high-dimensional structured emotion feature vectors.

[0039] The intelligent emotion analysis module is equipped with a fusion algorithm model of random forest, convolutional neural network and long short-term memory network. It takes emotion feature vector and synchronous treatment parameters as input, and learns the mapping relationship between emotion state and treatment parameters through model training. It dynamically assesses the patient's depression level, emotion fluctuation amplitude, treatment response efficiency and emotion improvement potential, and generates quantitative emotion score and qualitative analysis report.

[0040] The feedback and control module receives the output results from the intelligent emotion analysis module, sends personalized parameter adjustment suggestions to the transcranial focused ultrasound treatment device, pushes emotion assessment reports and clinical intervention prompts to the medical staff workstation, and sends customized emotion regulation guidance plans to the patient's mobile terminal, thus constructing a closed-loop control system of "data collection-intelligent analysis-precise feedback-clinical intervention".

[0041] The data storage and management module adopts a dual architecture of local encrypted storage unit and cloud distributed server. It categorizes and stores raw collected data, preprocessed feature data, emotion analysis results, treatment parameter records, and interactive operation logs. It supports multi-dimensional querying, exporting, and tracing by patient ID, treatment stage, emotion level, and time range. The data storage cycle covers clinical follow-up, efficacy evaluation, and research needs.

[0042] The human-computer interaction module is configured with a professional workstation touch screen for medical staff and an interactive interface for the patient's mobile terminal. It displays the patient's emotional quantitative score, emotional change trend curve, treatment parameter details, and intervention suggestion list in real time. It supports medical staff to adjust the analysis model parameters, set emotional warning thresholds, and enter clinical assessment opinions. It also supports patients to submit self-assessment data, view the progress of emotional improvement, and receive treatment cooperation guidance and rehabilitation incentive content.

[0043] This invention also includes a precise emotional state assessment module. This module integrates multi-dimensional features output by the emotional feature extraction module and constructs a comprehensive assessment model through a multi-factor weighted fusion algorithm, using a formula... Calculate the patient's real-time comprehensive emotional score, among which The overall emotional score is rated from 0 to 100. These are the weighting coefficients for physiological signal characteristics. The standardized score for physiological signals is set to a value from 0 to 100. These are the weighting coefficients for behavioral data features. The standardized scores for behavioral data range from 0 to 100. These are the weighting coefficients for subjective scale features. The standardized scores for the subjective scale range from 0 to 100. For treatment response feature weighting coefficients, The standardized score for treatment response ranges from 0 to 100, and the weighting coefficients are optimized through training with multi-center clinical data and meet the following requirements. + + + =1, the comprehensive score intuitively quantifies the degree of depression relief and emotional stability of patients, and provides a precise quantitative basis for adjusting treatment plans and evaluating efficacy.

[0044] This invention also includes a treatment response prediction and parameter optimization module. This module, based on historical evaluation data and treatment parameter records from the intelligent emotion analysis module, predicts the patient's subsequent treatment response using a time-series trend analysis algorithm and a formula. Calculate the optimal adjustment value for ultrasound therapy intensity, where The unit for adjusting the intensity of ultrasound therapy is W / cm². This is the treatment response sensitivity coefficient. The difference in the overall emotional score between two adjacent treatment cycles is... , The duration of the current treatment cycle is measured in minutes. The current treatment pulse repetition frequency is measured in Hz. Based on the adjustment amount, personalized treatment parameter optimization suggestions are generated. When the mood score improves, the intensity is adjusted appropriately to consolidate the therapeutic effect. When the mood score does not improve, the parameters are optimized to improve the treatment's effectiveness.

[0045] This invention also includes a remote monitoring and anomaly early warning module. This module uses 5G ultra-wideband communication technology to realize real-time data transmission between the medical staff and the patient's treatment scenario. It sets three levels of early warning thresholds for emotional fluctuations, physiological abnormalities, and treatment response. When it detects that the patient's emotional score fluctuates beyond the set range within 48 hours, abnormal peaks appear in physiological signals, or the emotional score does not improve by the set value for two consecutive treatment cycles, the level early warning mechanism is immediately triggered. It sends audible and visual alarms and pop-up prompts to the medical staff and pushes comforting guidance and medical advice to the patient. It also records the time of the abnormality, relevant data, and early warning level, supporting medical staff to remotely view real-time data, issue emergency intervention instructions, and adjust treatment plans.

