A dynamic early warning system for physiological characteristics of prostate cancer endocrine therapy patients

Abstract

The application relates to the technical field of software, and discloses a dynamic early warning system for physiological characteristics of prostate cancer endocrine therapy patients, which comprises a core processing module configured to: dynamically allocate analysis weights for different categories of monitoring data based on the current endocrine therapy duration of a patient, wherein the analysis weight of high-frequency physiological symptom category data is increased in a treatment induction period, and the analysis weight of function and behavior category data is increased in a long-term maintenance period; setting an individualized baseline of monitoring data; adaptively adjusting the analysis weight and the individualized baseline according to periodic clinical calibration data; calculating the cumulative integral of the deviation of monitoring data relative to the individualized baseline in a sliding time window for a preset function or behavior index, and calculating a risk score; the system can distinguish between trends that need long-term attention, changes that are worth reminding, and dangers that must be immediately handled, and different levels of responses are given.

Description

Technical Field

[0001] This invention relates to the field of software technology, and in particular to a dynamic early warning system for physiological characteristics of patients undergoing endocrine therapy for prostate cancer. Background Technology

[0002] Endocrine therapy for prostate cancer is typically a long-term, chronic intervention, often lasting for several years and sometimes requiring lifelong management. During this process, the patient's physiological state and daily behavior constantly change, resulting in significant inter-stage variations in the large amount of monitoring data generated.

[0003] Currently, most home health management systems or general follow-up software still generally use static fixed thresholds for risk identification algorithms.

[0004] In addition, existing home management systems mostly use discrete single-point sampling-based judgment logic in their data processing methods. That is, they only compare the current value collected at a certain point in time with a preset standard to make a judgment, without conducting continuous analysis of historical data.

[0005] It is evident that existing data processing technologies are ill-suited to the evolving data characteristics throughout the entire prostate cancer ADT treatment cycle, and generally lack the ability to identify long-term, subtle but continuously accumulating risks. Traditional systems based on static rules and discrete judgments struggle to effectively reflect the true changes in patients' risk status when faced with long-term home management scenarios. Summary of the Invention

[0006] This invention aims to address the technical shortcomings of existing prostate cancer ADT home management systems, which commonly employ static fixed thresholds and single-point discrete sampling risk assessment algorithms. Specifically, existing systems are prone to generating numerous false positive alarms during the treatment induction phase due to high-frequency, low-risk fluctuations in physiological indicators, while failing to identify slowly changing but continuously accumulating progressive risks during the long-term maintenance phase, thus resulting in the underestimation of some hidden risks. A dynamic early warning system for physiological characteristics of prostate cancer patients undergoing endocrine therapy includes:

[0007] The data acquisition and access module is used to acquire patient monitoring data, including: high-frequency physiological symptom data, functional and behavioral data, and subjective state data in home settings, as well as periodic clinical calibration data in hospital settings.

[0008] The core processing module is configured as follows:

[0009] Based on the patient's current endocrine therapy duration, the analysis weights for different categories of monitoring data are dynamically allocated. Specifically, the analysis weights for high-frequency physiological symptom data are increased during the treatment induction phase, and the analysis weights for functional and behavioral data are increased during the long-term maintenance phase.

[0010] Establish individualized baselines for monitoring data;

[0011] The analysis weights and the individualized baseline are adaptively adjusted based on the periodic clinical calibration data.

[0012] For preset functional or behavioral indicators, the cumulative integral of the deviation of monitoring data from an individualized baseline within a sliding time window is calculated to obtain a risk score; and

[0013] The early warning output module is used to generate and output risk early warning information based on the dynamically allocated analysis weights, the results of the adaptive adjustment, and the risk score.

[0014] The data acquisition and access module further includes a physiological signal acquisition unit, which is used to acquire high-frequency physiological symptom data, including:

[0015] Heart rate variability: Heart rate signals are continuously acquired using a photoplethysmography (PPG) sensor, and SDNN and RMSSD are calculated.

[0016] Sleep structure parameters: Based on accelerometer and heart rate signals, the sleep process is analyzed to obtain the duration of deep sleep, the proportion of rapid eye movement (REM) sleep and the number of nighttime awakenings (WASO).

