Tumor patient symptom group collaborative management system

Abstract

The invention relates to the technical field of tumor treatment, in particular to a tumor patient symptom group collaborative management system, comprising a data acquisition module used for acquiring patient information and environment information through an intelligent device; the feature extraction module is used for data acquisition of the data acquisition module, and the feature extraction module performs feature extraction on the patient information and the environment information to obtain time domain features, frequency domain features and nonlinear features corresponding to symptoms; the evaluation platform identifies the core symptom group of the patient according to the time domain feature, the frequency domain feature and the nonlinear feature and calculates a total load index; the evaluation platform also performs symptom deterioration early warning; the collaboration platform is used for storing a disease knowledge base and identifying and obtaining a corresponding multi-mode intervention scheme according to the condition of a patient; and the patient terminal is used for pushing the multi-mode intervention scheme to the patient. According to the invention, the symptom group of the tumor patient can be managed as a whole, a personalized and real-time intervention scheme is provided for the patient, and the treatment effect is improved.

Description

Technical Field

[0001] This invention relates to the field of tumor treatment technology, and in particular to a collaborative management system for symptom clusters in tumor patients. Background Technology

[0002] Symptom cluster management in cancer patients is an important component of the cancer diagnosis and treatment system. Cancer patients often experience multiple interrelated symptoms simultaneously during treatment, such as pain-fatigue-sleep disturbances, nausea-vomiting-anorexia, etc., collectively referred to as "symptom clusters." Symptom cluster management in cancer patients generally employs a comprehensive approach, including pharmacological, non-pharmacological, and psychological interventions, to prevent, alleviate, or control the physiological and psychological symptoms that occur during disease progression, treatment, and recovery. Current technologies have the following drawbacks:

[0003] 1. Currently, the medical system usually treats individual symptoms separately, ignoring the synergistic effect between symptoms, resulting in poor treatment outcomes and even medication conflicts.

[0004] 2. Patient symptom data is mainly obtained through regular outpatient visits, which lacks real-time information and makes it impossible to achieve early warning and timely intervention.

[0005] 3. Symptom assessment relies heavily on patients' subjective descriptions (such as NRS scores) and lacks objective quantitative data support, resulting in insufficient accuracy.

[0006] 4. Intervention plans are mostly general guidelines, making it difficult to adjust them in real time according to individual patient differences and dynamic changes in symptom clusters. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides a collaborative management system for cancer patient symptom clusters, which manages cancer patient symptom clusters as a whole, providing patients with personalized and real-time intervention plans to improve treatment outcomes.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A collaborative management system for symptom clusters in cancer patients includes a data acquisition module, a feature extraction module, an assessment platform, a collaboration platform, and a patient terminal.

[0010] The data acquisition module is used to acquire patient information and environmental information through smart devices. The patient information includes medical record information, symptom information, medication information, emotional information, and basic information. The environmental information includes indoor humidity and indoor temperature.

[0011] The feature extraction module is used for data acquisition by the data acquisition module, and the feature extraction module performs feature extraction on the patient information and the environmental information to obtain time-domain features, frequency-domain features and nonlinear features corresponding to the symptoms;

[0012] The assessment platform is used for data acquisition by the feature extraction module, and the assessment platform identifies the patient's core symptom clusters based on the time-domain features, the frequency-domain features, and the nonlinear features, and the assessment platform calculates the total burden index based on the core symptom clusters; the assessment platform analyzes the core symptom clusters based on time series data to provide early warning of symptom deterioration;

[0013] The collaborative platform is used for storing the disease knowledge base. The collaborative platform acquires data from the feature extraction module and the evaluation platform to identify and obtain corresponding multimodal intervention plans based on the patient's condition.

[0014] The patient terminal is used to acquire data from the assessment platform and the collaboration platform in order to push the multimodal intervention plan to the patient.

