Day chemotherapy center appointment treatment method and system
The method and system optimize day-care chemotherapy by integrating AI algorithms for scheduling, personalized treatment, and real-time monitoring, addressing inefficiencies and improving patient experience and treatment efficiency.
Patent Information
- Application Number
- CN202410625066.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-20
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-05-20
AI Technical Summary
The existing day-care chemotherapy process faces challenges in scheduling patient appointments and doctor shifts, inefficient information collection and processing, uneven resource allocation, and manual real-time monitoring, leading to complex and error-prone operations that affect treatment efficiency and patient experience.
A method and system utilizing deep reinforcement learning, natural language processing, generative adversarial networks, multi-agent algorithms, and time series analysis to optimize appointment scheduling, personalize treatment plans, and dynamically adjust treatment parameters based on patient data and behavior patterns, ensuring seamless resource allocation and real-time monitoring.
This approach improves appointment efficiency, reduces conflicts and waiting times, enhances resource utilization, provides personalized treatment plans, increases patient compliance, and ensures timely and effective chemotherapy processes through intelligent monitoring and reporting.
Smart Images

Figure CN118448008B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to an appointment treatment method and system for a day chemotherapy center. Background Art
[0002] Day-care chemotherapy centers are a widely used model in modern cancer treatment, which aims to improve the convenience of medical services and resource utilization by allowing patients to receive chemotherapy during the day and go home to rest at night. The existing day-care chemotherapy process usually includes steps such as patient appointment, information collection, personalized treatment plan formulation, appointment confirmation and reminder, pre-treatment preparation and coordination, and real-time monitoring during chemotherapy. These steps are independent and are managed and coordinated by medical staff manually or semi-automatically.
[0003] However, the existing day-case chemotherapy process still has some problems in actual operation. First, the scheduling of patient appointments and doctor shifts is complex and error-prone, and lacks an intelligent optimization mechanism. Secondly, the collection and processing of patient information is inefficient, and historical data cannot be fully utilized for personalized treatment. Furthermore, the pre-treatment preparation and coordination links lack systematic management, and resources are unevenly allocated, resulting in inefficient and discontinuous treatment processes. In addition, real-time monitoring and adjustments during chemotherapy mainly rely on manual operations, making it difficult to achieve dynamic and accurate adjustments to treatment parameters. These problems affect the overall effect of day-case chemotherapy and patient experience. Summary of the invention
[0004] In view of this, an embodiment of the present invention provides a method and system for appointment treatment in a day chemotherapy center, which is used to improve the efficiency of appointment treatment in a day chemotherapy center.
[0005] The present invention provides a method for appointment treatment in a day chemotherapy center, comprising: collecting patient appointment requests, and performing request scheduling analysis on the patient appointment requests to obtain an appointment number and a doctor's schedule;
[0006] Matching a target patient tag according to the patient appointment request, and extracting historical patient medical data through the target patient tag;
[0007] Extracting current diagnosis and treatment information corresponding to the patient appointment request, and matching target medical items based on the historical patient medical data and the current diagnosis and treatment information;
[0008] Predicting patient behavior patterns on the historical patient medical data to obtain predicted behavior patterns, and generating reminder strategies based on the target medical items and the predicted behavior patterns;
[0009] Performing project node segmentation on the target medical project to obtain a plurality of project nodes, inputting the plurality of project nodes into a preset multi-agent algorithm for node sorting, and obtaining a target diagnosis and treatment node sorting table;
[0010] An appointment status report is generated based on the target diagnosis and treatment node ranking table and the reminder strategy, and the appointment status report is transmitted to the target terminal corresponding to the patient appointment request.
[0011] In the present invention, the step of collecting patient appointment requests, performing request scheduling analysis on the patient appointment requests, and obtaining appointment numbers and doctor schedules includes:
[0012] Collecting the patient's appointment request, and performing data analysis on the patient's appointment request to obtain analysis data, wherein the analysis data includes: the patient's name, age, and expected consultation time;
[0013] Collecting the current appointment data set in real time, wherein the appointment data set includes appointment requests of multiple patients and collecting the working hours of doctors;
[0014] Inputting the appointment requests of the multiple patients and the working hours of the doctors into a preset deep reinforcement learning algorithm to construct a simulation environment to obtain a simulated medical environment;
[0015] Extracting parameter information of the simulated medical environment, wherein the parameter information includes: a state space, an action space, and a coordinable space, wherein the state space includes appointment status, doctor free time, and free treatment rooms, the action space includes appointment arrangement, doctor scheduling, and treatment room allocation, and the coordinable space includes appointment coordination, doctor coordination, and treatment room coordination;
[0016] According to the parameter information, a request scheduling analysis is performed on the parsed data in the simulated medical environment to obtain the appointment number and the doctor's schedule.