[0046] This invention also includes a model self-learning optimization module. This module continuously collects multi-center clinical treatment data, emotion analysis results, and treatment effect feedback to construct a large-scale labeled dataset. Through incremental learning algorithms, it iteratively trains the fusion algorithm model of the intelligent emotion analysis module, dynamically optimizing the model's feature extraction weights, classification decision boundaries, and evaluation parameters. At the same time, based on differences in patient age, disease course, disease severity, treatment targets, and basic health conditions, it generates personalized analysis sub-models to gradually improve the accuracy of emotion assessment, the accuracy of treatment response prediction, and the adaptability of parameter suggestions, thereby meeting the treatment needs of patients with different types of depression.

[0047] This invention also includes a multimodal data fusion enhancement module. This module employs an attention mechanism and a feature-level fusion algorithm to deeply fuse four types of data: physiological signals, behavioral data, subjective scale data, and treatment parameters. It dynamically allocates feature weights to highlight key information strongly correlated with emotional state, weakens noise data interference, extracts specific EEG features from regions associated with EEG signals and ultrasound treatment targets, and combines facial expressions and voice tone for emotional consistency verification. This strengthens the temporal series correlation of subjective scale data, generates more recognizable high-dimensional emotional feature vectors, and further improves the evaluation performance and anti-interference ability of the intelligent emotion analysis module.

[0048] This invention also includes a medical-nursing collaborative interaction module. This module supports hierarchical permission management for multidisciplinary medical teams, defining the operation permissions and data access scope for different roles such as attending physicians, nurses, rehabilitation therapists, and psychotherapists. Attending physicians can view complete emotion analysis data, make suggestions for adjusting treatment parameters, and sign intervention plans. Nurses can receive early warning information, assist patients in completing self-assessment scales, and record clinical observation results. Rehabilitation therapists can obtain emotion improvement trends and develop personalized emotion regulation training plans. Psychotherapists can view details of emotion fluctuations and provide targeted psychological interventions. The operation logs of all roles are recorded in real time and cannot be tampered with, supporting online communication and opinion collaboration within the team, generating unified patient treatment and emotion management plans, and improving the efficiency of multidisciplinary collaboration.

[0049] This invention also includes a patient emotion regulation guidance module. Based on the assessment results of the intelligent emotion analysis module and the treatment stage, this module automatically matches personalized emotion regulation plans. For low mood, it pushes breathing relaxation training, mindfulness meditation guided audio, and positive psychological suggestion graphic tutorials. For mood fluctuation, it provides cognitive behavioral regulation techniques, stress management suggestions, and emotion diary templates. For good treatment response, it pushes rehabilitation progress incentive content, interest activity recommendations, and social interaction guidance. The plan push is reasonably arranged in combination with the patient's daily routine and treatment time to avoid interfering with treatment and rest. At the same time, it records the patient's usage time, completion rate, and feedback rating of the regulation plan, providing data support for subsequent plan optimization.

[0050] This invention also includes a data security and privacy protection module. This module uses an end-to-end encryption algorithm to encrypt and transmit sensitive information such as collected data, analysis results, and treatment parameters throughout the entire process. It uses blockchain technology to achieve traceability and tamper-proofing of key medical data, employs a dual verification mechanism of biometrics and dynamic passwords to ensure account security for both medical staff and patients, strictly adheres to relevant standards for medical data privacy protection, desensitizes patient identity information, retains only necessary clinical identification information, and sets up data access audit logs to comprehensively record data access, query, export, modification, and other operations to prevent the risks of data leakage, misuse, and tampering.

[0051] This invention also includes a multi-center data sharing and model iteration module. This module adopts standardized data formats and interface protocols to support the interconnection and secure sharing of platform data among different medical institutions. It establishes a multi-center joint research database, performs data aggregation and analysis after removing patient privacy information, and uses federated learning algorithms to train the intelligent emotion analysis model across institutions by combining multi-center data without sharing original sensitive data. It integrates clinical experience and data resources from different regions and different levels of diagnosis and treatment, continuously improving the model's generalization ability, clinical applicability, and robustness, and providing data support and technical guarantee for the standardization, precision, and individualization of transcranial focused ultrasound treatment for depression.

