Medicine management method and system for old people based on AI auxiliary decision making

Abstract

The invention relates to the technical field of medical informatization, in particular to an old people medicine management method and system based on AI auxiliary decision making, and the method comprises the steps: obtaining health archives, medical history data and current medicine use conditions of old people; based on the health record, the medical history data and the current medicine use condition, generating a personalized medicine use recommendation scheme by using an AI algorithm; generating a medication list and medication reminding according to the personalized medication recommendation scheme; outputting the medication list and medication reminding to the elderly or the family members of the elderly; acquiring medication feedback information of the elderly; based on the medication feedback information, an AI algorithm model is updated, a detailed personal health record is established, and analysis is performed in combination with an AI algorithm, so that the system can fully consider special conditions, such as age, gender, weight and liver and kidney function conditions, of each old person, thereby providing more accurate medication suggestions. Therefore, the risk of adverse reaction is greatly reduced, and the effectiveness of medication is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and more specifically, to a drug management method and system for the elderly based on AI-assisted decision-making. Background Art

[0002] As the global population ages, medication management for the elderly has become an increasingly prominent social issue. Elderly people often suffer from multiple chronic diseases and need to take multiple medications for a long time, which brings a series of challenges, including increased medication complexity, increased risk of drug interactions, and reduced medication compliance. Traditional medication management methods for the elderly mainly rely on doctors' experience and judgment and patients' self-management, which has many limitations.

[0003] In recent years, with the development of information technology, some intelligent drug management systems have begun to be applied to the field of medication management for the elderly. These systems usually include electronic prescriptions, medication reminders, and drug interaction check functions, which have improved the efficiency and safety of medication management to a certain extent. However, these systems still have some obvious shortcomings.

[0004] First, existing drug management systems often use static rule engines that fail to fully consider individual differences and dynamic changes in the health status of the elderly. For example, some systems may only recommend the standard dosage in the drug instructions, while ignoring the special circumstances of decreased liver and kidney function that may exist in the elderly. This "one-size-fits-all" approach may lead to improper medication and increase the risk of adverse reactions.

[0005] Second, existing systems perform poorly in dealing with the complexities of polypharmacy. Polypharmacy, a common problem among the elderly, involves not only interactions between drugs, but also interactions between drugs and diseases. Existing systems can usually only detect known drug interactions, but are unable to fully evaluate and optimize complex polypharmacy situations.

[0006] Furthermore, most existing systems lack effective feedback mechanisms and self-learning capabilities. They are unable to adjust medication regimens based on the actual medication effects and reactions of the elderly, nor can they learn and improve decision-making models from a large amount of medication data. This results in the system being unable to adapt to individual differences and time changes, and the long-term use effect is poor.

[0007] In addition, existing systems also have deficiencies in medication compliance management. Although some systems provide medication reminder functions, they lack comprehensive consideration of the multiple factors that affect medication compliance and personalized intervention strategies. This makes the system limited in its effectiveness in improving medication compliance among the elderly.

[0008] Finally, existing systems generally lack the ability to predict the health status of the elderly. They mainly focus on current medication management, but cannot foresee potential health risks and changes in medication needs, thus missing opportunities for preventive interventions. Summary of the invention

[0009] The elderly medication management method and system based on AI-assisted decision-making proposed in this invention effectively solves many problems existing in the prior art by introducing advanced artificial intelligence technology. The method comprehensively utilizes deep learning, reinforcement learning, and data mining technologies to achieve personalized, dynamic, and intelligent elderly medication management.

[0010] The present invention provides a method for elderly medication management based on AI-assisted decision-making, including:

[0011] The acquisition steps include:

[0012] Obtain the older adult’s health records, medical history data, and current medication use;

[0013] Processing steps include:

[0014] Generate personalized medication recommendations using AI algorithms based on the health records, medical history data, and current medication use;

[0015] Generate a medication list and medication reminder according to the personalized medication recommendation plan;

[0016] Output steps include:

[0017] Outputting the medication list and medication reminder to the elderly or their family members;

[0018] Feedback steps include:

[0019] Obtaining medication feedback from older adults;

[0020] Based on the medication feedback information, the AI ​​algorithm model is updated.

