Endocrine disease data processing system and method based on AI intelligence

Abstract

The invention relates to the field of endocrine disease treatment, in particular to an endocrine disease data processing system and method based on AI intelligence. The endocrine disease data processing system based on AI intelligence comprises a treatment scheme recommendation module, a treatment effect evaluation module and a treatment measure decision module. The AI technology is integrated, an initial treatment scheme suitable for the patient is obtained according to clinical data and genome data of the patient, the treatment effect of the patient is evaluated in the actual treatment process, and the treatment scheme is dynamically adjusted in combination with monitoring of the psychological state of the patient; according to the method, personalized treatment schemes can be provided for the patients with the endocrine diseases, the treatment accuracy is improved, unnecessary medicine side effects can be reduced, and therefore the treatment effect and the life quality of the patients are improved.

Description

Technical Field

[0001] The present invention relates to the field of endocrine disease treatment, and in particular to an endocrine disease data processing system and method based on AI intelligence. Background Art

[0002] With the rapid development of artificial intelligence (AI) technology, its application in the field of medical health is becoming more and more extensive, especially in the diagnosis and treatment of endocrine diseases; the endocrine system is responsible for regulating many important functions of the body, including growth, metabolism, blood pressure and body temperature; the occurrence of endocrine diseases is often related to abnormal hormone levels, and these diseases include diabetes, thyroid disease, adrenal disease, etc.; these diseases not only affect the patient's physical health, but also seriously affect the patient's quality of life.

[0003] In the traditional treatment of endocrine diseases, doctors mainly formulate treatment plans based on the patient's clinical data and laboratory test results; however, more and more studies have shown that the patient's psychological state, including emotions and stress levels, has a significant impact on the occurrence, development and treatment effects of endocrine diseases; psychological factors such as anxiety, depression and stress can affect the patient's hormone secretion, thereby affecting the course of the disease and treatment response; for example, in diabetic patients, long-term stress and negative emotions are associated with poor blood sugar control; psychological stress can increase the secretion of cortisol and lead to insulin resistance, thereby affecting blood sugar levels; in thyroid diseases, anxiety and depression are related to hyperthyroidism and hypothyroidism; psychological stress can affect the secretion of thyroid hormones and lead to thyroid dysfunction.

[0004] Given the important impact of psychological factors on endocrine diseases, traditional treatment plans need to be combined with psychological intervention and evaluation to improve treatment outcomes; however, this comprehensive treatment requires a lot of medical resources and expertise, and is difficult to standardize and scale in practice; therefore, developing an AI-based endocrine disease data processing system and method that can automatically collect and analyze patients’ clinical data, genomic data, and psychological states, and provide patients with personalized treatment plans and psychological intervention measures, has important clinical significance and social value. Summary of the invention

[0005] The present invention integrates AI technology to obtain an initial treatment plan suitable for the patient based on the patient's clinical data and genomic data, and evaluates the patient's treatment effect during the actual treatment process, and dynamically adjusts the treatment plan in combination with the monitoring of the patient's psychological state. This method can provide personalized treatment plans for patients with endocrine diseases, which not only improves the accuracy of treatment, but also reduces unnecessary drug side effects, thereby improving the treatment effect and the patient's quality of life.

[0006] The technical solution of the present invention is as follows: Endocrine disease data processing methods based on AI intelligence include: Obtain and preprocess the patient's clinical data and genomic data. The clinical data includes the patient's basic information, medical history, and laboratory test data. The preprocessed clinical data and genomic data are used as inputs to the treatment recommendation model to output the patient's initial treatment plan. The initial treatment plan includes the type of drug, drug dosage, and treatment cycle required for the patient's treatment. The patient applies the initial treatment plan to treat the disease. For any monitoring time point during the patient's treatment cycle, the patient's target detection index is obtained. The target detection index includes several indicators that need to be detected corresponding to the patient's disease. The detection index score is calculated based on the patient's target detection index; at the same time, the patient's self-reported state score and negative emotion score are obtained. Based on the patient's detection index score, self-reported state score and negative emotion score, the treatment effect score at the current monitoring time point is calculated; An initial threshold for the effect score is set. For any monitoring time point, the initial threshold for the effect score is adjusted based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold. It is determined whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold. If so, no operation is performed. If not, the initial treatment plan is adjusted. At the same time, an emotion score threshold is set to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold. If so, no operation is performed. If not, psychological intervention measures are taken for the patient.

