Chemotherapy sign and emotion entropy coupled psychological elasticity dynamic portraying method

Abstract

The invention relates to the technical field of medical information processing, and particularly discloses a chemotherapy sign and emotional entropy coupled psychological elasticity dynamic portraying method, which comprises the following steps of: acquiring continuous time sequence sign data and emotional state data of a chemotherapy patient, calculating an emotional entropy change rate, and constructing an individualized emotion-physiological coupling kinetic model; the psychological state of the patient is predicted in real time; a psychological toughness probe microexperiment mechanism is introduced, when an emotion-physiological resonance state is recognized, the psychological regulation ability of a patient is quantitatively evaluated, and an instant elastic recovery spectrum is generated; a dynamic psychological elasticity portrait is generated by combining the resonance state recognition result and the instant elasticity recovery spectrum, and personalized nursing suggestions are provided; meanwhile, model parameters are optimized through feedback signals, and long-term accuracy and effectiveness of the system are ensured; according to the method, dynamic, accurate and personalized monitoring and support on the psychological elasticity of the patient are realized, and the initiative and timeliness of psychological intervention are improved.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and more specifically, to a method for creating a dynamic psychological profile by coupling chemotherapy symptoms with emotional entropy values. Background Technology

[0002] Chemotherapy is a crucial treatment for malignant tumors, but while killing tumor cells, it often brings a series of physiological toxic side effects and causes significant psychological stress for patients. Psychological resilience, the ability of an individual to adapt well to adversity, trauma, tragedy, and other major stressors, is a key factor influencing the quality of life and treatment adherence of chemotherapy patients. Therefore, effectively monitoring and supporting the psychological resilience of chemotherapy patients is an indispensable part of modern oncology care. Digital health technologies, particularly methods utilizing information and communication technologies for health data processing and diagnostic support, offer new possibilities for achieving this goal.

[0003] Currently, clinical monitoring of the psychological state of chemotherapy patients mainly relies on periodically completed psychological assessment scales, such as the Hospital Anxiety and Depression Scale (HADS) or the Essential Cancer Patient Quality of Life Scale (EORTC QLQ-C30). Some advanced digital health solutions attempt to collect physiological data such as heart rate and skin conductance using wearable devices, or collect patients' daily emotional self-assessments through mobile applications, and perform independent threshold alarms or trend analysis on these data in order to detect abnormal emotional fluctuations.

[0004] However, existing technologies have significant limitations. First, scale-based assessments are infrequent and highly subjective, making it difficult to capture instantaneous changes in emotions and resulting in significant monitoring delays. Second, separating physiological and psychological data ignores the complex interactions between mind and body, failing to reveal the deep coupling between emotional fluctuations and physiological stress responses. Furthermore, commonly used fixed or group-based threshold alarm mechanisms do not consider the significant individual differences among patients, leading to insufficient specificity and sensitivity in monitoring and a high risk of false alarms or missed alarms. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a psychological resilience dynamic profiling method coupling chemotherapy signs and emotional entropy values ​​is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: This invention provides a method for dynamic profiling of psychological resilience by coupling chemotherapy signs and emotional entropy values, including the following steps: S1, acquiring continuous temporal sign data of the target patient during chemotherapy, and simultaneously acquiring temporal emotional state data of the target patient, and calculating the rate of change of emotional entropy of the target patient based on the temporal emotional state data.

[0007] S2. Based on continuous time-series vital sign data and emotional entropy change rate data, an individualized emotion-physiological coupling dynamic model is constructed to characterize the dynamic relationship between physiological fluctuations and emotional stability in target patients.

[0008] S3. Input continuous time-series vital sign data into the individualized emotion-physiology coupled dynamics model in real time to generate emotion state prediction data, and compare the emotion state prediction data with the real-time acquired emotion entropy change rate data to obtain prediction deviation data.

[0009] S4. Analyze the time series characteristics of the prediction deviation data to identify whether there are co-amplification patterns related to physiological or therapeutic rhythms. When a co-amplification pattern is identified, generate resonance state identification results including resonance intensity index and instability proximity index.

[0010] S5. In response to the generation of resonance state recognition results, initiate a psychological resilience probe micro-experiment to the target patient through the terminal device to explore the effectiveness of specific psychological regulation pathways, and collect emotional state response data after the execution of the psychological resilience probe micro-experiment.

[0011] S6. Analyze emotional state response data to quantify the dynamic characteristics of the emotional recovery process and calculate the immediate elastic recovery spectrum, which includes efficacy scores of multiple psychological regulation pathways.

[0012] S7. By coupling the resonance state recognition results and the instantaneous elastic recovery spectrum, a dynamic psychological resilience profile of the target patient is generated.

[0013] S8. Using the resonance state identification results and instantaneous elastic recovery spectrum as feedback signals, the parameters of the individualized emotion-physiology coupling dynamics model are updated to obtain the optimized individualized emotion-physiology coupling dynamics model.

[0014] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention realizes predictive monitoring of patients' psychological state by constructing an individualized emotion-physiology coupled dynamic model. Compared with traditional periodic scale assessment, this method can identify potential psychological instability risks in advance based on the dynamic correlation of real-time physiological-emotional data, and change the timing of psychological intervention from post-event response to pre-event prevention, thereby improving the initiative and timeliness of psychological support.

[0015] This invention innovatively introduces a psychological resilience probe micro-experiment mechanism, organically combining the diagnostic and intervention processes. By triggering lightweight, goal-oriented interactive tasks at key time points, it not only provides immediate emotional regulation support but, more importantly, quantitatively assesses the patient's immediate performance in various psychological regulation abilities under stress. This provides precise and multidimensional decision-making basis for subsequent personalized interventions, achieving a closed loop of intervention as assessment.