[0021] Preferably, the obtaining step specifically includes:

[0022] Obtain information on the elderly’s age, gender, height, weight, diseases, medications, health conditions, and medication contraindications;

[0023] Based on the information, establish medical health records for the elderly.

[0024] Preferably, the processing step of using an AI algorithm to generate a personalized medication recommendation plan specifically includes:

[0025] Calculate the utility value of different drugs for different diseases based on the health records, medical history data and current drug use;

[0026] generating multiple groups of drug intervention plans according to the utility values;

[0027] Based on preset evaluation criteria, evaluating the multiple groups of drug intervention plans;

[0028] The drug intervention plan with the highest evaluation score is selected as the personalized medication recommendation plan.

[0029] Preferably, the preset evaluation criteria include:

[0030] Drug efficacy;

[0031] Adverse reactions;

[0032] Drug interactions;

[0033] Medication adherence;

[0034] Wherein, the adverse reactions are evaluated by the following steps:

[0035] Set the weight value for adverse reaction symptom assessment;

[0036] Each type of adverse reaction symptom was scored;

[0037] The average score of the sum of all scores is calculated as the final score.

[0038] Preferably, the feedback step specifically includes:

[0039] Obtain information on changes in health status of older adults after using recommended medications;

[0040] Obtain information on adverse reactions in the elderly;

[0041] Obtaining medication adherence information from older adults;

[0042] Based on the health status change information, adverse reaction information and medication compliance information, a medication effect score is calculated.

[0043] Preferably, the calculation formula for the medication effect score is:

[0044] H=W*L*K

[0045] Among them, H is the medication effect score, W is the adverse reaction symptom score, L is the health value change, and K is the influencing factor of medication compliance and treatment effect.

[0046] Preferably, the method further comprises the following steps:

[0047] Based on the medication feedback information, determining whether the medication regimen needs to be adjusted;

[0048] If adjustment is required, the processing steps and output steps are repeated until a medication regimen that meets the conditions is generated.

[0049] Preferably, the method further comprises the step of data mining:

[0050] Based on the acquired health records, medical history data, current medication usage, and medication feedback information, a medication dataset for the elderly was constructed;

[0051] The data set is analyzed using a machine learning algorithm to predict changing trends in the health status of the elderly.

[0052] Preferably, the AI ​​algorithm includes a deep reinforcement learning algorithm, specifically including:

[0053] Establish drug name collection and feedback collection;

[0054] Performing iterative calculation based on the drug name set and the feedback set;

[0055] Based on the iteration results, the optimal drug combination is generated.

[0056] The drug management system for the elderly based on AI-assisted decision-making includes:

[0057] A health status assessment module is used to assess the health status of the elderly based on their health records;

[0058] Medical history information management module, used to manage and analyze the medical history information of the elderly;

[0059] Medication preference analysis module, which is used to analyze the medication preferences of the elderly and generate a personalized medication recommendation list;

[0060] Medication management module, which uses AI algorithms to generate medication control recommendations based on health status, medical history, and medication preferences;

[0061] Feedback module, which is used to collect and process medication feedback information from the elderly and optimize the AI ​​algorithm model accordingly;

[0062] The system realizes drug management for the elderly by executing the described method.

[0063] The beneficial effects of the present invention are mainly reflected in the following aspects:

[0064] First, the present invention realizes truly personalized medication recommendations. By establishing a detailed personal health profile and combining it with AI algorithms for analysis, the system can fully consider the special circumstances of each elderly person, such as age, gender, weight, and liver and kidney function, and thus provide more accurate medication recommendations. This greatly reduces the risk of adverse reactions and improves the effectiveness of medication.

[0065] Secondly, the present invention effectively solves the complex problem of multiple medications. Through the deep reinforcement learning algorithm, the system can find the optimal drug combination based on the consideration of multiple drug interactions. This not only reduces the risk of drug interactions, but also maximizes the therapeutic effect, solving the shortcomings of the existing technology in dealing with complex medication situations.