[0017] In the present invention, the step of matching the target patient tag according to the patient appointment request and extracting the historical patient medical data through the target patient tag includes:
[0018] Predicting patient labels on the parsed data to obtain initial patient labels;
[0019] Extracting similar case data based on the initial patient label, extracting common features from the similar case data to obtain target features;
[0020] Performing label correction on the initial patient label based on the target feature to obtain the target patient label;
[0021] The historical patient medical data is extracted from a preset historical patient database based on the target patient tag.
[0022] In the present invention, the step of extracting the current diagnosis and treatment information corresponding to the patient appointment request and matching the target medical item based on the historical patient medical data and the current diagnosis and treatment information includes:
[0023] Extracting the current diagnosis and treatment information corresponding to the patient's appointment request, wherein the current diagnosis and treatment information includes: condition description, symptoms and diagnosis results;
[0024] Constructing a patient condition portrait based on the condition description, the symptoms, and the diagnosis results to obtain the patient condition portrait;
[0025] Based on the patient's condition portrait, similar data is extracted from the historical patient medical data to obtain target similar data;
[0026] The current diagnosis and treatment information is matched with medical items based on the target similarity data to obtain the target medical items.
[0027] In the present invention, the step of predicting the patient behavior pattern of the historical patient medical data to obtain the predicted behavior pattern, and generating a reminder strategy based on the target medical item and the predicted behavior pattern includes:
[0028] Extracting appointment performance features from the historical patient medical data to obtain an appointment performance feature set, wherein the appointment performance feature set includes: appointment time, number of cancellations, treatment frequency, attendance rate, patient feedback, and daily activity patterns;
[0029] Predicting patient behavior patterns on the historical patient medical data based on the appointment performance feature set to obtain a predicted behavior pattern;
[0030] Predicting the patient status of the predicted behavior pattern to obtain predicted patient status data, wherein the predicted patient status data includes: patient compliance and appointment change probability;
[0031] Generate a reminder schedule based on the appointment change probability, and generate patient visit reminders based on the patient compliance;
[0032] The reminder strategy is generated through the medical appointment reminder items and the reminder schedule.
[0033] In the present invention, the step of segmenting the target medical project into project nodes to obtain a plurality of project nodes, inputting the plurality of project nodes into a preset multi-agent algorithm for node sorting, and obtaining a target diagnosis and treatment node sorting table includes:
[0034] Performing project node segmentation on the target medical project to obtain multiple project nodes;
[0035] Perform intelligent agent matching on each of the project nodes respectively to obtain the intelligent agent corresponding to each of the project nodes;
[0036] Constructing task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes;
[0037] Sort the nodes according to the task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes to obtain an initial node sorting;
[0038] The initial node ranking is input into the multi-agent algorithm for multiple rounds of iterations to obtain the target diagnosis and treatment node ranking table.
[0039] In the present invention, the step of generating an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmitting the appointment status report to the target terminal corresponding to the patient appointment request includes:
[0040] Extracting data from the target diagnosis and treatment node sorting table to obtain specific execution time;
[0041] Modify the reminder schedule corresponding to the reminder strategy according to the specific execution time to obtain a target reminder schedule;
[0042] Matching the target reminder schedule with reminder content to obtain target reminder content, wherein the target reminder content includes text message content, phone notification content, and application notification content;
[0043] An appointment status report is generated based on the target reminder content, and the appointment status report is transmitted to a target terminal corresponding to the patient appointment request.
[0044] The present invention also provides a day chemotherapy center appointment treatment system, comprising:
[0045] An analysis module is used to collect patient appointment requests and perform request scheduling analysis on the patient appointment requests to obtain appointment numbers and doctor schedules;
[0046] A matching module, used for matching a target patient tag according to the patient appointment request, and extracting historical patient medical data through the target patient tag;
[0047] An extraction module, used to extract the current diagnosis and treatment information corresponding to the patient appointment request, and match the target medical project based on the historical patient medical data and the current diagnosis and treatment information;
[0048] A prediction module, configured to predict patient behavior patterns based on the historical patient medical data, obtain predicted behavior patterns, and generate reminder strategies based on the target medical items and the predicted behavior patterns;
[0049] A segmentation module is used to segment the project nodes of the target medical project to obtain multiple project nodes, input the multiple project nodes into a preset multi-agent algorithm to sort the nodes, and obtain a target diagnosis and treatment node sorting table;
[0050] A generation module is used to generate an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmit the appointment status report to the target terminal corresponding to the patient appointment request.