[0016] This invention establishes an adaptive evolutionary intelligent system. By using the identification results of each resonance event and the elasticity recovery assessment results as feedback, the core dynamic model is continuously optimized, enabling the system to continuously learn and adapt to the patient's individualized, time-varying psychophysiological response patterns. This self-evolutionary capability ensures the accuracy and effectiveness of the system's long-term application, achieving a dynamic, accurate, and personalized profile of the patient's psychological resilience. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the method steps of the present invention. Detailed Implementation

[0019] 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.

[0020] Please see Figure 1 This invention provides a method for dynamic profiling of psychological resilience by coupling chemotherapy signs and emotional entropy values, including: S1, acquiring continuous temporal sign data of the target patient during chemotherapy, and simultaneously acquiring temporal emotional state data of the target patient, and calculating the rate of change of emotional entropy of the target patient based on the temporal emotional state data.

[0021] In a specific embodiment of the present invention, the specific steps for calculating the rate of change of the emotional entropy of the target patient based on time-series emotional state data include: periodically collecting the target patient's self-assessment data of emotional state or voice data through an embedded micro-interactive interface.

[0022] Emotional dimension analysis is performed on self-reported emotional state data or voice data to generate temporal emotional state vectors.

[0023] The probability distribution of temporal sentiment state vectors is calculated within a sliding time window, and the temporal sentiment entropy is calculated based on the probability distribution.

[0024] By performing time-series emotional entropy differentiation, the rate of change of emotional entropy of the target patients is obtained.

[0025] Specifically, the engineering implementation process for acquiring the time-series emotional state data of the target patient and calculating the rate of change of emotional entropy based on the time-series emotional state data is as follows. First, the engineering objective of this step is to continuously quantify the patient's emotional state in a low-interference manner and further calculate the index reflecting the rate of change of emotional instability or disorder, namely the rate of change of emotional entropy, to provide the core "emotional" dimension input for the subsequent construction of the dynamic model.

[0026] The first step involves periodically collecting self-assessment data or voice data of the target patient's emotional state through a pre-designed embedded micro-interactive interface. This embedded micro-interactive interface is integrated into the patient's terminal device, for example, popping up once a day at fixed times (e.g., 9 AM, 3 PM, and 8 PM), prompting the patient to choose one of several representative emotional words or record a short 10-15 second voice description of their feelings that day. This non-invasive micro-interactive design aims to obtain high-frequency emotional snapshots with minimal burden.

[0027] The second step involves performing emotional dimension analysis on the collected self-assessment data or speech data to obtain a temporal emotional state vector. If emotional words are selected, the system maps each emotional word to a coordinate point (valence, arousal) in a two-dimensional emotional space. Valence represents the positive or negative aspect of the emotion, and arousal represents the intensity of the emotion. If the data is speech data, an acoustic feature such as pitch and speech rate is extracted using a speech emotion recognition engine. Then, a pre-trained emotion classifier outputs the corresponding emotional valence and arousal values ​​for that speech segment. Through continuous data collection, a temporal emotional state vector is obtained. ,in yes Valence at any moment yes Awareness level at any given moment.

[0028] It should be noted that in the emotional word selection mode, valence and arousal are acquired through a preset mapping mechanism: the system first allows the patient to select from a set of representative emotional words through an embedded micro-interactive interface, and then the system directly maps each emotional word selected by the patient to a specific coordinate point in a two-dimensional emotional space; where valence, as one dimension of the coordinate, is represented by numerical quantification to indicate the positive or negative attribute corresponding to the emotional word (i.e., whether the emotion is positive and pleasant or negative and unpleasant), and arousal, as the other dimension of the coordinate, is represented by numerical quantification to indicate the intensity attribute corresponding to the emotional word (i.e., the degree of activation or energy level of the emotion), thereby transforming the discrete text labels subjectively selected by the patient into a continuous temporal emotional state vector that can be processed by the computer.

[0029] The third step involves calculating the information entropy difference between adjacent time points in the temporal sentiment state vector, and then performing a time derivative on this information entropy difference to obtain the final sentiment entropy change rate data. In practice, the two-dimensional sentiment space is first divided into an M×N discrete grid. Then, the temporal sentiment state vector is statistically analyzed over a past sliding time window (e.g., the past 24 hours). The probability of falling into each grid cell Based on this probability distribution, the sentiment entropy of the current time window is calculated using the information entropy formula. : The summation iterates through all grid cells. (Emotional entropy) The higher the value, the wider and more disordered the patient's emotional distribution within that time window. The system continuously calculates the emotional entropy for each sliding window, forming a time series of emotional entropy. Finally, the emotional entropy sequence By performing first-order difference or numerical differentiation, the rate of change of emotional entropy can be obtained. This data directly reflects the instantaneous rate at which the patient's level of emotional distress changes over time.

[0030] S2. Based on continuous time-series vital sign data and emotional entropy change rate data, an individualized emotion-physiological coupling dynamic model is constructed to characterize the dynamic relationship between physiological fluctuations and emotional stability in target patients.

[0031] In a specific embodiment of the present invention, the specific steps for constructing an individualized emotion-physiological coupling dynamic model to characterize the dynamic relationship between physiological fluctuations and emotional stability of a target patient include: processing continuous time-series vital sign data into a system input vector and processing emotional entropy change rate data into a system output response.

[0032] A system identification method is used to fit the dynamic relationship between the system input vector and the system output response, generating a set of preliminary coupled model parameters.

[0033] The parameters of the initial coupling model are validated and calibrated using historical data windows. When the prediction error of the model is lower than the error threshold, the parameters of the initial coupling model are confirmed, and an individualized emotion-physiology coupling dynamic model is formed.

[0034] Specifically, the engineering implementation process for constructing an individualized emotion-physiology coupled dynamic model for a target patient, based on continuous time-series vital sign data and emotional entropy change rate data, is as follows. First, the engineering objective of this step is to establish a dynamic model capable of digitally describing the unique correlation between a specific patient's physiological fluctuations and changes in emotional stability. This model is engineering-wise considered a single-input single-output or multi-input single-output system. To achieve this goal, the following operations are required.