[0052] The following two examples further illustrate the specific implementation of this system:

[0053] Example 1: Application of treatment scenario for moderate depression patients in the psychiatric department of a general hospital

[0054] This embodiment was applied in the psychiatry department of a tertiary general hospital, targeting a 35-year-old patient with moderate depression, a two-year history of illness, and no other underlying diseases. Transcranial focused ultrasound therapy combined with an emotion analysis platform was used for the entire intervention, with a treatment cycle of 8 weeks, 3 treatments per week, and each treatment lasting 40 minutes. The specific operation is as follows:

[0055] I. Execution of Core Processes and Key Steps

[0056] System initialization and parameter configuration: Medical staff create patient files and enter basic information such as age, disease course, and severity of illness through the professional workstation touchscreen of the human-computer interaction module. The system loads the pre-set transcranial focused ultrasound treatment plan, sets the treatment target to the left dorsolateral prefrontal cortex, the initial ultrasound intensity to 1.2 W / cm², the pulse repetition frequency to 1 MHz, and sets the emotion warning threshold to trigger a medium-risk warning if the overall emotion score is below 40 and a high-risk warning if it is below 30. The system completes communication connections and self-checks with the data acquisition module, treatment parameter synchronization module, and core processing module to ensure normal sensor response and stable data transmission.

[0057] Multimodal data acquisition and treatment parameter synchronization: The data acquisition module is installed according to a preset layout. The ECG sensor is attached to the patient's chest, the EEG acquisition electrodes are fixed to the corresponding area of ​​the scalp according to the international 10-20 system layout, the skin conductance sensor is worn on the tip of the patient's index finger, a high-definition facial image camera is installed 1.5 meters in front of the treatment room, and a high-fidelity voice microphone is placed 0.5 meters in front of the patient. After treatment begins, heart rate variability and EEG are collected simultaneously. Physiological signals such as frequency band energy and skin conductivity fluctuations, behavioral data such as displacement of 68 key facial feature points, speech fundamental frequency, and speech rate, are collected. Patients submit subjective emotional feedback every 20 minutes using a digital self-assessment scale. The treatment parameter synchronization module acquires parameters such as treatment target coordinates, ultrasound intensity, and pulse repetition frequency in real time via industrial Ethernet and 5G dual links, adding millisecond-level timestamps to the collected data to achieve precise spatiotemporal correlation.

[0058] Emotion Feature Extraction and Intelligent Analysis: The emotion feature extraction module preprocesses the raw data, using Kalman filtering to remove motion artifacts caused by slight limb movements of the patient. It employs a CNN algorithm to extract dynamic features from facial expressions, such as the degree of drooping of the corners of the mouth and the extent of brow furrowing. Voice emotion recognition technology analyzes parameters such as the degree of tone low, energy fluctuation range, and pause frequency. The self-assessment scale data is dimensionally split and converted into standardized scores for three categories: depressed mood, decreased interest, and anhedonia. The intelligent emotion analysis module loads a fusion algorithm model, inputting multi-dimensional feature vectors and synchronous treatment parameters. The model then generates a quantitative emotion score through computation.

[0059] Emotional assessment and treatment parameter optimization: The precise emotional state assessment module uses formulas... Calculate the real-time comprehensive sentiment score, where =0.3, =45 points =0.25, =42 points =0.25, =38 points =0.2, =40 points, substituting into the equation gives... =0.3×45+0.25×42+0.25×38+0.2×40=13.5+10.5+9.5+8=41.5 points. After 3 weeks of treatment, the patient's comprehensive emotional score improved to 58 points. The treatment response prediction and parameter optimization module used the formula... Calculate the intensity adjustment for ultrasound therapy, where =0.001, =58-41.5=16.5, =40 minutes =1MHz, substituting, we get =0.001×16.5×40×1=0.66W / cm², generating an optimized suggestion to adjust the ultrasound intensity to 1.4W / cm². After review by medical staff, the suggestion is sent to the transcranial focused ultrasound treatment device through the feedback control module.