[0015] Furthermore, the feature extraction module statistically analyzes the patient's symptoms, medication information, emotional information, indoor humidity, and indoor temperature on the time axis from the patient information and the environmental information to obtain time-domain features; the feature extraction module transforms the patient information and the environmental information and decomposes them into different frequency components to obtain frequency-domain features of symptom rhythm; the feature extraction module extracts entropy features from the patient information and the environmental information to analyze symptom fluctuations, post-medication symptom conditions, and the correlation between emotional fluctuations and symptoms to obtain nonlinear features.

[0016] Furthermore, the evaluation platform includes a data processing module, an identification module, and an early warning module.

[0017] The data processing module acquires data from the feature extraction module, and the data processing module sets different weights for the time-domain features, the frequency-domain features, and the nonlinear features based on the medical record information and the basic information. The data processing module also fuses the time-domain features, the frequency-domain features, and the nonlinear features based on the weights to obtain a feature data matrix.

[0018] The identification module analyzes the feature data matrix using machine learning algorithms to identify core symptom clusters, and the identification module performs standardized severity scoring on the core symptom clusters to calculate the total burden index.

[0019] The early warning module analyzes the core symptom cluster and the total burden index based on time series data to obtain early warning signals.

[0020] Furthermore, the data processing module sequentially performs feature alignment, feature standardization, and feature dimensionality reduction on the time-domain features, frequency-domain features, and nonlinear features, and then fuses them into a feature data matrix:

[0021] Formula (1),

[0022] in, For the first Attention weights for each feature; To convert the scores into a probability distribution; To calculate the attention score; This is the attention weight vector; To introduce nonlinearity through the tanh activation function; This is the weight matrix; For the first Feature vectors of each modality; This is the feature data matrix.

[0023] Furthermore, the identification module analyzes the feature data matrix using machine learning algorithms, outputs classification labels to obtain core symptom clusters, and calculates the total burden index based on subjective symptom intensity and multimodal feature prediction probabilities.

[0024] Formula (2),

[0025] in, Total load index; The number of core symptoms; Symptom weights; Standardized symptom scoring; For characteristic correction coefficients; Predict the probability of symptom clusters.

[0026] Furthermore, the early warning module constructs an LSTM prediction model and analyzes the core symptom cluster and the total load index through change point detection to obtain the mutation point of the total load index. The early warning module predicts the occurrence time of the next mutation point based on the duration of two adjacent mutation points. When the occurrence time of the next mutation point is less than a preset duration, the early warning module issues an early warning signal.

[0027] Furthermore, the collaborative platform includes a knowledge base module, an annotation module, and a treatment recommendation module.

[0028] The knowledge base module is used to store a disease knowledge base, which includes knowledge of oncology, pharmacy, nursing, nutrition, and psychology. The disease knowledge base also stores historical cases and corresponding treatment plans.

[0029] The annotation module is used to annotate the data in the knowledge base module to annotate the synergistic effects and contraindications of different intervention measures on multiple symptoms, and the annotation module constructs a knowledge graph based on the annotation information;

[0030] The treatment recommendation module is used for data acquisition by the feature extraction module, the knowledge base module, and the annotation module. The treatment recommendation module constructs a personal feature map of the patient based on the feature extraction module, and the treatment recommendation module matches the personal feature map with the knowledge map through text recognition to obtain a multimodal intervention plan.

[0031] Furthermore, the treatment recommendation module obtains multiple candidate solutions, sets corresponding matching weights for each piece of information in the patient information, and performs similarity matching between the historical cases corresponding to the candidate solutions and the patient information based on the matching weights. The treatment recommendation module selects the candidate solution with the highest similarity as the multimodal intervention solution.

[0032] Furthermore, the collaborative platform also includes an adjustment module, which is used for data acquisition by the treatment recommendation module, and the adjustment module analyzes the sub-schemes in the multimodal intervention plan to prioritize the sub-schemes in the multimodal intervention plan according to the severity of the patient's symptoms.

[0033] Furthermore, the collaborative platform also includes a verification module, which is used to send the data from the treatment recommendation module and the adjustment module to the medical staff console to review the multimodal intervention plan.