[0035] The first step is to process continuously acquired time-series vital sign data into system input vectors. These time-series vital sign data, such as the frequency domain indices of heart rate variability (HRV) or the peak rate of change of skin conductance (SDA), are standardized and constructed into a vector that varies with time. Changing vector Simultaneously, the emotional entropy change rate data obtained through synchronous calculation is subjected to the same time alignment and smoothing filtering processing and used as the system's output response. For example, the system input vector It can include multiple physiological indicators such as low-frequency heart rate variability (LF) and high-frequency heart rate variability (HF), and the system output response... This is a single value representing the rate of change of emotional entropy.

[0036] The second step is to use a system identification method to analyze the system input vector. and system output response The relationship between inputs and outputs is fitted to generate preliminary coupled model parameters. In engineering practice, an autoregressive moving average (ARMAX) model can be used for fitting. This model establishes the transitive relationship between inputs and outputs using the following mathematical formula: ,in, yes The system output response at any given moment, i.e., the rate of change of emotional entropy; yes The system input vector at time step, i.e., time-series vital sign data; These are random perturbations that the model cannot explain, representing white noise; It is the input delay order of the system, which is usually set in the range of 1-5 sampling periods; It is a delay operator; , and This is a polynomial concerning the delay operator, whose coefficients constitute the initial coupled model parameters that need to be identified. Using the initial training dataset, the formula that minimizes the sum of squared prediction errors is automatically solved using the least squares or gradient descent method. , and The coefficient value.

[0037] The third step is to validate and calibrate the generated preliminary coupled model parameters. The engineering purpose of this step is to test the model's generalization ability and fine-tune it to ensure its predictive accuracy on new data. A pre-defined historical data window is defined from the patient's historical data as a validation set; data within this window is not used in model training. The system input vector from the validation set is then input into the established preliminary coupled model to generate the model's predicted output. Then, the predicted output... Compared with the actual system output response in the validation set The model parameters are compared, and the root mean square error (RMSE) or mean absolute percentage error (MAPE) is calculated. If the error metric is below a preset engineering threshold (e.g., RMSE less than 0.15), the model parameters are considered valid, forming the final individualized emotion-physiology coupled dynamics model. If the error metric exceeds the preset engineering threshold, the model order is adjusted or the amount of training data is increased, and the process returns to step two to re-identify the model parameters until the engineering accuracy requirements are met.

[0038] S3. Input continuous time-series vital sign data into the individualized emotion-physiology coupled dynamics model in real time to generate emotion state prediction data, and compare the emotion state prediction data with the real-time acquired emotion entropy change rate data to obtain prediction deviation data.

[0039] In a specific embodiment of the present invention, the specific steps for generating emotional state prediction data and comparing the emotional state prediction data with the real-time acquired emotional entropy change rate data to obtain prediction deviation data include: constructing a real-time data buffer pool to store continuous time-series vital sign data and historical emotional entropy change rate data for the most recent N sampling periods, and performing time alignment and resampling processing on heterogeneous data sources.

[0040] Based on an autoregressive structure of an individualized emotion-physiology coupled dynamics model, recursive prediction operations are performed using data from a real-time data buffer pool to calculate the predicted value of the emotion state at the current moment.

[0041] The instantaneous difference between the predicted emotional state value and the real-time acquired emotional entropy change rate data is calculated, and the instantaneous difference is smoothed using the exponential moving average algorithm to generate prediction deviation data.

[0042] Specifically, the first step is to build and maintain a real-time data buffer pool to synchronize and preprocess heterogeneous data. Because vital sign data (such as heart rate variability (HRV), typically measured in seconds or milliseconds) and emotional state data (such as the rate of change of emotional entropy, typically calculated based on minute-level windows) have different sampling frequencies, the system first needs to allocate a first-in-first-out (FIFO) circular buffer in memory. This buffer is configured to store a historical data sequence of length N (N is typically 2-3 times the model order, e.g., 10-20 points). In engineering implementation, the system uses the emotional entropy calculation cycle (e.g., per minute) as the baseline beat. Whenever a new emotional entropy change rate data point is generated... During generation, the system extracts statistical features (such as the mean of the LF / HF ratio) within the corresponding time window from the high-frequency vital sign data stream to form the current system input vector. Then, and Push the data into the buffer pool and discard the oldest data to ensure that the data at the model input is strictly aligned on the time axis.

[0043] The second step involves performing recursive prediction operations based on the model structure to generate emotional state prediction data. This utilizes the individualized emotion-physiology coupled dynamic model identified in step S2. The system performs single-step forward prediction. At time t... The system calls historical input data from the buffer pool. and historical output data Substitute into the model equation: ,in, and These are the fixed model parameters. Through this recursive operation, the theoretical rate of change of emotional entropy at the current moment is calculated, i.e., the emotional state prediction data. This value represents the rate of emotional fluctuation that a patient is expected to exhibit based on their current level of physiological arousal, under normal physiological-psychological coupling mechanisms.

[0044] The third step is to calculate the instantaneous difference and perform smoothing to generate the final prediction bias data. First, calculate the instantaneous residual. The formula is: ,in, This refers to the real-time monitored rate of change in actual emotional entropy. Since the original instantaneous residuals may contain measurement noise or various instantaneous artifacts, direct use could lead to false triggering of the system. Therefore, in engineering practice, an exponential moving average algorithm is used to filter the residual sequence online and calculate the final prediction bias data. : ,in, Smoothing coefficient (value range 0 < <1 (e.g., 0.3) is used to balance sensitivity to new data and robustness to noise. The final output sequence This is the prediction bias data, which eliminates high-frequency noise and stably represents the degree to which a patient's current emotional state deviates from their individualized baseline model. When When the value is close to 0, it indicates that the patient is in a steady state; when A significant increase suggests that an abnormal coupling amplification mechanism may have occurred within the system.

[0045] S4. Analyze the time series characteristics of the prediction deviation data to identify whether there are co-amplification patterns related to physiological or therapeutic rhythms. When a co-amplification pattern is identified, generate resonance state identification results including resonance intensity index and instability proximity index.