[0051] In the technical solution provided by the present invention, by collecting patient appointment requests and performing request scheduling analysis of deep reinforcement learning algorithms, optimized appointment numbers and doctor schedules can be intelligently generated. This method can not only efficiently process a large number of appointment requests and maximize the resource utilization of doctors and treatment rooms, but also effectively reduce appointment conflicts and waiting time, greatly improving the overall efficiency and accuracy of appointments. Secondly, based on the patient appointment request, the target patient label is matched and the historical patient medical data is extracted, so that the patient's key medical information can be accurately extracted through natural language processing and deep learning algorithms to generate detailed patient medical files. This not only improves the efficiency of information collection, but also ensures that doctors have comprehensive patient information as a reference when formulating treatment plans, providing a solid data foundation for personalized treatment. By matching the current diagnosis and treatment information and historical medical data corresponding to the patient's appointment request with the target medical project, a high-quality personalized treatment plan is generated using the generative adversarial network (GANs) algorithm, and through continuous optimization and evaluation, it is ensured that each treatment plan can fully consider the patient's specific situation and needs, thereby improving the accuracy and effectiveness of treatment. Machine learning and user behavior prediction algorithms are used to conduct in-depth analysis of historical patient medical data, extract appointment performance feature sets, and predict patient compliance and appointment changes. Based on these prediction results, a personalized reminder strategy is generated, including reminder frequency, content and method, to ensure that patients can receive diagnosis and treatment on time and as required. In this way, not only the accuracy of reminders is improved, but also personalized services can be provided according to individual differences, appointment cancellations and delays are reduced, and patient compliance and overall satisfaction are significantly improved. For the preparation and coordination before treatment, an algorithm based on multi-agent and cooperative game theory is introduced. Each preparation link is responsible for an independent intelligent agent to ensure seamless connection between each link and efficient use of resources. Through collaboration and optimization, the intelligent agent dynamically adjusts the resource allocation and process sequence of each link, improves the efficiency of the overall process, ensures that all preparations before treatment can be carried out efficiently and orderly, and lays a solid foundation for the chemotherapy process. The patient's real-time physiological data is efficiently analyzed in time series to predict future changes in physiological state. According to these prediction results, the DRL algorithm dynamically adjusts the monitoring and treatment parameters during chemotherapy. Once an abnormality is detected, the DRL algorithm will immediately calculate the optimal response strategy and notify the doctor to take corresponding measures. This method ensures the safety and effectiveness of the chemotherapy process and provides patients with comprehensive medical protection. In addition, by generating an appointment status report and transmitting it to the patient's target terminal, the patient can be informed of the appointment status and specific treatment arrangements in a timely manner, which improves the transparency of information and the patient's peace of mind. By integrating the target diagnosis and treatment node sorting table and reminder strategy, a detailed appointment status report is generated, including the execution time of each medical step, the items that the patient needs to prepare, the reminder time point and the specific reminder content.The appointment status report not only improves patients' understanding of the treatment process, but also ensures that patients can receive treatment on time and as required, further improving the success rate of treatment and patient satisfaction. Overall, the appointment treatment method of the day chemotherapy center has introduced advanced artificial intelligence algorithms and management methods, from appointments, information collection, personalized treatment plan formulation, appointment confirmation and reminders, pre-treatment preparation and coordination, to real-time monitoring and optimization during chemotherapy, which has comprehensively improved the operational efficiency, medical service quality and patient experience of the day chemotherapy center, and has significant innovation and practical application value. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0053] Figure 1 The present invention is a flowchart of a method for making an appointment for a treatment method in a day chemotherapy center according to an embodiment of the present invention.
[0054] Figure 2 Schematic diagram of a treatment reservation system for a day chemotherapy center according to an embodiment of the present invention. DETAILED DESCRIPTION
[0055] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0056] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.