[0034] The beneficial effects of this invention are:

[0035] 1. The data acquisition module, through intelligent devices, can comprehensively and accurately obtain patients' medical records, symptom information, medication information, emotional information, and basic information. It also incorporates indoor humidity and temperature as feature data, improving the accuracy of subsequent symptom cluster collaborative management assessments. The feature extraction module extracts symptom-related time-domain, frequency-domain, and nonlinear features from multimodal data, reducing the cost of identifying core symptom clusters. This allows the assessment platform to accurately correlate core symptom clusters, improving the accuracy of the total burden index calculation. Furthermore, the assessment platform can provide early warnings of symptom deterioration based on core symptom clusters, sending warning signals to medical staff and patients in advance to prevent patients from missing the optimal treatment time. This shifts from passive response to proactive prediction and early warning, preventing problems before they arise. With the collaborative platform, personalized dynamic management plans are provided based on the patient's condition, improving treatment efficacy and quality of life.

[0036] 2. The data processing module fuses time-domain, frequency-domain, and nonlinear features based on weights, enabling a more comprehensive representation of the features and improving the response of important features. The identification module uses machine learning algorithms to process the feature data matrix, enabling efficient analysis of large-scale symptom data, revealing symptom association patterns that are difficult for humans to detect, and continuously optimizing symptom cluster identification as data accumulates, achieving dynamic updates. The early warning module obtains the duration of two adjacent mutation points based on the core symptom cluster and the total burden index, thereby predicting the time of the next mutation point and enabling prediction of the worsening trend of the symptom cluster.

[0037] 3. A disease knowledge base is constructed by integrating knowledge from oncology, pharmacy, nursing, nutrition, and psychology, and also stores historical cases and corresponding treatment plans. The synergistic effects and contraindications of different interventions on multiple symptoms are labeled, providing accurate data for the treatment recommendation module and improving the accuracy of multimodal intervention plans. Furthermore, the treatment recommendation module can match candidate plans based on patient information, making multimodal intervention plans more suitable for patients and improving treatment outcomes. By adjusting the priority of sub-plans within the multimodal intervention plan, the module recommends implementing the most effective and urgent measures first, effectively reducing patient suffering. Attached Figure Description

[0038] Figure 1 This is a structural block diagram of a tumor patient symptom cluster collaborative management system according to a preferred embodiment of the present invention.

[0039] In the diagram, 1-Data acquisition module, 2-Feature extraction module, 3-Evaluation platform, 31-Data processing module, 32-Identification module, 33-Early warning module, 4-Collaboration platform, 41-Knowledge base module, 42-Annotation module, 43-Treatment recommendation module, 44-Adjustment module, 45-Verification module, and 5-Patient terminal. Detailed Implementation

[0040] 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.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0042] See Figure 1 The present invention provides a preferred embodiment of a collaborative management system for symptom clusters in cancer patients, comprising a data acquisition module 1, a feature extraction module 2, an assessment platform 3, a collaborative platform 4, and a patient terminal 5.

[0043] Data acquisition module 1 is used to acquire patient and environmental information through smart devices. Patient information includes medical records, symptoms, medication information, emotional information, and basic information, such as the patient's gender, age, height, and weight. Environmental information includes indoor humidity and indoor temperature.

[0044] In this embodiment, the data acquisition module 1 is equipped with a clinical data interface, and after authorization, it securely connects to the hospital's electronic medical record (EMR) to obtain data such as treatment plans, laboratory test results (such as inflammatory markers), and imaging assessments.

[0045] Data acquisition module 1 is equipped with a subjective data input interface, guiding patients to report daily symptom intensity (e.g., VAS / NRS scores), emotional state (e.g., PHQ-9, GAD-7), and sleep quality (PSQI) via a mobile app or mini-program. Gamification and voice input methods are used to improve compliance.

[0046] The data acquisition module 1 continuously monitors physiological parameters such as heart rate variability (HRV), electrical skin activity (EDA), body surface temperature, sleep structure (deep sleep, light sleep, REM), and daytime activity level through wearable devices, such as smart bracelets and watches.