[0046] In a specific embodiment of the present invention, the specific steps of analyzing the time series characteristics of the prediction deviation data to identify whether there is a synergistic amplification mode related to physiological or therapeutic rhythms include: performing spectral analysis on the prediction deviation data to obtain the deviation spectral distribution.

[0047] Physiological rhythm signals are extracted from continuous temporal vital sign data or external treatment cycle data are obtained to form rhythm reference signals.

[0048] The coherence coefficient between the deviation spectral distribution and the spectrum of the rhythm reference signal is calculated. When the coherence coefficient exceeds the coherence threshold in a specific frequency band and the amplitude of the predicted deviation data shows a monotonically increasing trend with an increase rate exceeding the preset increase rate threshold, the existence of the cooperative amplification mode is confirmed.

[0049] Specifically, the engineering process for identifying target patients entering an emotional-physiological resonance state involves matching the time-series characteristics of prediction bias data with preset rhythmic patterns. First, the engineering objective of this step is to further determine whether significant deviations in the individualized emotional-physiological coupling dynamics model prediction are caused by abnormal coupling amplification of physiological rhythms and emotional fluctuations, thereby accurately identifying potential precursors to psychological instability, i.e., the emotional-physiological resonance state.

[0050] The first step is to perform spectral analysis on the real-time generated prediction bias data to extract its spectral features and obtain the bias spectral distribution. The prediction bias data is obtained by subtracting the model's predicted sentiment state data from the actual acquired sentiment entropy change rate data at each sampling point. For a time window, such as the prediction bias data sequence over the past 3 to 5 minutes, a Fast Fourier Transform (FFT) is applied to transform it from the time domain to the frequency domain, yielding the bias spectral distribution. ,in For frequency. The spectral distribution of this deviation. The main frequency components of energy concentration in the prediction error were revealed.

[0051] The second step involves simultaneously acquiring real-time physiological rhythm data or external treatment cycle data from the target patient and processing it into a rhythm reference signal. Real-time physiological rhythm data can be extracted from continuous time-series vital sign data, such as energy fluctuations in specific frequency bands of the power spectrum of heart rate signals, typically focusing on the very low frequency (VLF) band (0.003-0.04 Hz), the low frequency band (0.04-0.15 Hz), and the high frequency band (0.15-0.4 Hz). External treatment cycle data, such as the periodic pulses of chemotherapy drug infusions, can be preset according to the treatment plan. These signals are also subjected to spectral analysis to obtain the spectrum of the rhythm reference signal. .

[0052] The third step is to distribute the deviation spectrum. Spectrum of rhythm reference signal Coherence analysis is performed. In engineering, the squared amplitude coherence function is used. To quantize two signals at a specific frequency The degree of linear correlation is calculated using the following formula: ,in, It is the cross-power spectral density of the prediction deviation data and the rhythm reference signal. and These are their self-power spectral densities. The value ranges from 0 to 1; the closer the value is to 1, the stronger the correlation at that frequency. When the coherence coefficient is calculated within one or more key frequency bands, such as the low-frequency band... If the amplitude of the prediction deviation data continuously exceeds a preset coherence threshold, such as 0.7, and within this time window, the amplitude of the prediction deviation data, i.e. the moving average of the absolute value of its time domain signal, shows a monotonically increasing trend and the growth rate exceeds a preset growth rate threshold, the system determines that the prediction deviation data exhibits a co-amplification characteristic.

[0053] It should be noted that the preset growth rate threshold is typically set at 0.05 (i.e., 5%). The basis for this value is that the absolute value moving average of the prediction deviation is essentially a low-pass filter for high-frequency random noise. When the system is stable or only subject to occasional interference, this value should fluctuate around zero or maintain a constant low amplitude oscillation. When the growth rate exceeds 5%, from the perspective of statistical process control, it means that the accumulation rate of error energy has significantly exceeded the natural variance range of the background noise, showing a definite divergence trend. In engineering practice, this threshold is a classic empirical value that balances sensitivity and robustness. It can effectively avoid misjudgments caused by small sensor jitters and can trigger an alarm in time before the deviation is irreversibly amplified and causes the system to collapse.

[0054] The fourth step involves calculating the resonance intensity index and the instability proximity index based on the analysis results that satisfy the synergistic amplification characteristics. The resonance intensity index can be defined as the product of the weighted average of the coherence coefficients within the synergistic amplification frequency band and the amplitude growth rate, used to quantify the abnormal intensity of the current physiological-emotional coupling. The instability proximity index is calculated by linearly or exponentially extrapolating the current amplitude growth trend to predict the time required for the predicted deviation data to reach a preset unacceptable threshold; the shorter the time, the higher the proximity. These two indices together constitute the resonance state identification result, triggering subsequent intervention and evaluation processes.

[0055] S5. In response to the generation of resonance state recognition results, initiate a psychological resilience probe micro-experiment to the target patient through the terminal device to explore the effectiveness of specific psychological regulation pathways, and collect emotional state response data after the execution of the psychological resilience probe micro-experiment.

[0056] In a specific embodiment of the present invention, the specific steps of initiating a psychological resilience probe micro-experiment to a target patient via a terminal device to explore the effectiveness of a specific psychological regulation pathway include: constructing a probe task library containing multiple probe tasks, wherein each probe task is associated with one or more psychological regulation pathways.

[0057] Based on the characteristics of the resonance state identification results, target probe tasks are selected from the probe task library.

[0058] Send guidance instructions for the target probe task to the target patient's terminal device and monitor the target patient's task execution interaction data to ensure the effective completion of the task.

[0059] Specifically, the psychological resilience probe micro-experiment is a low-load guided task used to explore the effectiveness of specific psychological regulatory pathways. The engineering implementation process of initiating a pre-set psychological resilience probe micro-experiment with the target patient is as follows. First, the engineering objective of this step is to accurately and automatically initiate a lightweight interactive task after the system identifies the emotional-physiological resonance state. This lightweight interactive task not only plays a preliminary intervention role, but its core function is to serve as a "diagnostic tool" to actively measure the patient's immediate performance of specific psychological regulatory abilities under the current stress state.