[0007] Preferably, the detection index score is obtained based on the patient's target detection index calculation, and the specific operation is as follows: For any one of the n indicators that need to be tested corresponding to the patient's disease in the target detection indicator, based on the normal control range corresponding to the indicator to , if the indicator is within the corresponding normal control range to If the score of this indicator is within is 10, i=1, 2, ..., n; if the indicator exceeds the corresponding normal control range to , then the calculation obtains the excess value , and using the formula Calculate the score of this indicator , among which, when When the calculated result is less than 0, the score of this indicator The value is 0; Finally, using the formula Calculate and obtain the detection index score .

[0008] Preferably, the patient's self-reported state score and negative emotion score are obtained, and based on the patient's detection index score, self-reported state score and negative emotion score, the treatment effect score at the current monitoring time point is calculated and obtained. The specific operations are as follows: Based on the patient's self-perception of the current monitoring time point and the previous monitoring time point, the patient scores the change in his or her own status to obtain the self-reported status score , the score is within 0-10; Patients were evaluated using the anxiety self-rating scale, depression self-rating scale and psychological stress scale to obtain anxiety scores. , depression score and stress score ; Using the formula Calculate the patient's negative emotion score ; Using the formula Calculate the treatment effect score ,in, and Scoring of detection indicators and self-reported status scores The initial weight of ,and > .

[0009] Preferably, the initial threshold of the effect score is adjusted based on the negative emotion score of the patient obtained at the current monitoring time point to obtain the effect score threshold. The specific operation is as follows: Based on the preset initial threshold of effect score and negative sentiment scores , using the formula Calculate the threshold value for obtaining the effect score ;in, is the adjustment coefficient used to control the initial threshold of the effect score The proportion of decline, Adjustment is done through a swarm optimization algorithm.

[0010] Preferably, the treatment plan recommendation model is established based on a multimodal neural network, including an input layer, a feature extraction layer, a fusion layer, a fully connected layer and an output layer; the input layer is used to receive preprocessed clinical data and genomic data; the feature extraction layer is used to extract features of clinical data and genomic data respectively, and obtain clinical feature representation and genomic feature representation; the fusion layer is used to merge the clinical feature representation and genomic feature representation output by the feature extraction layer into a fused feature vector; the fully connected layer is used to extract deeper nonlinear relationships in the fused feature vector to obtain a comprehensive feature representation; the output layer is used to output an initial treatment plan, which includes the type of drugs, drug dosage and treatment cycle required for the patient's treatment, wherein, for drug type prediction, a softmax activation function is used for classification, and the probability of each drug type is output; for drug dosage prediction, a linear activation function is used for regression, and the specific drug dosage is output; for treatment cycle prediction, a softmax activation function is used for classification, and the probabilities of short-term, medium-term and long-term are output respectively.

[0011] Preferably, the specific operations for training the treatment plan recommendation model are as follows: The clinical data, genomic data and historical treatment data of all endocrine disease patients are extracted from the historical medical database. The historical treatment data includes the patient's treatment cycle and the type and dosage of drugs used during the treatment cycle; the pre-treated clinical data and genomic data of any endocrine disease patient are used as a regimen recommendation training sample, and the historical treatment data of the endocrine disease patient is used as a label of the regimen recommendation training sample to obtain a number of regimen recommendation training samples; all regimen recommendation training samples are divided into a regimen recommendation training set and a regimen recommendation verification set, the regimen recommendation training set is used to train the treatment regimen recommendation model, and the treatment regimen recommendation model is verified by the regimen recommendation verification set to obtain the verification result; the training conditions are set to determine whether the obtained verification result meets the training conditions, and if so, the trained treatment regimen recommendation model is output; if not, the regimen recommendation training set is continued to be used to train the treatment regimen recommendation model.