[0047] Data acquisition module 1 uses home IoT devices such as smart scales, smart pillboxes (to record medication adherence), and environmental sensors (to monitor indoor temperature, humidity, and light).

[0048] The feature extraction module 2 is used to acquire data from the data acquisition module 1, and the feature extraction module 2 extracts features from patient information and environmental information to obtain time-domain features, frequency-domain features and nonlinear features corresponding to symptoms.

[0049] In this embodiment, the feature extraction module 2 collects patient information and environmental information, and statistically analyzes medication information, emotional information, indoor humidity and indoor temperature corresponding to the patient's symptoms on the time axis to obtain time-domain features. The feature extraction module 2 transforms the patient information and environmental information and decomposes them into different frequency components to obtain frequency-domain features of symptom rhythm. The feature extraction module 2 extracts entropy features from the patient information and environmental information, and analyzes symptom fluctuations, post-medication symptom conditions, and the correlation between emotional fluctuations and symptoms to obtain nonlinear features.

[0050] In this embodiment, time-domain features include baseline symptom intensity, symptom variability, intensity of acute symptom attacks, and rate of symptom progression. Time-domain features also include periodic symptoms such as sleep disturbances and mood swings. Nonlinear features include the dynamic complexity of symptoms, such as post-chemotherapy neurotoxicity, symptom self-similarity, such as chronic fatigue, and the degree of symptom confusion, such as respiratory rhythm disturbances.

[0051] In this embodiment, the data acquisition module 1 can comprehensively and accurately obtain the patient's medical record information, symptom information, medication information, emotional information, and basic information through intelligent devices. At the same time, indoor humidity and indoor temperature are used as one of the feature data, which can improve the accuracy of subsequent assessment of symptom cluster collaborative management. The feature extraction module 2 extracts time-domain features, frequency-domain features, and nonlinear features related to symptoms from multimodal data, reducing the identification cost of subsequent core symptom clusters.

[0052] The assessment platform 3 is used for data acquisition by the feature extraction module 2. The assessment platform 3 identifies the patient's core symptom clusters based on time-domain features, frequency-domain features, and nonlinear features. The assessment platform 3 calculates the total burden index based on the core symptom clusters. The assessment platform 3 analyzes the core symptom clusters based on time series data to provide early warning of symptom deterioration.

[0053] The assessment platform 3 includes a data processing module 31, an identification module 32, and an early warning module 33.

[0054] The data processing module 31 acquires data from the feature extraction module 2, and sets different weights for time-domain features, frequency-domain features, and nonlinear features based on medical record information and basic information. The data processing module 31 then fuses the time-domain features, frequency-domain features, and nonlinear features based on the weights to obtain a feature data matrix.

[0055] Data processing module 31 sequentially performs feature alignment, feature standardization, and feature dimensionality reduction on time-domain features, frequency-domain features, and nonlinear features, and then fuses them into a feature data matrix:

[0056] Formula (1),

[0057] in, For the first Attention weights for each feature; To convert the scores into a probability distribution; To calculate the attention score; This is the attention weight vector; To introduce nonlinearity through the tanh activation function; This is the weight matrix; For the first Feature vectors of each modality; This is the feature data matrix.

[0058] The data processing module 31 fuses time-domain features, frequency-domain features, and nonlinear features based on weights, enabling the features to be represented more comprehensively and improving the response of important features.

[0059] The identification module 32 analyzes the feature data matrix using machine learning algorithms to identify the core symptom clusters, and the identification module 32 performs standardized severity scoring on the core symptom clusters to calculate the total burden index.

[0060] The identification module 32 analyzes the feature data matrix using machine learning algorithms, outputs classification labels to obtain the core symptom clusters, and calculates the total burden index by predicting the probability of subjective symptom intensity and multimodal features.