[0060] The first step involves selecting targeted probe tasks from a pre-defined probe task library based on the features of the resonance state recognition results generated in the previous step. This library pre-constructs a series of micro-experiments, each strongly correlated with one or more specific psychological regulatory pathways. For example, the "breathing synchronization tracking task" is associated with the attention control pathway, and the "positive event recall task" is associated with the positive emotion reappraisal pathway. The matching logic is based on the intrinsic features of the resonance state recognition results. For instance, if the deviation spectrum distribution of the resonance state is mainly concentrated in the high-frequency band, it may indicate anxiety and hypervigilance; in this case, the system will prioritize probe tasks related to relaxation responses and somatic perception. The matching algorithm can be a rule-based decision tree or a simple lookup table mechanism, with the input being the resonance intensity index and the main resonance frequency, and the output being the ID of the optimal target probe task.

[0061] The second step involves sending guidance instructions for the selected target probe task to the patient's terminal, such as a smartphone or a dedicated bedside interactive device. These instructions include multimedia content, such as brief voice guidance, animated demonstrations, or text prompts, to clearly guide the patient through the task. Task duration is strictly controlled between 90 and 180 seconds to ensure low workload and minimal disruption to the normal treatment process. During task execution, the system monitors patient compliance data via the terminal's sensors, such as the touchscreen or microphone. For example, in a respiratory tracking task, the consistency between the user's swipe trajectory and the guidance animation; or in a voice task, the loudness and fluency of the voice, to determine whether the task has been effectively performed.

[0062] Third, the system records the entire execution process of the target probe task in the background, generating a structured probe experiment record. This record not only includes basic information such as the start and end times of the task and the task ID, but also crucially records detailed patient interaction data during task execution, as well as synchronously collected emotional state data, such as the emotional valence and arousal sequences output in real time through the voice emotion analysis module. This probe experiment record will serve as the core input data for the next step of real-time elastic recovery spectrum analysis, providing direct evidence for quantitatively assessing the patient's psychological adjustment ability.

[0063] S6. Analyze emotional state response data to quantify the dynamic characteristics of the emotional recovery process and calculate the immediate elastic recovery spectrum, which includes efficacy scores of multiple psychological regulation pathways.

[0064] In a specific embodiment of the present invention, the specific steps for calculating the immediate elastic recovery spectrum, which includes efficacy scores of multiple psychological regulation pathways, include: drawing an emotion recovery curve based on emotional state response data.

[0065] A function fit was performed on the emotion recovery curve to extract a set of morphological feature parameters that include the recovery time constant and the curve decay type.

[0066] Based on the type of psychological regulation pathway associated with the target probe task, the morphological feature parameters are mapped to the efficacy scores of the corresponding psychological regulation pathways, and all efficacy scores are aggregated into an instantaneous elastic recovery spectrum.

[0067] Specifically, the engineering process for calculating the immediate resilience recovery spectrum of the target patient by analyzing the curve shape and recovery pattern of the emotional state response data is as follows. First, the engineering purpose of this step is to quantitatively assess the process characteristics of the patient's emotional state recovering to the baseline level after completing the psychological resilience probe micro-experiment, thereby obtaining an immediate resilience recovery spectrum that can characterize the patient's current psychological adjustment ability in multiple dimensions.

[0068] The first step is to plot the emotion recovery curve based on the emotional state response data collected in the previous step. The emotional state response data is a time series, recording the continuous changes in the patient's emotional valence or emotional entropy over a period of time after the probe experiment, such as 3 to 5 minutes. Visualizing this time series data with time on the horizontal axis and emotion indicators on the vertical axis yields the emotion recovery curve. To reduce the influence of noise, the raw data is usually processed using a moving average or low-pass filter to obtain a smoothed emotion recovery curve.

[0069] The second step involves fitting a piecewise function to the smoothed emotion recovery curve to quantitatively extract its morphological parameters. In engineering, this emotion recovery curve is typically fitted to a first- or second-order decaying system response model. For example, an exponential decay function can be used for fitting. ,in, yes The sentiment index value at any given moment; It is the initial amplitude of the emotional fluctuation, determined by the difference between the emotional index value at the end of the probe experiment and the emotional baseline value; It is the time counted from the end of the probe experiment; It is the recovery time constant, reflecting the speed of emotional recovery. The smaller the size, the faster the recovery; This is the steady-state value of the sentiment index that eventually converges, representing the baseline sentiment after recovery. By fitting the curve using a nonlinear least squares method, the recovery time constant can be automatically calculated. .Apart from It may also extract other morphological parameters, such as overshoot, which is the maximum deviation of the emotional index from the steady-state value during the recovery process, and oscillation frequency, to describe the fluctuations during the recovery process. These parameters together constitute a quantitative description of the recovery pattern.

[0070] The third step involves mapping the multiple morphological feature parameters extracted in the previous step to the corresponding efficacy scores of the psychological regulation pathways, based on the known types of psychological regulation pathways associated with the target probe task. This is a many-to-many mapping process, implemented through preset scoring rules. For example, if the target probe task is associated with attention control pathways, a smaller recovery time constant is used. A higher attention shifting efficacy score may correspond to a higher overall efficiency score; conversely, a larger overshoot may indicate lower efficacy of the cognitive reappraisal pathway in suppressing negative emotional rebound. Each psychological regulation pathway, such as attention control, cognitive restructuring, and emotional suppression, receives an efficacy score ranging from 0 to 100 based on the combination of its most relevant morphological characteristic parameters. The set of efficacy scores calculated for different psychological regulation pathways ultimately constitutes a structured immediate resilience spectrum, providing core regulatory capacity dimensions as input for the subsequent generation of a dynamic psychological resilience profile.