[0066] Furthermore, the present invention introduces dynamic feedback and self-learning mechanisms. The system continuously collects medication feedback information and uses this information to update the AI ​​model, thereby achieving dynamic optimization of medication regimens. This adaptive capability enables the system to continuously improve the accuracy of its decisions, overcoming the limitation of existing systems that lack learning capabilities.

[0067] In terms of medication compliance management, the present invention proposes an innovative solution. The system not only provides intelligent reminders, but also provides personalized intervention strategies by analyzing various factors that affect compliance, such as medication frequency and dosage form complexity. This comprehensive compliance management method significantly improves medication compliance in the elderly.

[0068] In addition, the data mining and health trend prediction functions of the present invention provide the possibility for preventive intervention. By analyzing a large amount of historical data, the system can predict the changing trend of the health status of the elderly, so as to adjust the medication plan in advance and achieve early intervention and control of the disease.

[0069] Finally, the modular design and advanced AI technology of the present invention make the system have excellent scalability and adaptability, which means that the system can easily integrate new medical discoveries and technological advances, and continuously improve its performance and functions.

[0070] In general, the elderly medication management method and system based on AI-assisted decision-making proposed in the present invention innovatively solves multiple key problems in the prior art and realizes accurate, safe and efficient elderly medication management. This can not only significantly improve the quality of life of the elderly, but also reduce the burden on the medical system, which is of great significance for coping with the social challenges brought about by an aging population. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 The present invention is a flow chart of the method.

[0072] Figure 2 It is a logic block diagram of the processing steps of the present invention.

[0073] Figure 3 It is a logic block diagram of the feedback step and dynamic adjustment of the present invention.

[0074] Figure 4 It is a logical block diagram of the data mining step of the present invention. DETAILED DESCRIPTION

[0075] Please refer to Figure 1-4 The present invention provides a drug management method and system for the elderly based on AI-assisted decision-making. The method aims to provide personalized drug management services for the elderly through artificial intelligence technology to improve the safety and effectiveness of drug use. The specific implementation methods of the present invention will be described in detail below.

[0076] The method of the present invention includes an acquisition step, a processing step, an output step and a feedback step. In the acquisition step, the system acquires the health records, medical history data and current medication usage of the elderly. This information usually comes from multiple sources, such as hospital electronic medical record systems, family doctor records, self-reports of the elderly or their families, etc. Preferably, this information can be automatically acquired through a standardized interface or API to ensure the timeliness and accuracy of the data.

[0077] In the processing step, the system generates a personalized medication recommendation plan based on the acquired information using an AI algorithm. In one embodiment of the present invention, the AI ​​algorithm can be a neural network model based on deep learning, which has been trained with a large amount of elderly medication data and can consider multiple factors for decision making. For example, the model can consider factors such as the elderly's age, gender, weight, liver and kidney function, previous medical history, current medication status, and potential drug interactions.

[0078] Specifically, the AI ​​algorithm can calculate the drug recommendation score using the following formula:

[0079]

[0080] Among them, S is the drug recommendation score, n is the number of factors considered, and w i is the weight of the i-th factor, f i (x) is the scoring function of the i-th factor, and x is the input health and medication information of the elderly.

[0081] Based on the generated personalized medication recommendation plan, the system further generates a medication list and medication reminders. The medication list usually includes information such as the name of the drug, dosage, and frequency of use. Medication reminders can be set as scheduled reminders, such as three times a day, morning, noon, and evening, or more detailed reminder times can be set according to the characteristics of specific drugs.

[0082] In the output step, the system outputs the generated medication list and medication reminder to the elderly or their family members. The output methods can be diverse, such as push notifications through smartphone APP, text messages, voice reminders through smart speakers, etc. In another embodiment of the present invention, the system can also be connected to a smart medicine box to directly control the medicine box to open the corresponding medicine compartment at a specified time, further improving the accuracy of medication.