[0057] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0058] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , Figure 1 is a flow chart of a method for booking a treatment method for a day chemotherapy center according to an embodiment of the present invention. Figure 1 As shown, the following steps are included:
[0059] S101, collecting patient appointment requests, and performing request scheduling analysis on the patient appointment requests to obtain appointment numbers and doctor schedules;
[0060] S102, matching target patient tags according to patient appointment requests, and extracting historical patient medical data through target patient tags;
[0061] S103, extracting the current diagnosis and treatment information corresponding to the patient's appointment request, and matching the target medical project based on the historical patient medical data and the current diagnosis and treatment information;
[0062] S104, predicting patient behavior patterns based on historical patient medical data to obtain predicted behavior patterns, and generating reminder strategies based on target medical items and predicted behavior patterns;
[0063] S105, dividing the target medical project into project nodes to obtain multiple project nodes, inputting the multiple project nodes into a preset multi-agent algorithm to perform node sorting, and obtaining a target diagnosis and treatment node sorting table;
[0064] S106. Generate an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmit the appointment status report to the target terminal corresponding to the patient's appointment request.
[0065] It should be noted that when patients decide to undergo day chemotherapy, they can submit an appointment request through the hospital's online appointment. During this process, patients will fill in their basic personal information, expected visit time, and a brief description of their condition. This information will be collected and stored in a central database. Next, these patient appointment requests will be parsed to extract key information such as the patient's name, age, and expected visit time. These parsed data will be used for subsequent scheduling analysis. In order to ensure the effectiveness of the appointment and the rational allocation of resources, the current appointment data set will be collected in real time, including all submitted patient appointment requests and doctor's work schedule. Next, these appointment requests and doctor's work schedule will be input into the preset deep reinforcement learning algorithm to build a simulated medical environment. In this simulation environment, key parameter information such as state space, action space, and coordinable space will be extracted. The state space includes the current appointment status, doctor's free time, and the status of the idle treatment room; the action space covers operations such as appointment scheduling, doctor scheduling, and treatment room allocation; the coordinable space contains information related to appointment coordination, doctor coordination, and treatment room coordination. Based on these parameter information, the parsed data will be requested for scheduling analysis in a simulation environment, and finally an optimized appointment number and doctor schedule will be generated. After the scheduling analysis is completed, the target patient label will be matched according to each patient's appointment request. The initial patient label can be predicted by natural language processing and deep learning analysis of the information submitted by the patient. These labels may include the patient's disease type, symptom description, and treatment needs. Next, similar case data will be extracted from the historical patient database based on these initial labels, and common features will be extracted from these data to obtain target features. These target features will be used to correct the initial patient labels and generate more accurate target patient labels. Finally, based on these target patient labels, relevant historical patient medical data will be extracted from the preset historical patient database to provide data support for the subsequent personalized treatment plan formulation. Then, the current diagnosis and treatment information corresponding to the patient's appointment request will be extracted, which includes detailed description of the condition, symptoms, and preliminary diagnosis results. By analyzing this information, a patient's condition profile will be constructed. Based on this condition profile, similar data will be extracted from the historical patient medical data to find historical case data similar to the current patient's condition. These target similar data will be used to match the medical items of the current diagnosis and treatment information to ensure that each patient can obtain the most suitable medical item. For example, if a patient's diagnosis shows that he has lung cancer, all treatment options related to lung cancer will be extracted from historical data, and the most suitable chemotherapy program will be matched based on the patient's current condition and symptoms. After the target medical program is determined, the patient's behavior pattern will be predicted based on historical patient medical data.Through machine learning algorithms, it is possible to analyze the appointment performance characteristics in historical data, such as appointment time, number of cancellations, treatment frequency, attendance rate, patient feedback, and daily activity patterns. These characteristics will be used to predict patient behavior patterns, including patient compliance and the probability of appointment changes. Based on these prediction results, personalized reminder strategies will be generated, including the frequency, content, and method of reminders. For example, for patients with low compliance, the frequency of reminders may be increased to ensure that they can arrive at the diagnosis and treatment center on time. Next, the target medical project will be segmented into project nodes to obtain multiple specific project nodes. Each node corresponds to an independent medical step, such as physical examination, blood test, and imaging examination. An intelligent agent will be assigned to each project node, which is responsible for specific task execution and resource coordination. Through multi-agent and cooperative game theory algorithms, the task data and resource data corresponding to each project node will be constructed, and preliminary node sorting will be performed. Then, the initial sorting will be iteratively optimized for multiple rounds, and finally a target diagnosis and treatment node sorting table will be generated to ensure that each medical step is seamlessly connected and improve the efficiency of the overall process. Finally, based on the target diagnosis and treatment node sorting table and reminder strategy, a detailed appointment status report will be generated. This report includes the specific execution time of each medical step, what the patient needs to prepare, the reminder time point and the specific reminder content. The appointment report will be transmitted to the patient's target terminal through encrypted SMS, email or patient application notification to ensure that the patient can receive and view the report in time. For example, after a lung cancer patient makes an appointment for chemotherapy, a detailed appointment report will be sent to tell him the specific examination and treatment time, what needs to be prepared (such as fasting), and reminders for each time node. This comprehensive arrangement and reminder not only improves the patient's understanding and compliance with the treatment process, but also ensures the smooth progress of each step, improves the success rate of treatment and patient satisfaction.