[0071] S7. By coupling the resonance state recognition results and the instantaneous elastic recovery spectrum, a dynamic psychological resilience profile of the target patient is generated.

[0072] In a specific embodiment of the present invention, the specific steps for generating a dynamic psychological resilience profile of the target patient by coupling the resonance state recognition result and the instantaneous elastic recovery spectrum include: fusing the resonance intensity index and the instability proximity index to calculate the system vulnerability index.

[0073] The efficacy scores of each psychological regulation pathway were analyzed from the immediate resilience recovery spectrum to form a profile of regulation strategy efficacy.

[0074] By integrating the system vulnerability index and the effectiveness profile of the adjustment strategy with historical profile data, a dynamic psychological resilience profile is output in a visual form that includes trend curves and multi-dimensional charts.

[0075] Specifically, the engineering implementation process for generating a dynamic psychological resilience profile of the target patient based on the resonance state recognition results and the real-time elasticity recovery spectrum is as follows. First, the engineering objective of this step is to efficiently integrate and visualize the data from the previous steps, which assess the patient's current stress state and psychological adjustment ability, to form a view that can intuitively and dynamically reflect the patient's overall psychological resilience status.

[0076] The first step involves fusing the resonance intensity index and instability proximity index from the resonance state identification results to generate a single system vulnerability index. This system vulnerability index aims to quantify the vulnerability of the patient's current psychophysiological system. In engineering, due to the resonance intensity index... and instability proximity index Since their units and ranges differ, standardization is required first. The calculation is performed using a formula: ,in, It is a system vulnerability index, and its value usually ranges from 0 to 1; It is a dimensionless or composite dimensionless index of resonance intensity. It is an instability proximity index measured in units of time; This represents the min-max normalization function, which linearly maps the input data to the interval 0-1. and These are preset weighting coefficients that satisfy... The values ​​can be set based on clinical experience, for example, 0.6 and 0.4 respectively, to reflect the importance of different indicators. Here, we use... The reciprocal of the instability is because the shorter the time before instability, the higher the vulnerability, and the two are inversely proportional.

[0077] The second step involves analyzing the current efficacy scores of each psychological regulation pathway from the immediate resilience recovery spectrum to generate a regulation strategy efficacy profile. The immediate resilience recovery spectrum is a data structure containing efficacy scores for multiple psychological regulation pathways. The regulation strategy efficacy profile is an engineered representation of this data structure, typically implemented as a radar chart or bar chart. For example, efficacy scores for different pathways such as attention control, cognitive restructuring, and emotional suppression can be used as multiple dimensions of the radar chart, with the length of each dimension representing the current efficacy level of that pathway. This visually demonstrates the strengths and weaknesses of the patient's psychological regulation abilities.

[0078] The third step involves integrating the calculated system vulnerability index, the generated moderating strategy effectiveness profile, and historical profile data stored in the database through spatiotemporal correlation, ultimately forming and outputting a multi-dimensional, dynamic psychological resilience profile. In engineering implementation, this manifests as a user interface or data report. This profile typically comprises at least two core parts: a trend curve of the system vulnerability index over time, revealing the long-term evolution of the patient's psychological state; and a radar chart of the moderating strategy effectiveness profile at the current moment, showcasing immediate moderating capabilities. By displaying these two parts side-by-side and allowing users to review profile snapshots at any historical point in time, the system provides a comprehensive and dynamic view reflecting the patient's vulnerability and coping abilities under stress—the dynamic psychological resilience profile.

[0079] In a specific embodiment of the present invention, after generating a dynamic psychological resilience profile of the target patient, the method further includes: constructing a tiered intervention strategy library containing different intervention levels.

[0080] Based on the system vulnerability index value, recommended intervention strategies are matched and selected from the tiered intervention strategy library.

[0081] By combining recommended intervention strategies with dynamic psychological resilience profiles, a personalized nursing advice report is generated to assist healthcare professionals in making decisions and is then pushed to their terminals.

[0082] Specifically, after generating a dynamic psychological resilience profile, the engineering implementation process also includes generating personalized nursing advice reports to assist medical staff in decision-making, as follows. First, the engineering objective of this step is to transform the complex dynamic psychological resilience profile obtained from system analysis into actionable reference suggestions that have direct guiding significance for clinical medical staff, thereby achieving a closed loop from intelligent monitoring to clinical decision support.

[0083] The first step involves automatically matching and selecting recommended intervention strategies from a pre-defined tiered intervention strategy library based on the core system vulnerability index in the dynamic psychological resilience profile. This library is, in engineering terms, a structured database or rule set, pre-defined with multiple intervention levels, each corresponding to a range of system vulnerability index values. For example, when the system vulnerability index is below 0.3, a Level 1 strategy is matched, namely "routine observation and self-management resource delivery"; when the index is between 0.3 and 0.7, a Level 2 strategy is matched, namely "increasing the frequency of nurse attention and initiating online mindfulness practice"; and when the index is above 0.7, a Level 3 strategy is matched, namely "triggering a high-risk alert and recommending intervention by a mental health professional." During system execution, the newly generated system vulnerability index is used as input, and the corresponding recommended intervention strategy is directly output by querying the strategy library.

[0084] The second step involves data association and integration between the recommended intervention strategies selected in the previous step and the complete dynamic psychological resilience profile, automatically generating a personalized nursing recommendation report. This process involves data aggregation and formatted output. The system extracts key information from the dynamic psychological resilience profile, including the current system vulnerability index and its historical trend graph, and a radar chart showing the effectiveness of adjustment strategies highlighting strengths and weaknesses. This information is then combined with the selected recommended intervention strategy text. The report content is organized into a clearly structured document, including basic patient information, assessment time, risk level determination, intuitive graphical analysis results, and clear, priority-based nursing action recommendations. Finally, this personalized nursing recommendation report, which assists healthcare professionals in decision-making, is generated in electronic document format and can be automatically pushed to the electronic medical record system on healthcare professionals' workstations or mobile terminals as a basis for clinical intervention decision-making.