[0083] The feedback step is an important feature of this method. The system obtains medication feedback information from the elderly, which may include physical reactions after medication, whether adverse reactions occur, medication compliance, etc. Preferably, the feedback information can be automatically collected through smart devices, such as smart bracelets to monitor physiological indicators such as heart rate and blood pressure, or through regular questionnaires to collect subjective feelings.

[0084] Based on the feedback information obtained, the system updates the AI ​​algorithm model. This step enables the system to continuously learn and optimize to adapt to the individual characteristics and medication needs of each elderly person. For example, if a drug is found to have a mild adverse reaction to a specific elderly person, the system will appropriately reduce the recommendation weight of the drug in subsequent recommendations.

[0085] The acquisition step of the present invention is further refined into obtaining detailed personal information of the elderly and establishing a medical health record. Specifically, the system obtains information such as the elderly's age, gender, height, weight, diseases, medication, health status, and medication contraindications. This information constitutes the basic health record of the elderly.

[0086] Age is an important parameter because the physiological functions of the elderly change with age, which will affect the metabolism and action of drugs in the body. For example, for elderly people over 80 years old, the dosage of certain drugs may need to be appropriately reduced. Gender is also a factor to consider because the metabolism rate of certain drugs in men and women may be different.

[0087] Height and weight are used to calculate body mass index (BMI), which is an important indicator for assessing the nutritional status of the elderly and also affects the dosage selection of certain drugs. For example, for elderly people with a BMI greater than 30, some drugs may require increased dosage to achieve ideal blood drug concentrations.

[0088] Disease information is the core of medication management. The system needs to record in detail all chronic and acute diseases that the elderly currently suffer from. Medication status includes all medications currently in use, including prescription and over-the-counter drugs. Health status can include recent physical examination results, self-care ability assessment, etc. Medication contraindication information is particularly important, which includes the elderly's history of allergies to certain drugs, drugs that cannot be used for other reasons, etc.

[0089] Based on this information, the system establishes a medical health record for the elderly. This record is dynamic and will be updated as the health status of the elderly changes. Preferably, the system can set an automatic reminder function, such as reminding to update the health record every 3 months to ensure the timeliness of the information.

[0090] The process of using AI algorithms to generate personalized medication recommendations in the processing steps of the present invention is further refined into four sub-steps. First, the system calculates the utility values ​​of different drugs for different diseases based on health records, medical history data, and current drug usage. This calculation process can use the following formula:

[0091] E ij =α·T ij +β·S ij -γ·I ij

[0092] Among them, E ij is the utility value of drug i for disease j, T ij S is the score of the therapeutic effect of drug i on disease j, ij is the degree of relief of the current symptoms of the elderly by drug i, ij is the degree of interaction between drug i and the elderly’s current medications, and α, β, and γ are weight coefficients.

[0093] Next, the system generates multiple groups of drug intervention plans based on the calculated utility values. Each group of plans may contain a combination of one or more drugs. Preferably, the system will consider the synergistic effects of the drugs and try to select the combination that can produce the best therapeutic effect.

[0094] The system then evaluates these drug intervention plans based on preset evaluation criteria. The evaluation criteria may include multiple dimensions such as treatment effect, risk of side effects, convenience of medication, and economic efficiency of medication. Each dimension can be assigned a different weight to reflect its importance in decision-making. For example, for some serious diseases, the treatment effect may be more important, while for chronic disease management, the risk of side effects and convenience of medication may be more important.

[0095] Finally, the system selects the drug intervention plan with the highest evaluation score as the personalized medication recommendation plan. This plan not only takes into account the therapeutic effect of the drug, but also balances other factors to ensure the feasibility and safety of the plan.