[0012] Preferably, adjust The population optimization algorithm used is the simulated annealing algorithm, and the specific operations are as follows: Step 1: Randomly select an initial value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution ,in, Indicates the current initialization of the application The treatment effect of the patient during the treatment cycle when the initial threshold of the effect score is adjusted; Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where represents the fitness value of the neighborhood solution, represents the fitness value of the current solution, Indicates the current temperature; using the formula Update the temperature and get the temperature for the next iteration , where k is the temperature attenuation coefficient; Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal one. value.

[0013] Preferably, the AI-based endocrine disease data processing system includes: The treatment plan recommendation module includes a patient data acquisition unit and a treatment plan recommendation unit; the patient data acquisition unit is used to acquire the patient's clinical data and genomic data and perform preprocessing; the treatment plan recommendation unit is used to use the preprocessed clinical data and genomic data as input to the treatment plan recommendation model and output the patient's initial treatment plan; The treatment effect evaluation module includes a patient data monitoring unit and a treatment effect score calculation unit; the patient data monitoring unit is used to obtain the patient's target detection index at any monitoring time point during the patient's treatment cycle, and simultaneously obtain the patient's self-reported state score and negative emotion score; the treatment effect score calculation unit is used to calculate and obtain the detection index score based on the patient's target detection index, and calculate and obtain the treatment effect score at the current monitoring time point based on the patient's detection index score, self-reported state score and negative emotion score; The treatment measure decision module is used to adjust the initial threshold of the effect score at any monitoring time point based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold; determine whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold; if so, no operation is performed; if not, the initial treatment plan is adjusted; and at the same time, the emotion score threshold is set to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold; if so, no operation is performed; if not, psychological intervention measures are taken for the patient.

[0014] The present invention has the following advantages: The present invention integrates AI technology to obtain an initial treatment plan suitable for the patient based on the patient's clinical data and genomic data, and evaluates the patient's treatment effect during the actual treatment process, and dynamically adjusts the treatment plan in combination with the monitoring of the patient's psychological state. This method can provide personalized treatment plans for patients with endocrine diseases, which not only improves the accuracy of treatment, but also reduces unnecessary drug side effects, thereby improving the treatment effect and the patient's quality of life.