[0085] S8. Using the resonance state identification results and instantaneous elastic recovery spectrum as feedback signals, the parameters of the individualized emotion-physiology coupling dynamics model are updated to obtain the optimized individualized emotion-physiology coupling dynamics model.

[0086] In a specific embodiment of the present invention, the specific steps for updating the parameters of the individualized emotion-physiology coupling dynamics model to obtain the optimized individualized emotion-physiology coupling dynamics model include: marking the data segment corresponding to the resonance state recognition result as a reinforcement learning sample.

[0087] The performance scores in the instantaneous elastic recovery spectrum are weighted and fused to calculate the reward signal used as feedback for this event.

[0088] By employing reinforcement learning algorithms and utilizing reinforcement learning samples and reward signals, the parameters of the individualized emotion-physiology coupling dynamics model are iteratively adjusted to minimize prediction bias in similar future scenarios, thereby obtaining an optimized individualized emotion-physiology coupling dynamics model.

[0089] Specifically, the engineering implementation process for updating the parameters of the individualized emotion-physiology coupled dynamics model using resonance state identification results and instantaneous elastic recovery spectrum is as follows. First, the engineering objective of this step is to use newly occurring and evaluated "events" (i.e., resonance state identification and subsequent elastic recovery process) to perform online learning and optimization of the individualized emotion-physiology coupled dynamics model, enabling it to better predict and adapt to the patient's future physical and mental response patterns, forming a self-evolving closed-loop system.

[0090] The first step is to label the time-series data segment corresponding to the resonance state recognition result as a valid reinforcement learning sample. This data segment contains continuous time-series vital sign data that triggered resonance state recognition, corresponding emotional entropy change rate data, and prediction bias data generated by the system. In engineering, this labeling process involves extracting all relevant data within this specific time window (e.g., 5 minutes before and after resonance state recognition) from the real-time data stream, assigning each data segment a unique event ID, and storing it in a sample database specifically used for model updates. This sample represents the model's specific performance in a "prediction failure" or "high uncertainty" scenario.

[0091] The second step involves converting the immediate resilience spectrum calculated from this event into a quantified reward signal. The immediate resilience spectrum includes efficacy scores for multiple psychological regulatory pathways. Reward signal The aim is to quantify the "quality" of a patient's psychological recovery after the system triggers probe intervention. In engineering terms, the various efficacy scores in the immediate elastic recovery spectrum can be weighted and summed to obtain a comprehensive recovery efficacy score, which serves as a reward signal for this event. For example, calculations can be performed using formulas: ,in, This is a reward signal for this event. It is the first in the instantaneous elastic recovery spectrum The effectiveness score of each psychological regulation pathway It is the first The pre-defined weights of each psychological regulation pathway represent their importance in overall psychological resilience. A higher weight... The value indicates that the patient has good psychological adjustment ability after this stress, while the opposite indicates poor adjustment ability.

[0092] The third step involves employing reinforcement learning algorithms to adjust the internal parameters of the individualized emotion-physiology coupled dynamics model, aiming to minimize prediction bias in similar future situations. In engineering terms, this can be viewed as a model-based policy optimization process. When a labeled reinforcement learning sample and its reward signal... After being acquired, the system performs an update. Taking Q-learning or Deep Q-Network (DQN) algorithms as examples, the system state can be defined as the current physiological input and the internal state of the model, while the action is a specific way to adjust the model parameters. The magnitude of the prediction bias data for this event is related to the reward signal. This is combined to update the value function. The goal of the update is to significantly reduce the prediction bias of the adjusted model when it encounters similar physiological input patterns in the future. Specifically, the model parameters, i.e., the aforementioned polynomial coefficients, will be fine-tuned using gradient descent, with the gradient direction determined by the prediction error and reward signal of the current event. The decision is made jointly. After multiple iterations and updates, an optimized individualized emotional-physiological coupled dynamic model can be obtained. This optimized model has a stronger ability to capture and predict the patient's unique, nonlinear physical and mental response patterns.

[0093] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for creating a dynamic psychological profile by coupling chemotherapy symptoms with emotional entropy values, characterized in that, Includes the following steps: S1. Obtain continuous temporal vital signs data of the target patient during chemotherapy, and simultaneously obtain temporal emotional state data of the target patient. Calculate the rate of change of the target patient's emotional entropy based on the temporal emotional state data. S2. Based on continuous time-series vital sign data and emotional entropy change rate data, an individualized emotion-physiological coupling dynamic model is constructed to characterize the dynamic relationship between physiological fluctuations and emotional stability in target patients. S3. Input continuous time-series vital sign data into the individualized emotion-physiology coupling dynamics model in real time to generate emotion state prediction data, and compare the emotion state prediction data with the real-time acquired emotion entropy change rate data to obtain prediction deviation data. S4. Analyze the time series characteristics of the prediction deviation data to identify whether there are co-amplification patterns related to physiological or therapeutic rhythms. When a co-amplification pattern is identified, generate resonance state identification results including resonance intensity index and instability proximity index. S5. In response to the generation of resonance state recognition results, initiate a psychological resilience probe micro-experiment to the target patient through the terminal device to explore the effectiveness of specific psychological regulation pathways, and collect emotional state response data after the execution of the psychological resilience probe micro-experiment. S6. Analyze emotional state response data to quantify the dynamic characteristics of the emotional recovery process and calculate the instantaneous elastic recovery spectrum that includes efficacy scores of multiple psychological regulation pathways. S7. Couple the resonance state recognition results and the instantaneous elastic recovery spectrum to generate a dynamic psychological resilience profile of the target patient; S8. Using the resonance state identification results and instantaneous elastic recovery spectrum as feedback signals, the parameters of the individualized emotion-physiology coupling dynamics model are updated to obtain the optimized individualized emotion-physiology coupling dynamics model.