[0096] The method of the present invention can comprehensively consider the individual characteristics and medication needs of the elderly by introducing AI-assisted decision-making, and provide more accurate and personalized drug management services. At the same time, the decision-making model is continuously optimized through the feedback mechanism, so that the system can adapt to the dynamic changes in the health status of the elderly. This method is expected to significantly improve the safety and effectiveness of medication for the elderly, reduce the occurrence of adverse reactions, improve the quality of chronic disease management, and thus improve the quality of life of the elderly. The method of the present invention further refines the preset evaluation criteria, as described in claim 4, and these criteria include drug efficacy, adverse reactions, drug interactions, and medication compliance. The selection of these criteria is intended to comprehensively evaluate the effectiveness and safety of drug intervention programs.

[0097] Drug efficacy is the primary criterion for evaluation, which directly reflects the therapeutic effect of the drug on the target disease. In one embodiment of the present invention, drug efficacy can be quantified by combining clinical trial data and real-world evidence. For example, the drug efficacy score can be calculated using the following formula:

[0098] E=α·RCT+β·RWE

[0099] Wherein, E is the drug utility score, RCT is the standardized score of the randomized controlled trial results, RWE is the standardized score of the real-world evidence, α and β are weight coefficients, and α+β = 1. Preferably, α = 0.7 and β = 0.3 can be set to balance the test results under ideal conditions and the effects in practical applications.

[0100] Adverse reactions are key indicators for evaluating drug safety. The present invention proposes a novel adverse reaction evaluation method, comprising the following steps: First, setting weight values ​​for adverse reaction symptom evaluation. These weight values ​​can be determined based on the severity and frequency of adverse reactions. For example, adverse reactions can be divided into three levels: mild, moderate, and severe, and weight values ​​of 1, 2, and 3 are assigned respectively. Secondly, each type of adverse reaction symptom is scored. The score can be adjusted based on the individual characteristics of the elderly and previous medication history. Finally, the average score of the sum of each score is calculated as the final score.

[0101] Specifically, the adverse reaction score can be calculated using the following formula:

[0102]

[0103] Where ADR is the adverse reaction score, n is the number of adverse reaction symptoms, and w i is the weight value s of the ith symptom i Score the severity of the ith symptom (a scale of 0-10 may be used).

[0104] Drug interactions are an important consideration in medication management for the elderly, as they often take multiple drugs at the same time. The method of the present invention uses a drug interaction matrix to assess the potential risk of interactions. Each element in the matrix I ij Indicates the degree of interaction between drug i and drug j, using a scale of 0-3 (0 for no interaction, 3 for severe interaction). The system calculates the sum of the interaction scores of all drug pairs in the recommended regimen as one of the evaluation indicators.

[0105] Medication compliance is an important factor in ensuring treatment effectiveness, especially for elderly people whose memory may decline. The method of the present invention takes into account multiple factors that affect medication compliance, such as medication frequency, dosage form, and the complexity of the administration method. For example, the following formula can be used to calculate the medication compliance score:

[0106] A=1-(α·F+β·C+γ·M)

[0107] Among them, A is the medication compliance score, F is the standardized score of medication frequency, C is the standardized score of dosage form complexity, M is the standardized score of administration method complexity, α, β, γ are weight coefficients, and α+β+γ=1.

[0108] The method of the present invention further refines the types of information obtained in the feedback step. The system obtains information on changes in health status, adverse reactions, and medication compliance of the elderly after using the recommended drugs. The acquisition of this information can be achieved through a variety of channels.

[0109] Information on changes in health status can be obtained through regular physical examination results, physiological indicator monitoring data (such as blood pressure, blood sugar, etc.) and the subjective feelings of the elderly. In a preferred embodiment of the present invention, the system can be connected to intelligent medical devices, such as intelligent blood pressure meters, blood sugar meters, etc., to achieve automatic collection and uploading of health data.

[0110] Acquisition of adverse reaction information requires special attention, because the elderly may not actively report mild discomfort symptoms. The method of the present invention adopts a combination of active inquiry and passive monitoring. The system can regularly push simple health status questionnaires to the elderly or their caregivers, asking whether specific adverse reaction symptoms have occurred. At the same time, the system will also analyze the physiological indicator data of the elderly, and if abnormal fluctuations are found, they will be automatically marked as potential adverse reactions.