[0015] By timely identifying and intervening in patients' psychological problems, the system helps to reduce patients' psychological stress and enhance treatment compliance, thereby improving the overall treatment effect; it provides a new perspective and tool for the comprehensive management of endocrine diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a schematic diagram of the structure of the AI-based endocrine disease data processing system used in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0018] Embodiment 1, an endocrine disease data processing method based on AI intelligence, comprising: Obtain the patient's clinical data and genomic data and preprocess them. Different data sources may have different records and storage formats. The preprocessing step ensures that all data can be converted into a unified format so that the model can read and process them correctly. Clinical data includes basic information of the patient (such as age, gender, weight, etc.), medical history (including previous medical history, drug allergy history, etc.) and laboratory test data (such as blood tests, urine tests, hormone level test results, etc.). Genomic data involves the patient's genetic information, including gene variation, gene expression patterns, etc. Genomic data helps to understand the patient's response to specific drugs and disease susceptibility, and provides support for personalized medicine. The collection of these data is to establish a comprehensive patient information file so that doctors can better understand the patient's health status and disease characteristics, thereby providing patients with more accurate treatment recommendations. The preprocessed clinical data and genomic data are used as inputs to the treatment recommendation model to output the patient's initial treatment plan, which includes the type of drug, drug dosage and treatment cycle required for the patient's treatment. The output initial treatment plan provides patients with a scientific and personalized treatment starting point, which helps to improve treatment effects and patient satisfaction, while reducing the waste of medical resources. The patient applies the initial treatment plan to treat the disease. At any monitoring time point during the patient's treatment cycle, the patient's target detection index is obtained. The target detection index includes several indicators that need to be detected corresponding to the patient's disease. These detection indicators are set for the patient's specific disease and are key physiological or biochemical parameters for evaluating the treatment effect. For example, for diabetic patients, the indicators that need to be detected include fasting blood glucose, postprandial blood glucose, glycosylated hemoglobin, insulin levels, etc.; for patients with thyroid diseases, the indicators that need to be detected include thyroid-stimulating hormone, free thyroxine, etc.; for patients with adrenal diseases, the indicators that need to be detected include cortisol levels, epinephrine levels, norepinephrine levels, etc.; the detection index score is calculated based on the patient's target detection index. This score reflects the degree of deviation of each indicator from the normal level and is a treatment The patient's self-reported state score and negative emotion score are also obtained. The patient's self-reported state score provides direct feedback on the patient's health status and treatment experience. This subjective evaluation can capture changes that may not be reflected in objective physiological indicators, such as pain, fatigue, sleep quality, etc. Based on the patient's test indicator score, self-reported state score and negative emotion score, the treatment effect score at the current monitoring time point is calculated. This score combines objective physiological data and the patient's subjective feelings to provide a comprehensive treatment effect evaluation. The treatment effect score enables doctors to have a more comprehensive understanding of the treatment effect, including not only the improvement of physiological indicators, but also the patient's psychological state and personal feelings. This comprehensive evaluation helps doctors make more accurate adjustments in subsequent treatments to improve treatment effects and patients' treatment experience. Set the initial threshold of the effect score. The initial threshold of the effect score is a pre-set standard used to determine whether the treatment effect has reached the expected goal. This threshold is set based on historical data, clinical guidelines and professional judgment, and is used to evaluate the patient's treatment effect score. For any monitoring time point, adjust the initial threshold of the effect score based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold. Since negative emotions may affect the treatment effect, the negative emotion score of the patient obtained at the monitoring time point is used to adjust the initial threshold of the effect score. This adjustment takes into account the impact of psychological factors on the treatment effect, making the threshold more personalized and flexible; determine whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold; if so, no operation; if not, adjust the initial treatment plan; at the same time, set the emotion score threshold to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold. If so, no operation; if not, psychological intervention measures are taken for the patient. This judgment process ensures that the patient's psychological state is properly cared for and managed; if the patient's negative emotion score is higher than the threshold, timely psychological intervention can help the patient better manage emotions, which may improve the treatment effect and the patient's quality of life.

[0019] The detection index score is calculated based on the patient's target detection index. The specific operations are as follows: For any one of the n indicators that need to be tested corresponding to the patient's disease in the target detection indicator, based on the normal control range corresponding to the indicator to , if the indicator is within the corresponding normal control range to If the score of this indicator is within is 10, i=1, 2, ..., n; if the indicator exceeds the corresponding normal control range to , then the calculation obtains the excess value , and using the formula Calculate the score of this indicator , among which, when When the calculated result is less than 0, the score of this indicator The value is 0; Finally, using the formula Calculate and obtain the detection index score .

[0020] Obtain the patient's self-reported status score and negative emotion score. Based on the patient's test index score, self-reported status score and negative emotion score, calculate the treatment effect score at the current monitoring time point. The specific operations are as follows: Based on the patient's self-perception of the current monitoring time point and the previous monitoring time point, the patient scores the change in his or her own status to obtain the self-reported status score , the score is within 0-10; Patients were evaluated using the anxiety self-rating scale, depression self-rating scale and psychological stress scale to obtain anxiety scores. , depression score and stress score ; Using the formula Calculate the patient's negative emotion score ; Using the formula Calculate the treatment effect score ,in, and Scoring of detection indicators and self-reported status scores The initial weight of ,and > .