[0060] Feedback Regulation and Multidisciplinary Collaboration: The feedback regulation module pushes emotion assessment reports to healthcare professionals, indicating significant improvement in patient mood and suggesting maintaining the current treatment frequency while adjusting its intensity. It also pushes breathing relaxation training audio and positive psychological suggestion tutorials (text and images) to patients' mobile devices twice daily, for 15 minutes each time. The healthcare professional collaboration module assigns permissions to attending physicians, nurses, and psychotherapists. Attending physicians view complete analysis data and sign off on parameter adjustments; nurses receive alerts and assist patients in completing self-assessment scales; and psychotherapists develop weekly cognitive behavioral therapy plans based on details of emotional fluctuations. All operation logs are recorded in real time and are tamper-proof.

[0061] Data storage and model self-learning: The data storage and management module categorizes and stores raw collected data, preprocessed feature data, and emotion analysis results, supporting queries by treatment stage and emotion level. The model self-learning optimization module collects the patient's treatment data and effect feedback, combines it with a multi-center labeled dataset, and optimizes the model's feature extraction weights through incremental learning algorithms to generate a personalized analysis sub-model for middle-aged patients with moderate depression. After 8 weeks of treatment, the system automatically generates a digital report containing the entire emotion change curve, treatment parameter records, and intervention effect evaluation, completing "one record, one file" storage.

[0062] II. Data Representation and Interpretation

[0063] Table 1 Comparison of Emotion Analysis and Treatment Intervention Effects in General Hospital Scenarios Table 1 Comparison of Emotion Analysis and Treatment Intervention Effects in General Hospital Scenarios

[0064] Performance indicators Traditional treatment methods Platform method of the present invention Emotion assessment accuracy 65% 92% Treatment parameter adjustment adaptability 58% 89% Intervention response time 24 hours 15 minutes Patient mood improvement target achievement rate 60% 88% Data traceability and integrity 55% 100%

[0065] Table 1 clearly demonstrates the advantages of this invention in a general hospital setting. Traditional treatment methods rely on subjective observation and periodic self-assessment by medical staff, with an accuracy rate of only 65% ​​in emotion assessment. Treatment parameter adjustments lack data support and have low adaptability, resulting in a 24-hour delay in intervention response and fragmented, difficult-to-trace data records. This invention, through multimodal data fusion and intelligent analysis, improves assessment accuracy to 92%, treatment parameter adjustment adaptability to 89%, provides a rapid 15-minute response to intervention needs, and ensures 100% data traceability throughout the process. The 88% achievement rate of emotion improvement is significantly higher than traditional methods, fully demonstrating that the platform can accurately capture emotional changes, scientifically guide treatment adjustments, and improve the treatment effect and clinical management efficiency for patients with moderate depression.

[0066] Example 2: Application of remote follow-up scenario for patients with mild depression in community healthcare

[0067] This embodiment was applied in a community health service center for a 28-year-old patient with mild depression, whose condition had lasted for 6 months and who was unable to frequently visit a tertiary hospital due to work reasons. The patient underwent transcranial focused ultrasound treatment combined with a remote emotion analysis platform for follow-up. The treatment cycle was 6 weeks, with treatment twice a week, each session lasting 30 minutes. The specific operation is as follows:

[0068] I. Execution of Core Processes and Key Steps

[0069] System initialization and remote configuration: Community healthcare staff create patient files through the human-computer interaction module, inputting basic information and then obtaining standardized treatment and analysis templates from tertiary hospitals through the multi-center data sharing and model iteration module. The treatment target is set as the bilateral dorsolateral prefrontal cortex, with an initial ultrasound intensity of 0.9 W / cm² and a pulse repetition frequency of 0.8 MHz. The emotional warning threshold is set as follows: a comprehensive emotional score below 50 triggers a low-risk warning, and below 40 triggers a medium-risk warning. An encrypted connection is established between the data acquisition module and the cloud server, with patient account security ensured through dual verification of biometrics and dynamic passwords.

[0070] Multimodal Data Acquisition and Remote Synchronization: The data acquisition module features a portable design, with patients wearing portable ECG monitoring devices, scalp EEG patches, and finger electrodermal sensors. The community treatment room is equipped with a high-definition facial camera and microphone. Physiological signals and behavioral data are collected simultaneously during treatment. Patients submit subjective feedback in real-time via a mobile self-assessment scale, and the data is transmitted end-to-end with encryption to the core processing module. The treatment parameter synchronization module uses 5G ultra-wideband communication to acquire treatment parameters in real-time and synchronize them to the remote tertiary hospital's medical staff, adding timestamps to the data for correlation.