[0066] By executing the above steps, by collecting patient appointment requests and performing request scheduling analysis with deep reinforcement learning algorithms, optimized appointment numbers and doctor schedules can be intelligently generated. This approach can not only efficiently handle a large number of appointment requests and maximize the resource utilization of doctors and treatment rooms, but also effectively reduce appointment conflicts and waiting time, greatly improving the overall efficiency and accuracy of appointments. Secondly, based on the patient appointment request, the target patient label is matched and historical patient medical data is extracted, so that the patient's key medical information can be accurately extracted through natural language processing and deep learning algorithms to generate detailed patient medical records. This not only improves the efficiency of information collection, but also ensures that doctors have comprehensive patient information as a reference when formulating treatment plans, providing a solid data foundation for personalized treatment. By matching the current diagnosis and treatment information and historical medical data corresponding to the patient's appointment request with the target medical project, the generative adversarial network (GANs) algorithm is used to generate high-quality personalized treatment plans, and through continuous optimization and evaluation, it is ensured that each treatment plan can fully consider the patient's specific situation and needs, thereby improving the accuracy and effectiveness of treatment. Machine learning and user behavior prediction algorithms are used to conduct in-depth analysis of historical patient medical data, extract appointment performance feature sets, and predict patient compliance and appointment changes. Based on these prediction results, a personalized reminder strategy is generated, including reminder frequency, content and method, to ensure that patients can receive diagnosis and treatment on time and as required. In this way, not only the accuracy of reminders is improved, but also personalized services can be provided according to individual differences, appointment cancellations and delays are reduced, and patient compliance and overall satisfaction are significantly improved. For the preparation and coordination before treatment, an algorithm based on multi-agent and cooperative game theory is introduced. Each preparation link is responsible for an independent intelligent agent to ensure seamless connection between each link and efficient use of resources. Through collaboration and optimization, the intelligent agent dynamically adjusts the resource allocation and process sequence of each link, improves the efficiency of the overall process, ensures that all preparations before treatment can be carried out efficiently and orderly, and lays a solid foundation for the chemotherapy process. The patient's real-time physiological data is efficiently analyzed in time series to predict future changes in physiological state. According to these prediction results, the DRL algorithm dynamically adjusts the monitoring and treatment parameters during chemotherapy. Once an abnormality is detected, the DRL algorithm will immediately calculate the optimal response strategy and notify the doctor to take corresponding measures. This method ensures the safety and effectiveness of the chemotherapy process and provides patients with comprehensive medical protection. In addition, by generating an appointment status report and transmitting it to the patient's target terminal, the patient can be informed of the appointment status and specific treatment arrangements in a timely manner, which improves the transparency of information and the patient's peace of mind. By integrating the target diagnosis and treatment node sorting table and reminder strategy, a detailed appointment status report is generated, including the execution time of each medical step, the items that the patient needs to prepare, the reminder time point and the specific reminder content.The appointment status report not only improves patients' understanding of the treatment process, but also ensures that patients can receive treatment on time and as required, further improving the success rate of treatment and patient satisfaction. Overall, the appointment treatment method of the day chemotherapy center has introduced advanced artificial intelligence algorithms and management methods, from appointments, information collection, personalized treatment plan formulation, appointment confirmation and reminders, pre-treatment preparation and coordination, to real-time monitoring and optimization during chemotherapy, which has comprehensively improved the operational efficiency, medical service quality and patient experience of the day chemotherapy center, and has significant innovation and practical application value.
[0067] In a specific embodiment, the process of executing step S101 may specifically include the following steps:
[0068] (1) Collecting patient appointment requests and performing data analysis on the patient appointment requests to obtain analysis data, wherein the analysis data includes: patient name, age, and expected appointment time;
[0069] (2) Real-time collection of the current appointment data set, where the appointment data set includes appointment requests from multiple patients and the doctor’s working hours;
[0070] (3) Input multiple patients’ appointment requests and doctors’ working hours into a preset deep reinforcement learning algorithm to construct a simulation environment and obtain a simulated medical environment;
[0071] (4) Extracting parameter information of the simulated medical environment, where the parameter information includes: state space, action space, and coordinable space. The state space includes appointment status, doctor free time, and vacant treatment rooms. The action space includes appointment arrangement, doctor scheduling, and treatment room allocation. The coordinable space includes appointment coordination, doctor coordination, and treatment room coordination.