[0017] Hot flash frequency: By combining the skin conductance response (GSR) signal with the characteristics of sudden heart rate increase, the system automatically identifies and marks the time of occurrence, duration and frequency of vasomotor symptoms.

[0018] The data acquisition and access module further includes a function and behavior signal analysis unit, which is used to acquire function and behavior data, including:

[0019] Muscle strength related indicators: The amplitude changes of microtremors during upper limb activities are captured using a triaxial accelerometer, and a grip strength meter can be optionally equipped to obtain grip strength data;

[0020] Gait stability characteristics: Gait cycle analysis was used to calculate gait frequency variability and the proportion of time in the dual support phase.

[0021] Standing / Walking Time: Identify the patient's movement patterns as they transition from a sedentary state to a standing state, and calculate the time required to transition from a static sitting posture to a stable standing posture;

[0022] Daily activity level: Statistics on daily effective steps and duration of moderate-to-vigorous physical activity (MVPA).

[0023] The data acquisition and access module further includes a subjective state interaction unit, which collects subjective state data actively reported by patients via a mobile terminal. The subjective state data includes:

[0024] Standardized fatigue rating;

[0025] Psychological stress score;

[0026] Recent fall incidents;

[0027] Hot flash self-rating scale.

[0028] The data acquisition and access module also includes a hospital low-frequency periodic calibration module. This module acquires clinical testing data with a time resolution of monthly or quarterly, periodically calibrates the baseline parameters of the home monitoring algorithm, and obtains periodic clinical calibration data. The periodic clinical calibration data includes:

[0029] Prostate-specific antigen (PSA) levels;

[0030] Serum testosterone levels;

[0031] Bone metabolism-related indicators include bone mineral density T-score and / or bone turnover markers.

[0032] Establish individualized baselines for monitoring data, including:

[0033] Using the patient's most recent hospital periodic calibration data as a reference, and on the premise that the relevant low-frequency clinical indicators are in a stable or normal range, the average value of the home continuous monitoring data in the first preset period after the calibration node is selected as the individualized baseline of the monitoring data.

[0034] The formula for the risk score is:

[0035]

[0036] in,

[0037] This represents the monitoring indicator value corresponding to time t.

[0038] For individualized baselines;

[0039] The time-weighted factor;

[0040] T is the length of the integration window.

[0041] The risk classification and response output unit includes:

[0042] The weighted scores of each monitoring indicator and their corresponding cumulative risk scores are all within the preset safety range, and no abnormal trends or event signals have been detected.

[0043] The relevant monitoring data is silently stored and archived, and the long-term trend is displayed to patients or medical staff in a low-frequency manner without triggering immediate intervention prompts.

[0044] The risk classification and response output unit includes:

[0045] If a single or short-term monitoring value does not exceed the traditional alarm threshold, but the cumulative risk score exceeds the corresponding trend warning threshold, it indicates a long-term, gradual risk of functional or status degradation.

[0046] Trigger trend warnings and provide patients with targeted self-management or rehabilitation advice based on risk characteristics.

[0047] The risk classification and response output unit includes:

[0048] Any of the following conditions were detected:

[0049] A near-fall event or other high-risk behavioral event occurs;

[0050] The rapid and abnormal changes in high-frequency physiological signals and the subjective state of discomfort exhibit nonlinear synergistic changes over time, indicating an amplification of potential acute risks.

[0051] Immediately trigger a high-priority risk response, including but not limited to outputting an instant alert to the patient's terminal and sending a high-priority risk notification request to the back-end medical management terminal, indicating that further evaluation or manual intervention is required.

[0052] The present invention has the following beneficial effects:

[0053] This invention utilizes low-frequency, high-reliability hospital data to dynamically correct the risk assessment boundary of home monitoring, avoiding sensitivity imbalance caused by long-term drift of a single home data source; it improves early warning sensitivity when the risk of disease progression increases, and reduces the probability of false alarms when the objective risk is low.