[0061] ,

[0062] in, Total load index; The number of core symptoms; Symptom weights; Standardized symptom scoring; For characteristic correction coefficients; Predict the probability of symptom clusters.

[0063] The identification module 32 uses machine learning algorithms to process the feature data matrix, which can efficiently analyze large-scale symptom data, reveal symptom association patterns that are difficult for humans to detect, and continuously optimize symptom cluster identification as data accumulates, achieving dynamic updates.

[0064] The early warning module 33 analyzes the core symptom clusters and total burden index based on time series data to obtain early warning signals.

[0065] The early warning module 33 constructs an LSTM prediction model and analyzes the core symptom cluster and total load index through change point detection to obtain the mutation point of the total load index. The early warning module 33 predicts the time of occurrence of the next mutation point based on the duration of two adjacent mutation points. When the occurrence time of the next mutation point is less than the preset duration, the early warning module 33 issues an early warning signal.

[0066] The early warning module 33 obtains the duration of two adjacent mutation points based on the core symptom cluster and the total burden index, thereby being able to predict the time of the next mutation point and predict the worsening trend of the symptom cluster.

[0067] The collaborative platform 4 is used for storing the disease knowledge base. The collaborative platform 4 acquires data from the feature extraction module 2 and the evaluation platform 3 to identify and obtain corresponding multimodal intervention plans based on the patient's condition.

[0068] The collaborative platform 4 includes a knowledge base module 41, an annotation module 42, a treatment recommendation module 43, an adjustment module 44, and a verification module 45.

[0069] The knowledge base module 41 is used to store disease knowledge bases, which include knowledge of oncology, pharmacy, nursing, nutrition, and psychology. The disease knowledge base also stores historical cases and corresponding treatment plans.

[0070] The annotation module 42 is used to annotate the data of the knowledge base module 41 to annotate the synergistic effects and contraindications of different interventions on multiple symptoms, and the annotation module 42 constructs a knowledge graph based on the annotation information.

[0071] The disease knowledge base is constructed by incorporating knowledge from oncology, pharmacy, nursing, nutrition, and psychology, and also stores historical cases and corresponding treatment plans. At the same time, the synergistic effects and contraindications of different intervention measures on multiple symptoms are labeled, thereby providing accurate data for the treatment recommendation module 43 and improving the accuracy of multimodal intervention plans.

[0072] The treatment recommendation module 43 is used for data acquisition from the feature extraction module 2, the knowledge base module 41, and the annotation module 42. The treatment recommendation module 43 constructs a personal feature map of the patient based on the feature extraction module 2, and the treatment recommendation module 43 matches the personal feature map with the knowledge map through text recognition to obtain a multimodal intervention plan.

[0073] In this embodiment, the treatment recommendation module 43 obtains multiple candidate solutions. The treatment recommendation module 43 assigns corresponding matching weights to each piece of patient information, and based on these matching weights, it performs similarity matching between the candidate solutions and the corresponding historical cases of the patients. The treatment recommendation module 43 selects the candidate solution with the highest similarity as the multimodal intervention solution. The treatment recommendation module 43 can match candidate solutions based on patient information, making the multimodal intervention solution more suitable for the patient and improving treatment effectiveness.

[0074] The adjustment module 44 is used for data acquisition by the treatment recommendation module 43, and it analyzes the sub-plans in the multimodal intervention program to prioritize them according to the severity of the patient's symptoms. By prioritizing the sub-plans in the multimodal intervention program through the adjustment module 44, the most effective and urgent measures are recommended to be implemented first, effectively reducing the patient's suffering.

[0075] In this embodiment, the multimodal intervention plan is not a single drug, but a combination of measures. For example, for the "pain-fatigue-sleep disorder" group, it is recommended to "a light aerobic exercise plan (15 minutes of walking) + mindfulness meditation audio (20 minutes before bedtime) + optimize the timing of painkiller administration"; for the "nausea-vomiting-anorexia" group, it is recommended to "ginger dietary therapy suggestions + acupressure guidance video + adjust the dosage of antiemetics + small and frequent meals plan".