[0072] (5) Based on the parameter information, request scheduling analysis is performed on the parsed data in a simulated medical environment to obtain the appointment number and doctor schedule.
[0073] Specifically, patients fill in personal information, including name, age, and desired visit time. This data is collected and initially parsed to extract key information, namely the patient's name, age, and desired visit time. Next, the current appointment data set is collected in real time. These data sets include all submitted appointment request information, and the doctor's work schedule is also collected. This data will be stored in a central database so that it can be accessed and updated at any time. In this process, not only a large number of appointment requests need to be processed, but also the work schedule of each doctor needs to be reasonably arranged to avoid resource waste or scheduling conflicts. In order to effectively schedule these appointment requests, a deep reinforcement learning algorithm (DRL) is introduced. Multiple patients' appointment requests and doctors' work hours are input into the preset deep reinforcement learning algorithm to construct a simulated medical environment. In this simulated environment, a lot of training and optimization can be performed to find the best appointment scheduling plan. In the simulated environment, key parameter information is extracted, including state space, action space, and coordinable space. The state space includes the current appointment status, doctor's free time and free treatment room information; the action space includes operations such as appointment scheduling, doctor scheduling and treatment room allocation; the coordinable space covers the relevant information of appointment coordination, doctor coordination and treatment room coordination. By defining these parameters, we can fully understand the current resource status and possible operations, thus providing a basis for optimizing scheduling. For example, suppose a day chemotherapy center receives 10 new appointment requests, each of which contains the patient's detailed information and the expected visit time. At the same time, the working schedule of 5 doctors on that day and the use of treatment rooms are recorded. This information is input into the deep reinforcement learning algorithm to build a simulated medical environment. In this environment, the state space includes the detailed information of each appointment request, the doctor's free time and the use status of the treatment room; the action space includes operations such as assigning an appointment to a doctor, adjusting the doctor's schedule, and allocating treatment rooms; the coordinable space covers how to coordinate these appointment requests, doctor's time and the use of treatment rooms. In this simulation environment, training will be repeated, and the optimal scheduling solution will be found through continuous adjustment and optimization. After each simulation run, the parameter information of the simulation environment will be extracted, and the request scheduling analysis will be performed based on this information. Finally, an optimization plan including appointment numbers and doctor schedules can be generated. This plan ensures that each appointment request can be arranged reasonably, the doctor's time is used efficiently, and the use of treatment rooms is optimized. For example, a patient, Mr. Zhang, submitted an appointment request and hopes to have chemotherapy next Wednesday morning. According to Mr. Zhang's request and the doctor's work schedule, it is found that a doctor, Dr. Li, has free time next Wednesday morning and a treatment room is available.Mr. Zhang’s appointment request, Dr. Li’s free time, and the use of treatment rooms were input into the deep reinforcement learning algorithm for simulation. After multiple simulation runs and optimizations, an appointment number was generated, and Mr. Zhang was scheduled to undergo chemotherapy at 10 a.m. next Wednesday. Dr. Li was responsible for the treatment and an available treatment room was assigned. This method not only improves the efficiency and accuracy of appointments, but also effectively reduces appointment conflicts and waiting time. Through the continuous optimization of the deep reinforcement learning algorithm, it is possible to ensure that each doctor’s working hours are reasonably arranged and the use of treatment rooms is maximized while processing a large number of appointment requests. This intelligent scheduling method not only improves the patient experience, but also improves the utilization of medical resources.
[0074] In a specific embodiment, the process of executing step S102 may specifically include the following steps:
[0075] (1) Predict patient labels for parsed data to obtain initial patient labels;
[0076] (2) Extract similar case data based on the initial patient labels, extract common features from similar case data, and obtain target features;
[0077] (3) Correct the initial patient label based on the target feature to obtain the target patient label;
[0078] (4) Extract historical patient medical data from a preset historical patient database based on the target patient label.