[0054] This invention distinguishes between long-term trend risks and acute event risks; identifies potential functional decline in advance before obvious abnormalities appear; avoids excessive intervention in low-risk states, while enabling rapid response to high-risk states. Attached Figure Description

[0055] Figure 1 This is a first flowchart of a dynamic early warning system for physiological characteristics of patients undergoing endocrine therapy for prostate cancer, provided in an embodiment of the present invention. Detailed Implementation

[0056] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0057] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the dynamic early warning system for physiological characteristics of prostate cancer patients undergoing endocrine therapy in this invention includes:

[0058] The data acquisition and access module is used to acquire patient monitoring data, including: high-frequency physiological symptom data, functional and behavioral data, and subjective state data in home settings, as well as periodic clinical calibration data in hospital settings.

[0059] The core processing module is configured as follows:

[0060] Based on the patient's current endocrine therapy duration, the analysis weights for different categories of monitoring data are dynamically allocated. Specifically, the analysis weights for high-frequency physiological symptom data are increased during the treatment induction phase, and the analysis weights for functional and behavioral data are increased during the long-term maintenance phase.

[0061] Establish individualized baselines for monitoring data;

[0062] The analysis weights and the individualized baseline are adaptively adjusted based on the periodic clinical calibration data.

[0063] For preset functional or behavioral indicators, the cumulative integral of the deviation of monitoring data from an individualized baseline within a sliding time window is calculated to obtain a risk score; and

[0064] The early warning output module is used to generate and output risk early warning information based on the dynamically allocated analysis weights, the results of the adaptive adjustment, and the risk score.

[0065] Data acquisition and access module

[0066] The data acquisition and access layer of this system adopts a dual-dimensional input architecture, corresponding to high-frequency continuous monitoring in home scenarios and low-frequency periodic calibration in hospital scenarios, so as to achieve synergy between long-term risk monitoring and clinical benchmark correction.

[0067] This module is deployed on the patient's mobile terminal (such as a smartphone) and associated wearable devices (such as smart bracelets, smartwatches, body fat scales, etc.) to collect dynamic physiological and behavioral data with a time resolution of hours or days.

[0068] 1.1 Physiological signal acquisition unit:

[0069] (1) Heart rate variability (HRV): Heart rate signals are continuously collected by photoplethysmography (PPG) sensor and indicators such as SDNN and RMSSD are calculated to quantify the adaptive response of the autonomic nervous system to endocrine therapy and related physiological stress.

[0070] (2) Sleep structure parameters: Based on accelerometer and heart rate signals, the sleep process is analyzed to obtain structural parameters such as deep sleep duration, REM sleep ratio and nighttime awakenings (WASO).

[0071] (3) Frequency of hot flashes: By combining the skin conductance response (GSR) signal and the characteristics of sudden heart rate increase, the occurrence time, duration and frequency of vasomotor symptoms are automatically identified and marked.

[0072] 1.2 Functional and Behavioral Signal Analysis Unit:

[0073] (1) Muscle strength related indicators: The amplitude changes of micro-tremors during upper limb activities are captured by a triaxial accelerometer, and a grip strength meter can be selected to obtain grip strength data to reflect muscle strength status and its changing trend.

[0074] (2) Gait stability characteristics: By analyzing the gait cycle, the gait frequency variability and the proportion of time in the double support phase are calculated to assess walking stability and potential fall risk.

[0075] (3) Standing / walking time: The system automatically identifies the patient's movement pattern from a prolonged sitting state to a standing state, calculates the time required from a static sitting posture to a stable standing posture, and uses it as a proxy indicator of lower limb explosive power and functional status.

[0076] (4) Daily activity volume: The number of effective steps and duration of moderate to high intensity physical activity (MVPA) are counted daily to assess overall physical fitness level and activity trend.

[0077] 1.3 Subjective State Interaction Unit:

[0078] The system collects subjective status information voluntarily reported by patients through an interactive electronic patient-reported outcome (ePRO) questionnaire on a mobile device, including but not limited to:

[0079] (1) Standardized fatigue rating (such as a simplified version of the BFI scale);

[0080] (2) Psychological stress score;

[0081] (3) The occurrence of near-fall events (i.e., stumbling but not actually landing).

[0082] (4) Hot flash self-rating scale

[0083] 1.4 Hospital Low-Frequency Cyclic Calibration Module

[0084] This module acquires clinical testing data with a monthly or quarterly time resolution through hospital information system (HIS) interfaces, electronic medical record integration, or OCR recognition technology on laboratory reports. This data is used to periodically calibrate the baseline parameters of the home monitoring algorithm. The clinical data includes, but is not limited to:

[0085] (1) Prostate-specific antigen (PSA) level: used to assess the biochemical recurrence status of tumors and treatment response.