[0111] The acquisition of medication compliance information is an innovative point of this method. In addition to the traditional self-reporting method, the present invention also proposes a medication compliance monitoring method based on a smart medicine box. The smart medicine box can record the time of each box opening and transmit the data to the system. The system calculates the medication compliance score by comparing the actual box opening time with the preset medication time.

[0112] Based on the acquired information, the system calculates the medication effect score. As described in claim 6, the calculation formula for the medication effect score is:

[0113] H=W·L·K

[0114] Among them, H is the medication effect score, W is the adverse reaction symptom score, L is the health value change, and K is the influencing factor of medication compliance and treatment effect.

[0115] Specifically, W can be calculated using the adverse reaction scoring method mentioned above. L can be calculated by comparing health indicators before and after medication, for example:

[0116]

[0117] Among them, H post is the health index value after taking the medicine, H pre It is the health index value before taking the medicine.

[0118] The K factor reflects the impact of medication compliance on treatment efficacy and can be calculated using the following formula:

[0119] K=α·A+β·T

[0120] Among them, A is the medication compliance score, T is the treatment effect score, α and β are weight coefficients, and α+β=1.

[0121] The method of the present invention also includes a dynamic adjustment step, which reflects the adaptive characteristics of the method and can continuously optimize the medication regimen according to the actual medication effect.

[0122] In the dynamic adjustment step, the system first determines whether the medication regimen needs to be adjusted based on the medication feedback information. The basis for judgment can be whether the medication effect score is lower than the preset threshold, or whether serious adverse reactions have occurred. If adjustment is required, the system will re-execute the processing step and the output step to generate a new medication regimen.

[0123] Preferably, the system will consider the problems with the previous plan when generating a new plan. For example, if the previous plan caused serious adverse reactions, the new plan will reduce the dosage of the relevant drugs or replace them with other drugs of the same class. If the problem is low medication compliance, the new plan may adjust the frequency of administration or dosage form to improve convenience.

[0124] The method of the present invention realizes closed-loop control of medication management through this dynamic adjustment mechanism, and can continuously optimize medication regimens and improve treatment effects and safety. This is especially important for elderly patients with chronic diseases who require long-term medication, because it can adapt to changes in medication needs caused by changes in the physiological functions of the elderly over time. As described in claim 8, the method of the present invention also includes a data mining step. This step aims to make full use of the large amount of data accumulated by the system, mine valuable information through advanced analytical techniques, and provide deeper insights into medication management for the elderly.

[0125] In the data mining step, the system first constructs a medication dataset for the elderly based on the acquired health records, medical history data, current medication usage, and medication feedback information. This dataset is a multi-dimensional structured database that contains time series information. Preferably, the dataset adopts a distributed storage structure to improve data processing efficiency and system scalability.

[0126] The constructed dataset includes but is not limited to the following main dimensions:

[0127] 1. Demographic characteristics: age, gender, education level, etc.

[0128] 2. Health status indicators: time series data of various physiological indicators.

[0129] 3. Disease information: diagnosis, course of disease, severity, etc.

[0130] 4. Medication information: drug name, dosage, frequency, duration of use, etc.

[0131] 5. Lifestyle information: eating habits, exercise status, sleep quality, etc.

[0132] 6. Medication feedback: adverse reactions, treatment effects, compliance, etc.

[0133] Based on this comprehensive data set, the method of the present invention uses a machine learning algorithm to analyze the data and predict the trend of changes in the health status of the elderly. The machine learning algorithm here can be an integration of multiple models to improve the accuracy and robustness of the prediction.

[0134] In a preferred embodiment of the present invention, the prediction model includes the following components:

[0135] 1. Time series analysis model: used to capture long-term trends and cyclical changes in health indicators. You can use the ARIMA (autoregressive integrated moving average) model or the more advanced LSTM (long short-term memory) neural network.

[0136] 2. Classification model: used to predict the risk of specific health events (such as acute disease onset). Ensemble learning methods such as random forests or gradient boosting trees can be used.