[0079] Specifically, the data parsed from the patient's appointment request is processed. Detailed patient information, including name, age, medical history, current symptoms, and expected visit time, is extracted from the appointment request. These parsed data will be used as input into the natural language processing (NLP) and deep learning algorithm models. NLP technology is used to perform semantic analysis and sentiment recognition on the parsed data to automatically extract key information, such as disease name, symptom description, and medication status. Then, based on the extracted information, deep learning models (such as convolutional neural networks or long short-term memory networks) are used to predict patient labels. The result of this process is initial patient labels, which can describe the patient's main health problems and treatment needs. For example, if a patient describes that he has recently developed symptoms of persistent coughing and weight loss, initial labels such as "lung disease" or "possible lung cancer" may be generated. Next, based on these initial patient labels, similar case data is extracted from the preset historical patient database. This database stores a large number of medical records of historical patients, including their medical history, diagnosis results, treatment plans, and treatment effects. Through the label matching algorithm, the historical case data most similar to the current patient label is found. For example, for the possible lung cancer patients mentioned above, all patient records with the label "lung cancer" or "lung disease" in the historical cases will be retrieved. After extracting similar case data, the data needs to be further processed to extract common features. This step is completed by the feature extraction algorithm, which analyzes the diagnostic information, treatment plans, treatment results, etc. of all similar cases to find the common features in these cases. These features can be the specific type of disease, the stage of disease development, commonly used treatment methods, etc. For example, by analyzing a set of lung cancer case data, it may be found that these patients generally received a certain type of chemotherapy and had similar changes and reactions during the treatment. With these target features, the initial patient label will be corrected. This process is to generate a more accurate target patient label by comparing and optimizing the initial label with the target feature. For example, if the initial label is "possible lung cancer", the label can be corrected to a more specific "early non-small cell lung cancer" or "late lung adenocarcinoma" through feature comparison. This label correction process ensures the accuracy and specificity of the label, providing a more accurate basis for the subsequent formulation of personalized treatment plans. Finally, based on the corrected target patient labels, historical patient medical data that highly matches these labels are extracted from the historical patient database. These data include detailed diagnosis records, treatment plans, treatment results, follow-up records, etc., which provide important reference and basis for the treatment of current patients. For example, for patients diagnosed with "early non-small cell lung cancer", the historical data of all "early non-small cell lung cancer" patients will be extracted, their treatment plans and effects will be analyzed, and the best treatment plans and precautions will be extracted and provided to the current patient's doctor for reference.Through such a complete set of processes, not only can the detailed information of patients be accurately extracted and analyzed, but also through the intelligent label prediction and correction process, it can ensure that each patient can get the most suitable treatment plan and medical services. For example, if a 50-year-old male patient submits an appointment request and describes that he has had symptoms of coughing, chest pain and weight loss in the past few months, this information is first processed by NLP to generate the initial label "possible lung cancer". Then, all case data matching the "possible lung cancer" label are extracted from the historical database, and common features are extracted. It is found that most of these cases are "non-small cell lung cancer". Next, the initial label is corrected to "early non-small cell lung cancer", and all relevant historical patient data are extracted based on this label. Finally, analyze these data to generate personalized treatment plans, including recommended chemotherapy regimens, expected treatment effects and precautions, and provide this information to the patient's attending physician.
[0080] In a specific embodiment, the process of executing step S103 may specifically include the following steps:
[0081] (1) Extract the current medical information corresponding to the patient's appointment request, where the current medical information includes: condition description, symptoms, and diagnosis results;
[0082] (2) Construct a patient's condition portrait based on the condition description, symptoms, and diagnosis results to obtain the patient's condition portrait;
[0083] (3) Based on the patient's condition profile, similar data is extracted from historical patient medical data to obtain target similar data;
[0084] (4) Match the current diagnosis and treatment information with medical items based on the target similarity data to obtain the target medical items.