[0086] (2) Serum testosterone level: used to verify the castration effect of drugs and the effectiveness of endocrine therapy.

[0087] (3) Bone metabolism related indicators: including bone mineral density (BMD) T value and / or bone turnover markers, which are used as base parameters for long-term bone loss and fracture risk assessment.

[0088] Core Algorithm Processing Flow

[0089] The system constructs a dynamic risk assessment process based on treatment time sequence for the above-mentioned multi-source physiological and behavioral data. Its core processing logic includes the following three consecutive steps: treatment time sequence stage mapping, dual-stream feature fusion, and nonlinear risk calculation.

[0090] Step 1: Dynamic weight allocation based on treatment time sequence

[0091] The system has a built-in time-series model related to therapeutic pharmacodynamics, which is used to read the duration parameter t of the patient's current ADT treatment, and based on this time parameter, dynamically allocate and continuously adjust the analysis weights W(t) of different categories of monitoring indicators.

[0092] It should be noted that the weighting coefficients are not fixed constants, but are used to illustrate the changes in the system's focus at different stages of treatment. Their specific values ​​can be adjusted according to clinical needs.

[0093] Scenario A: Induced Oscillation Period

[0094] Stage characteristics:

[0095] The patient is in the acute castration adaptation phase at the beginning of endocrine therapy, with frequent fluctuations in physiological and subjective symptoms and obvious short-term discomfort.

[0096] Weighting strategy:

[0097] The system automatically increases the analytical weight of high-frequency physiological signals and subjective state indicators to enhance sensitivity to acute discomfort and its potential behavioral effects, for example:

[0098] Hot flashes are associated with high-weighted configurations (e.g., W≈0.4).

[0099] Heart rate variability (HRV): Higher weighting (example: W≈0.3);

[0100] Functional and motor-related indicators (such as muscle strength and gait) are relatively downweighted (e.g., W≈0.1) and used as background reference signals.

[0101] During this phase, the system focuses on accurately identifying the psychological burden and risk of treatment adherence disruption caused by acute discomfort. Simultaneously, by applying bandpass or lowpass filtering to functional signals, the system reduces the interference of temporary gait abnormalities or activity fluctuations caused by short-term physiological discomfort on risk assessment, thereby minimizing false alarms.

[0102] Scenario B: Transitional Phase (3 months ≤ t ≤ 6 months)

[0103] Stage characteristics:

[0104] Patients gradually adapt to hormone suppression, with high-frequency symptom fluctuations coexisting with low-frequency functional changes, resulting in a mixed risk profile.

[0105] Weighting strategy:

[0106] The system achieves a smooth shift in the monitoring focus through a continuously time-varying weight function:

[0107] The weights of indicators related to high-frequency physiological symptoms gradually decrease over time;

[0108] The weights of functional and behavioral indicators (such as muscle strength, activity level, and gait stability) are gradually increasing.

[0109] By adopting the above approach, we can avoid the discontinuity in risk assessment caused by sudden changes in weights at the stage boundaries, and at the same time establish a stable characteristic transition basis for subsequent long-term cumulative risk assessment.

[0110] Scenario C: Maintenance Phase (t>6 months)

[0111] Stage characteristics:

[0112] Acute discomfort symptoms are significantly reduced or disappear, but the metabolic side effects and functional decline risks associated with endocrine therapy begin to accumulate over a long period.

[0113] Weighting strategy:

[0114] The system proactively shifts its monitoring focus, significantly increasing the weight given to functional and behavioral signals in its analysis. For example:

[0115] Muscle strength and gait stability related features: high weighting (example: W≈0.5);

[0116] Daily activity level and physical fitness trend indicators: medium to high weighting (example: W≈0.3);

[0117] High-frequency symptom indicators such as hot flashes: weighted down (example: W≈0.1).

[0118] In this phase, the system is primarily used to identify latent functional deterioration risks that do not exceed a single-point threshold but continue to worsen over time, such as sarcopenia progression and increased risk of falls.