[0021] Based on the negative emotion score of the patient obtained at the current monitoring time point, the initial threshold of the effect score is adjusted to obtain the effect score threshold. The specific operations are as follows: Based on the preset initial threshold of effect score and negative sentiment scores , using the formula Calculate the threshold value for obtaining the effect score ;in, is the adjustment coefficient used to control the initial threshold of the effect score The proportion of decline, Adjustment is done through a swarm optimization algorithm.

[0022] The treatment recommendation model is based on a multimodal neural network, which is an advanced deep learning model that can process and integrate information from a variety of different data types (such as text, images, audio, and sensor data) by fusing these data at different levels within the network, extracting and utilizing cross-modal features to improve the performance of tasks such as classification, detection, prediction, or generation; The treatment plan recommendation model includes an input layer, a feature extraction layer, a fusion layer, a fully connected layer and an output layer; the input layer is used to receive preprocessed clinical data and genomic data; the feature extraction layer is used to extract the features of clinical data and genomic data respectively, and obtain clinical feature representation and genomic feature representation; the fusion layer is used to merge the clinical feature representation and genomic feature representation output by the feature extraction layer into a fused feature vector; the fully connected layer is used to extract deeper nonlinear relationships in the fused feature vector to obtain a comprehensive feature representation; the output layer is used to output the initial treatment plan, which includes the type of drugs, drug dosage and treatment cycle required for the patient's treatment. For drug type prediction, the softmax activation function is used for classification and the probability of each drug type is output. For drug dosage prediction, the linear activation function is used for regression and the specific drug dosage is output. For treatment cycle prediction, the softmax activation function is used for classification and the probabilities of short-term, medium-term and long-term are output respectively.

[0023] The specific operations for training the treatment plan recommendation model are as follows: The clinical data, genomic data and historical treatment data of all endocrine disease patients are extracted from the historical medical database. The historical treatment data includes the patient's treatment cycle and the type and dosage of drugs used during the treatment cycle; the pre-treated clinical data and genomic data of any endocrine disease patient are used as a regimen recommendation training sample, and the historical treatment data of the endocrine disease patient is used as a label of the regimen recommendation training sample to obtain a number of regimen recommendation training samples; all regimen recommendation training samples are divided into a regimen recommendation training set and a regimen recommendation verification set, the regimen recommendation training set is used to train the treatment regimen recommendation model, and the treatment regimen recommendation model is verified by the regimen recommendation verification set to obtain the verification result; the training conditions are set to determine whether the obtained verification result meets the training conditions, and if so, the trained treatment regimen recommendation model is output; if not, the regimen recommendation training set is continued to be used to train the treatment regimen recommendation model.

[0024] Adjustment The swarm optimization algorithm used is the simulated annealing algorithm. The simulated annealing algorithm is a probabilistic global optimization algorithm inspired by the solid annealing process. It simulates the movement of particles in the metal annealing process to find the optimal solution to the problem. It allows a certain probability of accepting a poor solution during the search process, which helps to jump out of the local optimal solution and continue to explore a wider solution space. As the "temperature" parameter gradually decreases, the probability of accepting a poor solution also decreases, and finally converges to the global optimal or approximate optimal solution. The specific operations are as follows: Step 1: Randomly select an initial value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution ,in, Indicates the current initialization of the application The treatment effect of the patient during the treatment cycle when the initial threshold of the effect score is adjusted; Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where represents the fitness value of the neighborhood solution, represents the fitness value of the current solution, Indicates the current temperature; using the formula Update the temperature and get the temperature for the next iteration , where k is the temperature attenuation coefficient; Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal one. value.