[0076] The verification module 45 sends data from the treatment recommendation module 43 and the adjustment module 44 to the healthcare console for review of the multimodal intervention plan. The verification module 45 prevents system errors from affecting patient treatment, and final manual review ensures treatment safety. Simultaneously, the verification module 45 can provide remote guidance or arrange in-person appointments.

[0077] This embodiment can also include a medical staff terminal, which can be a computer or a smart mobile terminal. The medical staff terminal is used to acquire data from the data acquisition module 1, feature extraction module 2, evaluation platform 3, and collaboration platform 4, thereby enabling medical staff to obtain the patient's original data and promptly identify the patient's problems when there are doubts about the multimodal intervention plan.

[0078] Patient terminal 5 is used to acquire data from assessment platform 3 and collaboration platform 4 to push multimodal intervention plans to the patient. In this embodiment, the multimodal intervention plan is pushed to the patient after being reviewed by verification module 45.

[0079] This embodiment constructs an intelligent closed-loop system of monitoring, assessment, intervention, and feedback through data acquisition module 1, feature extraction module 2, assessment platform 3, collaboration platform 4, and patient terminal 5. This system manages symptom clusters as a whole, revealing the intrinsic connections between symptoms. Furthermore, it shifts from passive response to proactive prediction and early warning, preventing problems before they arise. The data used combines subjective reports with objective physiological data, making the assessment results more comprehensive and accurate. Simultaneously, this embodiment provides each patient with a "tailor-made" dynamic management plan, improving treatment efficacy and quality of life. For medical staff, it reduces the burden on healthcare workers and optimizes the allocation of medical resources.

Claims

1. A collaborative management system for symptom clusters in cancer patients, characterized in that, It includes a data acquisition module (1), a feature extraction module (2), an assessment platform (3), a collaboration platform (4), and a patient terminal (5). The data acquisition module (1) is used to acquire patient information and environmental information through smart devices. The patient information includes medical record information, symptom information, medication information, emotional information, and basic information. The environmental information includes indoor humidity and indoor temperature. The feature extraction module (2) is used for data acquisition by the data acquisition module (1), and the feature extraction module (2) performs feature extraction on the patient information and the environmental information to obtain time-domain features, frequency-domain features and nonlinear features corresponding to the symptoms; The assessment platform (3) is used for data acquisition by the feature extraction module (2), and the assessment platform (3) identifies the patient's core symptom clusters based on the time domain features, the frequency domain features and the nonlinear features, and the assessment platform (3) calculates the total burden index based on the core symptom clusters; the assessment platform (3) analyzes the core symptom clusters based on time series to provide early warning of symptom deterioration; The collaborative platform (4) is used for storing the disease knowledge base. The collaborative platform (4) obtains data from the feature extraction module (2) and the evaluation platform (3) to identify and obtain the corresponding multimodal intervention plan according to the patient's condition. The patient terminal (5) is used to acquire data from the assessment platform (3) and the collaboration platform (4) to push the multimodal intervention plan to the patient.

2. The collaborative management system for symptom clusters of cancer patients according to claim 1, characterized in that: The feature extraction module (2) collects the medication information, emotional information, indoor humidity and indoor temperature corresponding to the patient's symptoms on the time axis in the patient information and the environmental information to obtain time-domain features; The feature extraction module (2) transforms the patient information and the environmental information and decomposes them into different frequency components to obtain the frequency domain features of symptom rhythm. The feature extraction module (2) extracts entropy features from the patient information and the environmental information, analyzes symptom fluctuations, post-medication symptom conditions, and the correlation between emotional fluctuations and symptoms to obtain nonlinear features.