[0119] Step 2: Calculation of cumulative integral for weak trends

[0120] To address the risk of progressive functional decline (such as a slow decrease in muscle strength or activity level) that is easily overlooked by single-point threshold determination methods during continuous home monitoring, this system does not use threshold comparison logic at a single time point. Instead, it introduces a time-based cumulative integral calculation mechanism to identify long-term, subtle but consistent risk changes.

[0121] (a) Definition of Individualized Baseline

[0122] The system uses the patient's most recent hospital-based periodic calibration data as a reference. Provided that relevant low-frequency clinical indicators are confirmed to be stable or within the normal range, the average of continuous home monitoring data within the first preset period (e.g., 7 days) after the calibration node is selected as the individualized reference baseline for that physiological or functional indicator, denoted as:

[0123] This baseline is used to reflect the patient's relatively stable functional status at the current stage of treatment.

[0124] (II) Calculation method of cumulative integral

[0125] Within a preset sliding time window T (e.g., 30 days), the system continuously accumulates the deviation of the target monitoring indicator V(t) relative to the individualized baseline to quantify the risk load generated by long-term trend changes. Its risk score is defined as follows:

[0126]

[0127] in,

[0128] This represents the monitoring indicator value corresponding to time t.

[0129] A time-weighted factor is used to enhance the contribution of persistent deviations in risk assessment.

[0130] T is the length of the integration window, used to control the time scale of trend accumulation.

[0131] Risk assessment logic

[0132] The system does not use whether a single day's indicator exceeds the alarm threshold as the sole criterion for judgment, but rather makes a comprehensive judgment based on the cumulative score results:

[0133] When a single monitoring value If the patient's cumulative risk score exceeds the preset compensatory exhaustion threshold but the traditional alarm threshold is not triggered, the system determines that the patient has a hidden risk of functional decline and triggers a corresponding warning.

[0134] The compensation exhaustion threshold is used to characterize the situation where a patient's physiological compensatory capacity is gradually depleted and approaches a critical state during a long-term, mild functional decline.

[0135] Through the above mechanism, the system can identify potential progressive risks in advance, before clinically obvious abnormalities appear.

[0136] Application Examples

[0137] For example, in one embodiment, the patient's stand-up-walk time gradually increased from 1.2 seconds to 1.5 seconds over a continuous monitoring period. Although this change remained within the traditional reference range, the cumulative integration results showed a continuous monotonic downward trend in lower limb explosive power. Based on this, the system determined that the patient had a latent signal of increased fall risk and issued an early risk warning.

[0138] Step 3: Closed-loop calibration mechanism based on hospital data

[0139] This step is used to achieve closed-loop collaborative calibration between high-frequency home monitoring and low-frequency clinical data from hospitals, which is one of the key technical mechanisms that distinguishes this system from conventional home health management applications.

[0140] The system dynamically adjusts the risk assessment sensitivity of the home monitoring algorithm by introducing low-frequency "gold standard" data obtained in hospital settings, thereby improving the accuracy and clinical consistency of long-term risk identification.

[0141] (a) Calibration trigger logic

[0142] When the system accesses or updates the patient's latest hospital test data, including but not limited to PSA levels, serum testosterone levels, or bone metabolism-related parameters, the system automatically initiates a closed-loop calibration process to reverse-adjust the alarm thresholds, weight distributions, or baseline parameters of relevant indicators during the home monitoring phase.

[0143] (II) Calibration Strategy Example 1: Disease Progression Risk Enhancement Calibration

[0144] Calibration conditions:

[0145] When hospital test results indicate elevated PSA (signs of biochemical relapse) or abnormally elevated testosterone levels, it suggests a decline in tumor control or a risk of endocrine escape.

[0146] Calibration action:

[0147] The system automatically lowers the alarm thresholds for indicators related to overall health and symptom tolerance during home monitoring, including but not limited to fatigue scores, bone pain-related signals, or changes in activity levels, making the system more sensitive to minor abnormalities.

[0148] A deterioration in clinical indicators usually signifies a decline in a patient's physiological compensatory capacity. In this context, even small abnormal fluctuations during home monitoring may be associated with the risk of disease progression or metastasis. Therefore, improving alarm sensitivity can help detect potential signs of clinical deterioration earlier.