[0025] Example 2, an endocrine disease data processing system based on AI intelligence, such as Figure 1 As shown, including: The treatment plan recommendation module includes a patient data acquisition unit and a treatment plan recommendation unit; the patient data acquisition unit is used to acquire the patient's clinical data and genomic data and perform preprocessing; the treatment plan recommendation unit is used to use the preprocessed clinical data and genomic data as input to the treatment plan recommendation model and output the patient's initial treatment plan; The treatment effect evaluation module includes a patient data monitoring unit and a treatment effect score calculation unit; the patient data monitoring unit is used to obtain the patient's target detection index at any monitoring time point during the patient's treatment cycle, and simultaneously obtain the patient's self-reported state score and negative emotion score; the treatment effect score calculation unit is used to calculate and obtain the detection index score based on the patient's target detection index, and calculate and obtain the treatment effect score at the current monitoring time point based on the patient's detection index score, self-reported state score and negative emotion score; The treatment measure decision module is used to adjust the initial threshold of the effect score at any monitoring time point based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold; determine whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold; if so, no operation is performed; if not, the initial treatment plan is adjusted; and at the same time, the emotion score threshold is set to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold; if so, no operation is performed; if not, psychological intervention measures are taken for the patient.

[0026] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. An AI-based endocrine disease data processing method, characterized in that: include: Obtain and preprocess the patient's clinical data and genomic data. The clinical data includes the patient's basic information, medical history, and laboratory test data. The preprocessed clinical data and genomic data are used as inputs to the treatment recommendation model to output the patient's initial treatment plan. The initial treatment plan includes the type of drug, drug dosage, and treatment cycle required for the patient's treatment. The patient applies the initial treatment plan to treat the disease. For any monitoring time point during the patient's treatment cycle, the patient's target detection index is obtained. The target detection index includes several indicators that need to be detected corresponding to the patient's disease. The detection index score is calculated based on the patient's target detection index; at the same time, the patient's self-reported state score and negative emotion score are obtained. Based on the patient's detection index score, self-reported state score and negative emotion score, the treatment effect score at the current monitoring time point is calculated; An initial threshold for the effect score is set. For any monitoring time point, the initial threshold for the effect score is adjusted based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold. It is determined whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold. If so, no operation is performed. If not, the initial treatment plan is adjusted. At the same time, an emotion score threshold is set to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold. If so, no operation is performed. If not, psychological intervention measures are taken for the patient.

2. The AI-based endocrine disease data processing method according to claim 1, characterized in that: The detection index score is calculated based on the patient's target detection index. The specific operations are as follows: For any one of the n indicators that need to be tested corresponding to the patient's disease in the target detection indicator, based on the normal control range corresponding to the indicator to , if the indicator is within the corresponding normal control range to If the score of this indicator is within is 10, i=1, 2, ..., n; if the indicator exceeds the corresponding normal control range to , then the calculation obtains the excess value , and using the formula Calculate the score of this indicator , among which, when When the calculated result is less than 0, the score of this indicator The value is 0; Finally, using the formula Calculate and obtain the detection index score .

3. The AI-based endocrine disease data processing method according to claim 2, characterized in that: Obtain the patient's self-reported status score and negative emotion score. Based on the patient's test index score, self-reported status score and negative emotion score, calculate the treatment effect score at the current monitoring time point. The specific operations are as follows: Based on the patient's self-perception of the current monitoring time point and the previous monitoring time point, the patient scores the change in his or her own status to obtain the self-reported status score , the score is within 0-10; Patients were evaluated using the anxiety self-rating scale, depression self-rating scale and psychological stress scale to obtain anxiety scores. , depression score and stress score ; Using the formula Calculate the patient's negative emotion score ; Using the formula Calculate the treatment effect score ,in, and Scoring of detection indicators and self-reported status scores The initial weight of ,and > .

4. The AI-based endocrine disease data processing method according to claim 3 is characterized in that: Based on the negative emotion score of the patient obtained at the current monitoring time point, the initial threshold of the effect score is adjusted to obtain the effect score threshold. The specific operations are as follows: Based on the preset initial threshold of effect score and negative sentiment scores , using the formula Calculate the threshold value for obtaining the effect score ;in, is the adjustment coefficient used to control the initial threshold of the effect score The proportion of decline, Adjustment is done through a swarm optimization algorithm.