[0137] 3. Regression model: used to predict continuous health indicator values. Support vector regression (SVR) or multi-layer perceptron (MLP) can be used.

[0138] The outputs of these models will be considered together to generate a final forecast of health status change trends. The forecast results can be used to guide the adjustment of medication regimens and achieve preventive interventions. For example, if the model predicts that an elderly person may have problems with blood pressure control in the next three months, the system may recommend adjusting the dosage or type of antihypertensive drugs in advance.

[0139] The AI ​​algorithm in the method of the present invention includes a deep reinforcement learning algorithm. This algorithm is particularly suitable for dealing with problems such as medication management for the elderly that require long-term decision-making optimization.

[0140] The core idea of ​​deep reinforcement learning algorithm is to find the optimal decision-making strategy through continuous trial and learning. In the context of the present invention, "decision" refers to drug selection and dosage adjustment, and "reward" can be understood as the score of drug effect.

[0141] The specific implementation steps of the algorithm are as follows:

[0142] 1. Establish a drug name set and a feedback set. The drug name set A contains all possible recommended drugs, and the feedback set Q contains various feedback indicators after taking the drugs.

[0143] 2. Perform iterative calculations based on the drug name set and the feedback set. In each iteration, the algorithm will:

[0144] a) Select a combination of drugs

[0145] b) Simulate the effect of the combination

[0146] c) Update the Q value (expected reward) based on the simulation results

[0147] 3. Generate the optimal drug combination based on the iteration results.

[0148] Preferably, the present invention uses a deep Q network (DQN) as a specific algorithm implementation. DQN uses a neural network to approximate the Q function and can handle high-dimensional state spaces. The update formula of the Q function is as follows:

[0149]

[0150] Among them, s is the current state, a is the selected action (drug combination), r is the immediate reward, s' is the next state, α is the learning rate, and r is the discount factor.

[0151] The advantage of this approach is that it can take into account the long-term effects of medication, not just the immediate response. For example, a drug combination may not be effective in the short term, but will significantly improve health after long-term use. Deep reinforcement learning algorithms can capture this long-term effect and make better decisions.

[0152] The present invention also provides an elderly medication management system based on AI-assisted decision-making, which is a hardware carrier for implementing the above method. This system includes several key modules, each of which is responsible for a specific function in the method.

[0153] The health status assessment module is responsible for assessing the health status of the elderly based on their health records. This module may contain multiple submodules, such as data preprocessing submodule, feature extraction submodule and scoring submodule. The data preprocessing submodule is responsible for cleaning and standardizing the input health data; the feature extraction submodule is responsible for extracting key features from the raw data; and the scoring submodule calculates the health status score based on the extracted features.

[0154] The medical history information management module is used to manage and analyze the medical history information of the elderly. This module may use graph database technology to represent the relationship between various diseases, symptoms and treatments as a complex network, which is convenient for in-depth analysis and reasoning.

[0155] The medication preference analysis module is responsible for analyzing the medication preferences of the elderly and generating a personalized medication recommendation list. This module may use collaborative filtering and content-based recommendation techniques to generate recommendations by comprehensively considering the elderly's historical medication records and the characteristics of the drugs.

[0156] The medication management module is the core of the system, which uses AI algorithms to generate medication control recommendations. This module integrates the various algorithms described above, including deep reinforcement learning algorithms. Its input includes health status, medical history, and drug preferences, and its output is detailed medication recommendations.

[0157] The feedback module is responsible for collecting and processing medication feedback information from the elderly and optimizing the AI ​​algorithm model accordingly. This module may include submodules such as data acquisition interface, data analysis engine, and model updater. The data acquisition interface can be connected to various smart medical devices; the data analysis engine is responsible for extracting useful information from the original feedback data; and the model updater adjusts the parameters of the AI ​​model based on the analysis results.

[0158] These modules work closely together to implement the methods described in claims 1 to 9. The modular design of the system makes it well scalable and maintainable. For example, if a new health assessment algorithm is developed in the future, only the health assessment module needs to be updated without changing the entire system.