[0149] (III) Example 2 of Calibration Strategy: Functional Risk Denoising Calibration

[0150] Calibration conditions:

[0151] When hospital test results show that bone metabolism is stable, for example, bone mineral density (BMD) is within the normal or relatively stable range.

[0152] Calibration action:

[0153] The system accordingly increases its tolerance for occasional functional abnormalities during home monitoring, such as short-term near-fall events or mild gait fluctuations, and can reset the cumulative baseline of relevant functional indicators.

[0154] Assuming good objective bone health, short-term functional abnormalities are more likely to stem from environmental factors or incidental behavior than from pathological fractures or systemic functional degeneration. Reducing overreaction to such events can effectively decrease false alarms and improve system stability.

[0155] Through the aforementioned closed-loop calibration mechanism, the system achieves the following technical effects:

[0156] Utilize low-frequency, high-reliability hospital data to dynamically adjust the risk assessment boundaries for home monitoring;

[0157] Avoid sensitivity imbalance caused by long-term drift of a single home data source;

[0158] Improve early warning sensitivity when the risk of disease progression is high, and reduce the probability of false alarms when the objective risk is low.

[0159] Risk Classification and Response Output Unit

[0160] Based on the aforementioned dynamic weight allocation, weak trend cumulative integration, and closed-loop calibration results, the system classifies the patient's current risk status and outputs corresponding graded response instructions. The risk grade is dynamically updatable, supporting upgrades or rollbacks over time and as data changes.

[0161] In one embodiment, the system includes at least the following three levels of risk states and corresponding response mechanisms:

[0162] Routine monitoring status

[0163] Judgment conditions:

[0164] The weighted scores of each monitoring indicator and their corresponding cumulative risk scores are all within the preset safety range, and no abnormal trends or event signals have been detected.

[0165] System response:

[0166] The system silently stores and archives the relevant monitoring data, and displays long-term trends to patients or medical staff in a low-frequency manner (such as periodic weekly reports or trend summaries) without triggering immediate intervention prompts.

[0167] Trend warning status

[0168] Judgment conditions:

[0169] If a single or short-term monitoring value does not exceed the traditional alarm threshold, but the cumulative risk score exceeds the corresponding trend warning threshold, it indicates a long-term, gradual risk of functional or state degradation.

[0170] System response:

[0171] The system triggers trend warnings and provides patients with targeted self-management or rehabilitation suggestions based on risk characteristics, such as increasing specific types of functional training, adjusting activity structures, or increasing the monitoring frequency of relevant indicators.

[0172] High-risk event status

[0173] Judgment conditions:

[0174] The system detected any of the following:

[0175] A near-fall event or other high-risk behavioral event occurs;

[0176] Rapidly aberrant changes in high-frequency physiological signals (such as a significant decrease in heart rate variability) and subjective discomfort (such as elevated stress or pain scores) exhibit nonlinear synergistic changes over time, indicating a potential amplification of acute risks.

[0177] System response:

[0178] The system immediately triggers a high-priority risk response, including but not limited to outputting an instant alert to the patient's terminal and sending a high-priority risk notification request to the back-end medical management terminal to indicate that further evaluation or manual intervention is required.

[0179] Through the aforementioned risk classification and response output mechanism, the system achieves the following technical effects:

[0180] Distinguish between long-term trend risks and acute event risks;

[0181] Identify potential functional decline before obvious abnormalities appear;

[0182] Avoid excessive intervention in low-risk situations, while enabling rapid response in high-risk situations.

[0183] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic early warning system for physiological characteristics of patients undergoing endocrine therapy for prostate cancer, characterized in that, The system includes: The data acquisition and access module is used to acquire patient monitoring data, including: high-frequency physiological symptom data, functional and behavioral data, and subjective state data in home settings, as well as periodic clinical calibration data in hospital settings. The core processing module is configured as follows: Based on the patient's current endocrine therapy duration, the analysis weights for different categories of monitoring data are dynamically allocated. Specifically, the analysis weights for high-frequency physiological symptom data are increased during the treatment induction phase, and the analysis weights for functional and behavioral data are increased during the long-term maintenance phase. Establish individualized baselines for monitoring data; The analysis weights and the individualized baseline are adaptively adjusted based on the periodic clinical calibration data. For preset functional or behavioral indicators, the cumulative integral of the deviation of monitoring data from an individualized baseline within a sliding time window is calculated to obtain a risk score; and The early warning output module is used to generate and output risk early warning information based on the dynamically allocated analysis weights, the results of the adaptive adjustment, and the risk score. The formula for the risk score is: in, This represents the monitoring indicator value corresponding to time t. For individualized baselines; The time-weighted factor; T is the length of the integration window.