5. The AI-based endocrine disease data processing method according to claim 4 is characterized in that: The treatment plan recommendation model is established based on a multimodal neural network, including an input layer, a feature extraction layer, a fusion layer, a fully connected layer and an output layer; the input layer is used to receive preprocessed clinical data and genomic data; the feature extraction layer is used to extract the features of clinical data and genomic data respectively, and obtain clinical feature representation and genomic feature representation; the fusion layer is used to merge the clinical feature representation and genomic feature representation output by the feature extraction layer into a fused feature vector; the fully connected layer is used to extract deeper nonlinear relationships in the fused feature vector to obtain a comprehensive feature representation; the output layer is used to output the initial treatment plan, which includes the type of drugs, drug dosage and treatment cycle required for the patient's treatment. For drug type prediction, the softmax activation function is used for classification, and the probability of each drug type is output. For drug dosage prediction, the linear activation function is used for regression, and the specific drug dosage is output. For treatment cycle prediction, the softmax activation function is used for classification, and the probabilities of short-term, medium-term and long-term are output respectively.

6. The AI-based endocrine disease data processing method according to claim 5, characterized in that: The specific operations for training the treatment plan recommendation model are as follows: Extract clinical data, genomic data and historical treatment data of all patients with endocrine diseases from the historical medical database. The historical treatment data includes the patient's treatment cycle and the type and dosage of drugs used during the treatment cycle; use the pre-treated clinical data and genomic data of any endocrine disease patient to form a regimen recommendation training sample, and use the historical treatment data of the endocrine disease patient as the label of the regimen recommendation training sample to obtain several regimen recommendation training samples; divide all regimen recommendation training samples into a regimen recommendation training set and a regimen recommendation verification set, use the regimen recommendation training set to train the treatment regimen recommendation model, and use the regimen recommendation verification set to verify the treatment regimen recommendation model to obtain the verification results; Set the training conditions and determine whether the obtained verification results meet the training conditions. If so, output the trained treatment plan recommendation model; if not, continue to apply the treatment plan recommendation training set to train the treatment plan recommendation model.

7. The AI-based endocrine disease data processing method according to claim 6, characterized in that: Adjustment The population optimization algorithm used is the simulated annealing algorithm, and the specific operations are as follows: Step 1: Randomly select an initial value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution ,in, Indicates the current initialization of the application The treatment effect of the patient during the treatment cycle when the initial threshold of the effect score is adjusted; Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where represents the fitness value of the neighborhood solution, represents the fitness value of the current solution, Indicates the current temperature; using the formula Update the temperature and get the temperature for the next iteration , where k is the temperature attenuation coefficient; Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal one. value.

8. The AI-based endocrine disease data processing system is characterized by: The system is applied to the AI-based endocrine disease data processing method according to any one of claims 1 to 7, comprising: The treatment plan recommendation module includes a patient data acquisition unit and a treatment plan recommendation unit; the patient data acquisition unit is used to acquire the patient's clinical data and genomic data and perform preprocessing; the treatment plan recommendation unit is used to use the preprocessed clinical data and genomic data as input to the treatment plan recommendation model and output the patient's initial treatment plan; The treatment effect evaluation module includes a patient data monitoring unit and a treatment effect score calculation unit; the patient data monitoring unit is used to obtain the patient's target detection index at any monitoring time point during the patient's treatment cycle, and simultaneously obtain the patient's self-reported state score and negative emotion score; the treatment effect score calculation unit is used to calculate and obtain the detection index score based on the patient's target detection index, and calculate and obtain the treatment effect score at the current monitoring time point based on the patient's detection index score, self-reported state score and negative emotion score; The treatment measure decision module is used to adjust the initial threshold of the effect score at any monitoring time point based on the patient's negative emotion score obtained at the current monitoring time point to obtain the effect score threshold; determine whether the treatment effect score obtained at the current monitoring time point is higher than the effect score threshold; if so, no operation is performed; if not, the initial treatment plan is adjusted; and at the same time, the emotion score threshold is set to determine whether the patient's negative emotion score obtained at the current monitoring time point is lower than the emotion score threshold; if so, no operation is performed; if not, psychological intervention measures are taken for the patient.

Images