[0159] In general, the elderly medication management method and system based on AI-assisted decision-making provided by the present invention realizes accurate management of elderly medication by comprehensively utilizing a variety of advanced artificial intelligence technologies. It can not only provide personalized medication recommendations, but also dynamically adjust according to actual effects, and predict health trends through data mining. This method and system are expected to significantly improve the safety and effectiveness of elderly medication, improve the quality of chronic disease management, and thus improve the quality of life of the elderly.

[0160] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, replacement, and improvement made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A drug management method for the elderly based on AI-assisted decision-making, characterized in that: include: The acquisition steps include: Obtain the older adult’s health records, medical history data, and current medication use; Processing steps include: Generate personalized medication recommendations using AI algorithms based on the health records, medical history data, and current medication use; Generate a medication list and medication reminder according to the personalized medication recommendation plan; Output steps include: Outputting the medication list and medication reminder to the elderly or their family members; Feedback steps include: Obtaining medication feedback from older adults; Based on the medication feedback information, the AI ​​algorithm model is updated.

2. The method according to claim 1, characterized in that The acquisition step specifically includes: Obtain information on the elderly’s age, gender, height, weight, diseases, medications, health conditions, and medication contraindications; Based on the information, establish medical health records for the elderly.

3. The method according to claim 1, characterized in that The processing steps of using AI algorithms to generate personalized medication recommendation plans specifically include: Calculate the utility value of different drugs for different diseases based on the health records, medical history data and current drug use; generating multiple groups of drug intervention plans according to the utility values; Based on preset evaluation criteria, evaluating the multiple groups of drug intervention plans; The drug intervention plan with the highest evaluation score is selected as the personalized medication recommendation plan.

4. The method according to claim 3, characterized in that The preset evaluation criteria include: Drug efficacy; Adverse reactions; Drug interactions; Medication adherence; Wherein, the adverse reactions are evaluated by the following steps: Set the weight value for adverse reaction symptom assessment; Each type of adverse reaction symptom was scored; The average score of the sum of all scores is calculated as the final score.

5. The method according to claim 1, characterized in that The feedback step specifically includes: Obtain information on changes in health status of older adults after using recommended medications; Obtain information on adverse reactions in the elderly; Obtaining medication adherence information from older adults; A medication effect score is calculated based on the health status change information, adverse reaction information and medication compliance information.

6. The method according to claim 5, characterized in that The calculation formula of the medication effect score is: H=W*L*K Among them, H is the medication effect score, W is the adverse reaction symptom score, L is the health value change, and K is the influencing factor of medication compliance and treatment effect.

7. The method according to claim 1, characterized in that It also includes dynamic adjustment steps: Based on the medication feedback information, determining whether the medication regimen needs to be adjusted; If adjustment is required, the processing steps and output steps are repeated until a medication regimen that meets the conditions is generated.

8. The method according to claim 1, characterized in that It also includes the data mining step: Based on the acquired health records, medical history data, current medication usage, and medication feedback information, a medication dataset for the elderly was constructed; The data set is analyzed using a machine learning algorithm to predict changing trends in the health status of the elderly.

9. The method according to claim 1, characterized in that: The AI ​​algorithm includes a deep reinforcement learning algorithm, specifically including: Establish drug name collection and feedback collection; Performing iterative calculation based on the drug name set and the feedback set; Based on the iteration results, the optimal drug combination is generated.

10. Elderly medication management system based on AI-assisted decision-making, characterized by: include: A health status assessment module is used to assess the health status of the elderly based on their health records; Medical history information management module, used to manage and analyze the medical history information of the elderly; Medication preference analysis module, which is used to analyze the medication preferences of the elderly and generate a personalized medication recommendation list; Medication management module, which uses AI algorithms to generate medication control recommendations based on health status, medical history, and medication preferences; Feedback module, which is used to collect and process medication feedback information from the elderly and optimize the AI ​​algorithm model accordingly; The system implements drug management for the elderly by executing the method described in any one of claims 1-9.

Images