2. The method for dynamic profiling of psychological resilience by coupling chemotherapy signs and emotional entropy values ​​according to claim 1, characterized in that, The specific steps for calculating the rate of change of the target patient's emotional entropy based on temporal emotional state data include: The system periodically collects self-assessment data or voice data of the target patient's emotional state through an embedded micro-interactive interface. Perform emotional dimension analysis on self-reported emotional state data or voice data to generate temporal emotional state vectors; Calculate the probability distribution of the temporal sentiment state vector within the sliding time window, and calculate the temporal sentiment entropy based on the probability distribution; By performing time-series emotional entropy differentiation, the rate of change of emotional entropy of the target patients is obtained.

3. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms and emotional entropy values ​​according to claim 1, characterized in that, The specific steps for constructing an individualized emotion-physiology coupled dynamic model to characterize the dynamic relationship between physiological fluctuations and emotional stability in the target patient include: Continuous time-series vital sign data are processed into system input vectors, and emotional entropy change rate data are processed into system output responses; A system identification method is used to fit the dynamic relationship between the system input vector and the system output response, generating a set of preliminary coupled model parameters. The parameters of the initial coupling model are validated and calibrated using historical data windows. When the prediction error of the model is lower than the error threshold, the parameters of the initial coupling model are confirmed, and an individualized emotion-physiology coupling dynamic model is formed.

4. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms and emotional entropy values ​​according to claim 1, characterized in that, The specific steps for generating emotional state prediction data and comparing the emotional state prediction data with the real-time acquired emotional entropy change rate data to obtain prediction deviation data include: A real-time data buffer pool is constructed to store continuous time-series vital sign data and historical emotional entropy change rate data from the most recent N sampling periods, and time alignment and resampling processing are performed on heterogeneous data sources. Based on the autoregressive structure of the individualized emotion-physiology coupled dynamics model, the system uses data from the real-time data buffer pool to perform recursive prediction operations and calculate the predicted value of the emotion state at the current moment. The instantaneous difference between the predicted emotional state value and the real-time acquired emotional entropy change rate data is calculated, and the instantaneous difference is smoothed using the exponential moving average algorithm to generate prediction deviation data.

5. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms and emotional entropy values ​​according to claim 1, characterized in that, The specific steps for analyzing and predicting the time-series characteristics of the bias data to identify the existence of co-amplification patterns related to physiological or therapeutic rhythms include: Perform spectral analysis on the prediction deviation data to obtain the deviation spectral distribution; Physiological rhythm signals are extracted from continuous temporal vital sign data or external treatment cycle data are obtained to form rhythm reference signals; The coherence coefficient between the deviation spectral distribution and the spectrum of the rhythm reference signal is calculated. When the coherence coefficient exceeds the coherence threshold in a specific frequency band and the amplitude of the predicted deviation data shows a monotonically increasing trend with an increase rate exceeding the preset increase rate threshold, the existence of the cooperative amplification mode is confirmed.

6. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms and emotional entropy values ​​according to claim 1, characterized in that, The specific steps of the psychological resilience probe micro-experiment initiated by the terminal device to the target patient to explore the effectiveness of specific psychological regulatory pathways include: Construct a probe task library containing multiple probe tasks, where each probe task is associated with one or more psychological modulation pathways; Based on the characteristics of the resonance state identification results, target probe tasks are matched and selected from the probe task library; Send guidance instructions for the target probe task to the target patient's terminal device and monitor the target patient's task execution interaction data to ensure the effective completion of the task.

7. The method for dynamic profiling of psychological resilience by coupling chemotherapy signs and emotional entropy values ​​according to claim 6, characterized in that, The specific steps involved in calculating the immediate resilience spectrum, which includes efficacy scores for multiple psychological regulation pathways, include: A mood recovery curve is plotted based on emotional state response data; A function fit is performed on the emotion recovery curve to extract a set of morphological feature parameters including the recovery time constant and the curve decay type; Based on the type of psychological regulation pathway associated with the target probe task, the morphological feature parameters are mapped to the efficacy scores of the corresponding psychological regulation pathways, and all efficacy scores are aggregated into an instantaneous elastic recovery spectrum.

8. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms and emotional entropy values ​​according to claim 1, characterized in that, The specific steps for generating a dynamic psychological resilience profile of the target patient based on the coupled resonance state identification results and the instantaneous elastic recovery spectrum include: By integrating the resonance intensity index and the instability proximity index, the system vulnerability index is calculated. The efficacy scores of each psychological regulation pathway were analyzed from the immediate resilience recovery spectrum to form a profile of regulation strategy efficacy. By integrating the system vulnerability index and the effectiveness profile of the adjustment strategy with historical profile data, a dynamic psychological resilience profile is output in a visual form that includes trend curves and multi-dimensional charts.

9. The method for dynamic profiling of psychological resilience by coupling chemotherapy signs and emotional entropy values ​​according to claim 8, characterized in that, After generating a dynamic psychological resilience profile of the target patient, the process also includes: Construct a tiered intervention strategy library containing different intervention levels; Based on the value of the system vulnerability index, recommended intervention strategies are matched and selected from the tiered intervention strategy library; By combining recommended intervention strategies with dynamic psychological resilience profiles, a personalized nursing advice report is generated to assist healthcare professionals in making decisions and is then pushed to their terminals.

10. The method for creating a dynamic psychological profile by coupling chemotherapy symptoms with emotional entropy values ​​according to claim 1, characterized in that, The specific steps for updating the parameters of the individualized emotion-physiology coupled dynamics model to obtain the optimized individualized emotion-physiology coupled dynamics model include: The data segments corresponding to the resonance state recognition results are labeled as reinforcement learning samples; The performance scores in the instantaneous elastic recovery spectrum are weighted and fused to calculate the reward signal as feedback for this event; By employing reinforcement learning algorithms and utilizing reinforcement learning samples and reward signals, the parameters of the individualized emotion-physiology coupling dynamics model are iteratively adjusted to minimize prediction bias in similar future scenarios, thereby obtaining an optimized individualized emotion-physiology coupling dynamics model.

Images