[0071] Emotion Feature Extraction and Cross-Institutional Analysis: The emotion feature extraction module preprocesses the collected data, removing motion artifacts and environmental noise, and extracts specific EEG features, facial expression dynamic features, and vocal emotion features. The intelligent emotion analysis module loads a generalized model optimized through federated learning, inputs feature vectors and treatment parameters, and generates a quantitative emotion score. The remote monitoring and anomaly warning module monitors the data in real time. Two weeks after treatment, the patient's emotion score dropped to 42 points due to work stress, triggering a medium-risk warning. This immediately sends audible and visual alarms and pop-up notifications to community healthcare providers and attending physicians at tertiary hospitals.

[0072] Emotion Assessment and Parameter Optimization: The accurate emotion state assessment module calculates a comprehensive emotion score using a formula, whereby... =0.3, =52 points, =0.25, =48 points =0.25, =39 points, =0.2, =43 points, substituting into the equation gives... =0.3×52+0.25×48+0.25×39+0.2×43=15.6+12+9.75+8.6=45.95 points. The treatment response prediction and parameter optimization module calculates the adjustment amount using a formula. =0.001, =45.95-53=-7.05, =30 minutes =0.8MHz, substituting, we get =0.001×(-7.05)×30×0.8=-0.1692W / cm², generating a suggestion to adjust the ultrasound intensity to 0.8W / cm² and increase treatment frequency to 1 time per week. The remote attending physician reviews and issues the intervention instruction.

[0073] Feedback Regulation and Patient Self-Management: The feedback regulation module sends parameter adjustment instructions to community treatment devices, pushes stress management suggestions, emotion diary templates, and guided mindfulness meditation audio to patients, and reminds patients to complete 15 minutes of emotion regulation training daily. The patient emotion regulation guidance module records the duration of patient program use, completion rate, and feedback scores, providing data for subsequent optimization. The model self-learning optimization module iteratively optimizes personalized sub-models based on patient data, improving assessment accuracy. After 6 weeks of treatment, the system generates a follow-up report, synchronized to community and tertiary hospital databases, supporting cross-institutional data traceability.

[0074] II. Data Representation and Interpretation

[0075] Table 2 Comparison of Application Effects of Remote Follow-up Scenarios in Community Healthcare

[0076] Performance indicators Traditional remote follow-up methods Platform method of the present invention Remote synchronization of emotion assessment 40% 95% Timeliness of treatment parameter adjustment 36 hours 2 hours Patient treatment compliance 55% 86% Mood Improvement Stability Rate 50% 83% Data privacy protection compliance rate 60% 100%

[0077] Table 2 highlights the advantages of this invention in community-based remote follow-up scenarios. Traditional remote follow-up relies on telephone communication and in-person visits, with only 40% synchronization of emotional assessments, a 36-hour lag in parameter adjustments, and low patient compliance. This invention achieves 95% remote data synchronization through 5G ultra-wideband communication, completes treatment parameter adjustments within 2 hours, and improves compliance to 86% with personalized emotional regulation plans. A 100% data privacy protection rate ensures the security of patients' sensitive information, and an 83% emotional improvement and stabilization rate demonstrates that the platform can effectively support the treatment and follow-up of patients with mild depression in the community, addressing the pain points of poor cross-institutional diagnosis and treatment connections and insufficient precision in remote intervention, thus meeting the actual needs of community healthcare.

[0078] Reference Figure 2This graph visually illustrates the dynamic trends of patient emotional scores under different assessment methods. Traditional assessment methods rely on weekly patient self-assessments and subjective judgment by medical staff. Data updates are lagging and they fail to capture emotional fluctuations during treatment, resulting in slow and fluctuating score improvements that fail to reflect the actual effectiveness of treatment parameter adjustments. This invention's platform, through real-time multimodal data acquisition and intelligent analysis, accurately captures subtle emotional changes in patients each cycle. Combined with dynamic optimization of transcranial focused ultrasound treatment parameters, it allows emotional scores to show a steady upward trend. Starting from week 2, the score improvement is significantly higher than with traditional methods, exceeding 60 points by week 5. This fully demonstrates the platform's ability to provide timely feedback on treatment effectiveness, offering real-time and objective data support for medical staff to adjust treatment plans and driving treatment towards greater efficiency.