2. The system according to claim 1, characterized in that, The data acquisition and access module further includes a physiological signal acquisition unit, which is used to acquire high-frequency physiological symptom data, including: Heart rate variability: Heart rate signals are continuously acquired using a photoplethysmography (PPG) sensor, and SDNN and RMSSD are calculated. Sleep structure parameters: Based on accelerometer and heart rate signals, the sleep process is analyzed to obtain the duration of deep sleep, the proportion of rapid eye movement (REM) sleep and the number of nighttime awakenings (WASO). Hot flash frequency: By combining the skin conductance response (GSR) signal with the characteristics of sudden heart rate increase, the system automatically identifies and marks the time of occurrence, duration and frequency of vasomotor symptoms.

3. The system according to claim 1, characterized in that, The data acquisition and access module further includes a function and behavior signal analysis unit, which is used to acquire function and behavior data, including: Muscle strength related indicators: The amplitude changes of microtremors during upper limb activities are captured using a triaxial accelerometer, and a grip strength meter can be optionally equipped to obtain grip strength data; Gait stability characteristics: Gait cycle analysis was used to calculate gait frequency variability and the proportion of time in the dual support phase. Standing / Walking Time: Identify the patient's movement patterns as they transition from a sedentary state to a standing state, and calculate the time required to transition from a static sitting posture to a stable standing posture; Daily activity level: Statistics on daily effective steps and duration of moderate-to-vigorous physical activity (MVPA).

4. The system according to claim 1, characterized in that, The data acquisition and access module further includes a subjective state interaction unit, which collects subjective state data actively reported by the patient via a mobile terminal. The subjective state data includes: Standardized fatigue rating; Psychological stress score; Recent fall incidents; Hot flash self-rating scale.

5. The system according to claim 1, characterized in that, The data acquisition and access module also includes a hospital low-frequency periodic calibration module. This module acquires clinical testing data with a time resolution of monthly or quarterly, periodically calibrates the baseline parameters of the home monitoring algorithm, and obtains periodic clinical calibration data. The periodic clinical calibration data includes: Prostate-specific antigen (PSA) levels; Serum testosterone levels; Bone metabolism-related indicators include bone mineral density T-score and / or bone turnover markers.

6. The system according to claim 1, characterized in that, Establish individualized baselines for monitoring data, including: Using the patient's most recent hospital periodic calibration data as a reference, and on the premise that the relevant low-frequency clinical indicators are in a stable or normal range, the average value of the home continuous monitoring data in the first preset period after the calibration node is selected as the individualized baseline of the monitoring data.

7. The system according to claim 1, characterized in that, The system also includes a risk classification and response output unit, comprising: The weighted scores of each monitoring indicator and their corresponding cumulative risk scores are all within the preset safety range, and no abnormal trends or event signals have been detected. The relevant monitoring data is silently stored and archived, and the long-term trend is displayed to patients or medical staff in a low-frequency manner without triggering immediate intervention prompts.

8. The system according to claim 1, characterized in that, The system also includes a risk classification and response output unit, comprising: If a single or short-term monitoring value does not exceed the traditional alarm threshold, but the cumulative risk score exceeds the corresponding trend warning threshold, it indicates a long-term, gradual risk of functional or status degradation. Trigger trend warnings and provide patients with targeted self-management or rehabilitation advice based on risk characteristics.

9. The system according to claim 1, characterized in that, The system also includes a risk classification and response output unit, comprising: Any of the following conditions were detected: A near-fall event or other high-risk behavioral event occurs; The rapid and abnormal changes in high-frequency physiological signals and the subjective state of discomfort exhibit nonlinear synergistic changes over time, indicating an amplification of potential acute risks. Immediately trigger a high-priority risk response, including but not limited to outputting an instant alert to the patient's terminal and sending a high-priority risk notification request to the back-end medical management terminal, indicating that further evaluation or manual intervention is required.

Images