3. The collaborative management system for symptom clusters of tumor patients according to claim 1, characterized in that: The evaluation platform (3) includes a data processing module (31), an identification module (32), and an early warning module (33). The data processing module (31) acquires the data from the feature extraction module (2), and the data processing module (31) sets different weights for the time domain features, the frequency domain features and the nonlinear features based on the medical record information and the basic information, and the data processing module (31) fuses the time domain features, the frequency domain features and the nonlinear features based on the weights to obtain a feature data matrix; The identification module (32) analyzes the feature data matrix through machine learning algorithms to identify the core symptom clusters, and the identification module (32) performs standardized severity scoring on the core symptom clusters to calculate the total burden index. The early warning module (33) analyzes the core symptom cluster and the total load index based on time series to obtain early warning signals.

4. The collaborative management system for symptom clusters of tumor patients according to claim 3, characterized in that: The data processing module (31) sequentially performs feature alignment, feature standardization, and feature dimensionality reduction on the time-domain features, frequency-domain features, and nonlinear features, and then fuses them into a feature data matrix: Official (1), in, For the first Attention weights for each feature; To convert the scores into a probability distribution; To calculate attention scores; This is the attention weight vector; To introduce nonlinearity through the tanh activation function; This is the weight matrix; For the first Feature vectors of each modality; This is the feature data matrix.

5. The collaborative management system for symptom clusters of cancer patients according to claim 4, characterized in that: The identification module (32) analyzes the feature data matrix using a machine learning algorithm, outputs classification labels to obtain the core symptom cluster, and calculates the total burden index by predicting the subjective symptom intensity and multimodal features. Official (2), in, Total load index; The number of core symptoms; Symptom weights; Standardized symptom scoring; For characteristic correction coefficients; Predict the probability of symptom clusters.

6. The collaborative management system for symptom clusters of tumor patients according to claim 3, characterized in that: The early warning module (33) constructs an LSTM prediction model and analyzes the core symptom cluster and the total load index through change point detection to obtain the mutation point of the total load index. The early warning module (33) predicts the time of the next mutation point based on the duration of two adjacent mutation points. When the time of the next mutation point is less than the preset duration, the early warning module (33) issues an early warning signal.

7. The collaborative management system for symptom clusters of tumor patients according to claim 1, characterized in that: The collaborative platform (4) includes a knowledge base module (41), an annotation module (42), and a treatment recommendation module (43). The knowledge base module (41) is used for storing disease knowledge bases, which include knowledge of oncology, pharmacy, nursing, nutrition, and psychology, and also stores historical cases and corresponding treatment plans. The annotation module (42) is used to annotate the data of the knowledge base module (41) to annotate the synergistic effects and contraindications of different intervention measures on multiple symptoms, and the annotation module (42) constructs a knowledge graph based on the annotation information; The treatment recommendation module (43) is used for data acquisition of the feature extraction module (2), the knowledge base module (41) and the annotation module (42). The treatment recommendation module (43) constructs a personal feature map of the patient based on the feature extraction module (2), and the treatment recommendation module (43) matches the personal feature map with the knowledge map through text recognition to obtain a multi-modal intervention plan.

8. The collaborative management system for symptom clusters of tumor patients according to claim 7, characterized in that: The treatment recommendation module (43) matches and obtains multiple candidate solutions. The treatment recommendation module (43) sets corresponding matching weights for each piece of information in the patient information. Based on the matching weights, the treatment recommendation module (43) performs similarity matching between the historical cases corresponding to the candidate solutions and the patient information. The treatment recommendation module (43) obtains the candidate solution with the highest similarity as the multimodal intervention solution.

9. A collaborative management system for symptom clusters in cancer patients according to claim 7, characterized in that: The aforementioned collaborative platform (4) further includes an adjustment module (44), which is used for data acquisition by the treatment recommendation module (43), and the adjustment module (44) analyzes the sub-schemes in the multimodal intervention program to prioritize the sub-schemes in the multimodal intervention program according to the severity of the patient's symptoms.

10. A collaborative management system for symptom clusters in cancer patients according to claim 7, characterized in that: The aforementioned collaborative platform (4) also includes a verification module (45), which is used to send the data of the treatment recommendation module (43) and the adjustment module (44) to the medical staff terminal console to review the multi-mode intervention plan.

Images