[0079] Reference Figure 3 This diagram clearly demonstrates the core advantages of the platform's emotion analysis module design. Traditional methods can only analyze single-dimensional data, with an accuracy rate of less than 70% for physiological signals and behavioral data. Furthermore, they lack the ability to integrate multimodal data and correlate it with treatment parameters, failing to integrate the complementary value of multi-source data and leading to biased assessment results. The platform of this invention optimizes the feature extraction algorithm for single modules, improving the accuracy of analysis across physiological, behavioral, and scale dimensions. More importantly, it adds a multimodal fusion analysis module, integrating data from various dimensions through an attention mechanism. Simultaneously, it establishes a mapping relationship between emotion and treatment by combining this with a treatment parameter correlation analysis module, achieving a fusion analysis accuracy rate of 94%. This significantly improves the comprehensiveness and precision of emotion assessment, providing a more reliable basis for adjusting treatment plans.

[0080] Reference Figure 4 This diagram clearly illustrates the feature weight allocation logic of the multimodal data fusion platform of this invention. In emotion analysis, the platform does not evenly distribute the weights of each data feature, but dynamically adjusts them based on the training results of clinical data. Facial behavioral features contribute 28%, and physiological signal features contribute 25%. These two types of objective data directly reflect the patient's emotional physiological responses and external behavioral expressions, serving as the core basis for emotion assessment. Vocal behavioral features contribute 20%, supplementing the information on emotional expression at the linguistic level. While subjective scale features and treatment parameter-related features contribute relatively less, they improve the dimensions of emotion assessment, especially the 15% contribution of treatment parameter-related features, which establishes a link between emotion and treatment. This weight allocation method allows the platform to focus on key features and reduce noise interference, further improving the accuracy and relevance of emotion analysis.

[0081] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A mood analysis platform for patients with depression based on transcranial focused ultrasound therapy, characterized in that, Includes the following modules: The data acquisition module is equipped with a high-precision electrocardiogram sensor, electroencephalogram acquisition electrodes, skin conductance sensor, high-definition facial image camera, high-fidelity voice microphone and digital self-assessment scale interaction unit. It simultaneously collects the patient's physiological signals, behavioral data and subjective emotional feedback during treatment. After acquisition, signal amplification and filtering noise reduction, the data is transmitted to the core processing module. The treatment parameter synchronization module adopts a dual-link communication protocol of industrial Ethernet and 5G to establish a stable connection with the transcranial focused ultrasound treatment equipment and acquire core parameters in real time. The emotion feature extraction module performs systematic preprocessing on the raw data and integrates multi-source data to extract high-dimensional structured emotion feature vectors. The intelligent emotion analysis module is equipped with a fusion algorithm model of random forest, convolutional neural network and long short-term memory network. It takes emotion feature vector and synchronous treatment parameters as input, dynamically evaluates the patient's condition, and generates quantitative emotion scores and qualitative analysis reports. The feedback and control module receives intelligent analysis results, sends parameter adjustment suggestions to the treatment equipment, pushes assessment reports and intervention prompts to the medical staff, and sends emotion regulation guidance to the patient. The data storage and management module adopts a dual architecture of local encrypted storage and cloud distributed server, which stores various types of data in categories and supports multi-dimensional query, export and traceability; The human-computer interaction module is configured with an interactive interface between a professional workstation for medical staff and a mobile terminal for patients. It displays relevant data and information in real time and supports adjustments by medical staff, data submission by patients, and viewing of guidance content.

2. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes an emotion state assessment module, which integrates multi-dimensional features output by the emotion feature extraction module and constructs a comprehensive assessment model through a multi-factor weighted fusion algorithm, using a formula... Calculate the patient's real-time comprehensive emotional score, among which Assess the overall emotional state. These are the weighting coefficients for physiological signal characteristics. Standardized scores for physiological signals, These are the weighting coefficients for behavioral data features. Standardize the scores for behavioral data. These are the feature weighting coefficients of the subjective scale. Standardized scores for subjective scales. For treatment response feature weighting coefficients, Standardized score for treatment response.

3. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a treatment response prediction and parameter optimization module. This module, based on historical assessment data and treatment parameter records from the intelligent emotion analysis module, predicts the patient's subsequent treatment response effect through a time-series trend analysis algorithm and a formula. Calculate the optimal adjustment value for ultrasound therapy intensity, where This is the intensity adjustment amount for ultrasound therapy. This is the treatment response sensitivity coefficient. The difference in the overall emotional score between two adjacent treatment cycles is... , This refers to the duration of the current treatment cycle. Based on the current treatment pulse repetition frequency, personalized treatment parameter optimization suggestions are generated according to the adjustment amount.

4. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a remote monitoring and anomaly early warning module. This module uses 5G ultra-wideband communication technology to complete real-time data transmission between medical staff and patients in the treatment scenario. It sets three levels of early warning thresholds for emotional fluctuations, physiological abnormalities, and treatment response. When it detects that the patient's emotional score fluctuation exceeds the set range, abnormal peaks occur in physiological signals, or the emotional score improvement is insufficient for two consecutive treatment cycles, the level of early warning mechanism is immediately triggered. It simultaneously records the time of the abnormality, relevant data, and early warning level, supporting medical staff to remotely view real-time data, issue emergency intervention instructions, and adjust treatment plans.

5. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a model self-learning optimization module, which continuously collects multi-center clinical treatment data, emotion analysis results, and treatment effect feedback to build a large-scale labeled dataset. The module iteratively trains the fusion algorithm model of the intelligent emotion analysis module through incremental learning algorithms, dynamically optimizing the model's feature extraction weights, classification decision boundaries, and evaluation parameters. At the same time, it generates personalized analysis sub-models based on differences in patient age, disease course, disease severity, treatment targets, and basic health conditions.

6. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a multimodal data fusion enhancement module, which uses an attention mechanism and feature-level fusion algorithm to deeply fuse four types of data: physiological signals, behavioral data, subjective scale data, and treatment parameters. It dynamically allocates feature weights to highlight information that is strongly correlated with emotional state, extracts specific EEG features from the region associated with EEG signals and ultrasound treatment targets, and combines facial expressions and voice intonation for emotional consistency verification to generate a more recognizable high-dimensional emotional feature vector.

7. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a medical-nursing collaborative interaction module, which supports hierarchical permission management for multidisciplinary medical teams, defining the operation permissions and data access scope of different roles. Attending physicians can view complete emotion analysis data, make suggestions to adjust treatment parameters, and sign intervention plans. Nurses can receive early warning information and record clinical observation results. Rehabilitation therapists can obtain emotion improvement trends and develop personalized emotion regulation training plans. Psychotherapists can view details of emotion fluctuations. The operation logs of all roles are recorded in real time and cannot be tampered with. It supports online communication and opinion collaboration within the team, generating a unified patient treatment and emotion management plan.

8. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a patient emotion regulation guidance module. Based on the assessment results of the intelligent emotion analysis module and the treatment stage, this module automatically matches personalized emotion regulation plans. For low mood, it pushes breathing relaxation training, mindfulness meditation guided audio, and positive psychological suggestion graphic tutorials. For mood fluctuation, it provides cognitive behavioral regulation techniques, stress management suggestions, and emotion diary templates. For good treatment response, it pushes rehabilitation progress incentive content, interest activity recommendations, and social interaction guidance. The plan push is reasonably arranged in combination with the patient's daily routine and treatment time, and the patient's regulation plan is recorded at the same time.

9. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a data security and privacy protection module, which uses end-to-end encryption algorithms to encrypt the transmission and storage of sensitive information throughout the process, uses blockchain technology to trace and prevent tampering of key medical data, adopts a dual verification mechanism of biometrics and dynamic passwords, strictly follows relevant standards for medical data privacy protection, desensitizes patient identity information, sets up data access audit logs, and comprehensively records data operations.

10. The mood analysis platform for patients with depression based on transcranial focused ultrasound therapy according to claim 1, characterized in that, It also includes a multi-center data sharing and model iteration module. This module adopts standardized data formats and interface protocols to support the interconnection and secure sharing of platform data from different medical institutions. It establishes a multi-center joint research database, performs data aggregation and analysis after removing patient privacy information, and uses federated learning algorithms to train the intelligent emotion analysis model across institutions by combining multi-center data without sharing the original sensitive data. It integrates clinical experience and data resources from different regions and different levels of diagnosis and treatment.

Images