[0085] Specifically, for example, a patient submits an appointment request, describing that he has had symptoms of persistent cough, weight loss, and chest pain in the past few months, and is initially diagnosed with lung cancer. The text data will be parsed to extract the core symptoms and diagnosis results, and key information such as "cough", "weight loss", "chest pain" and "lung cancer" will be identified. Next, the extracted information will be further processed using deep learning models (such as convolutional neural networks or long short-term memory networks) to construct the patient's medical profile. The medical profile is a detailed digital representation of the patient's health status, including the patient's medical description, symptom combination and diagnosis results. This medical profile can help to fully understand the patient's health status and provide a basis for subsequent data matching and analysis. In this way, not only can the patient's current health status be recorded, but this information can also be standardized to ensure that the medical data of different patients can be effectively compared. For example, if a patient's medical profile shows that he has the characteristics of "lung cancer", "cough", "weight loss" and "chest pain", these characteristics will be combined to form a complete medical profile. Next, based on this medical profile, similar data will be extracted from the historical patient medical database. Machine learning algorithms will be used to calculate the similarity of the disease profiles and find historical case data similar to the disease profile of the current patient. For example, it may be found that there are multiple patients in the database with similar disease profiles, who also suffer from lung cancer and have similar symptoms and diagnosis results. By extracting similar data, the detailed medical records of these similar patients can be identified, including the treatment plans they have received, treatment effects, follow-up records, etc. These target similar data provide valuable reference information for the treatment of the current patient. These similar data will be further analyzed to extract common treatment characteristics and effective treatment plans. For example, if it is found that most similar patients have received a specific chemotherapy regimen and achieved good treatment effects, this regimen can be used as one of the candidate treatment plans for the current patient. Finally, based on the target similar data, the current patient's diagnosis and treatment information will be matched with medical projects. The patient's disease profile will be compared with the extracted similar data to match the most suitable medical project and treatment plan. For example, for the lung cancer patient mentioned above, the most suitable chemotherapy regimen, treatment cycle and expected treatment effect can be determined by analyzing similar data. The generated target medical project includes not only specific chemotherapy regimens, but also may include required examinations, follow-up plans and possible side effect management measures.
[0086] In a specific embodiment, the process of executing step S104 may specifically include the following steps:
[0087] (1) Extracting appointment performance features from historical patient medical data to obtain an appointment performance feature set, where the appointment performance feature set includes: appointment time, number of cancellations, treatment frequency, visit rate, patient feedback, and daily activity patterns;
[0088] (2) Predict patient behavior patterns based on historical patient medical data based on the appointment performance feature set to obtain predicted behavior patterns;
[0089] (3) Predicting the patient status of the predicted behavior pattern to obtain predicted patient status data, wherein the predicted patient status data includes: patient compliance and appointment change probability;
[0090] (4) Generate a reminder schedule based on the probability of appointment changes, and generate patient visit reminders based on patient compliance;
[0091] (5) Generate reminder strategies through medical appointment reminders and reminder schedules.
[0092] Specifically, we must first ensure that relevant appointment performance features can be accurately extracted from historical patient medical data. We will access the central database to read a large number of historical patient appointment records, treatment records, follow-up records and other data. These data contain various behaviors and feedback information generated by patients during appointments and treatments. Using data mining and machine learning techniques, we conduct in-depth analysis of these historical data to extract useful appointment performance features. The appointment performance feature set includes multiple key indicators, such as appointment time, number of cancellations, treatment frequency, visit rate, patient feedback, and daily activity patterns. These features can reflect the patient's behavior patterns during appointments and treatments. For example, we will record the appointment time of each patient to analyze whether they prefer to choose morning or afternoon appointments; record the number of cancellations to understand whether patients often cancel or reschedule appointments; record treatment frequency and visit rate to evaluate patient compliance and on-time attendance; record patient feedback to analyze their satisfaction with treatment and any complaints; and record daily activity patterns to understand patients' living habits and activity levels. Next, based on these appointment performance feature sets, we will predict patient behavior patterns for historical patient medical data. Through machine learning algorithms, such as random forests, support vector machines, or neural networks, it is possible to identify the behavioral patterns of different types of patients. For example, it may be found that some patients are very punctual for their appointments and rarely cancel or reschedule, while other patients frequently cancel or postpone their appointments. By training on a large amount of data, a predictive model can be built that can predict the future behavior patterns of patients based on their performance characteristics. Once the predictive model is built, the data of new patients is predicted to obtain their behavior patterns. This predicted behavior pattern will be used for further patient status prediction. The predicted behavior pattern will be used to evaluate each patient's compliance and the probability of appointment change. Compliance refers to the likelihood that the patient will arrive on time and receive treatment as planned, while the probability of appointment change is the likelihood that the patient may cancel or reschedule an appointment. For example, it may be predicted that a patient has a 90% compliance, which means that he is likely to arrive on time, but there is also a 10% probability that he will cancel or reschedule his appointment. Based on these predictions, a reminder schedule is generated. For patients with a higher probability of appointment change, the frequency of reminders may be increased to ensure that the patient remembers to arrive on time. A detailed reminder plan is developed based on the patient's compliance and appointment change probability. For example, if a patient has low compliance, reminders may be sent a few days before the appointment and again on the day of the appointment to ensure that the patient does not miss the appointment. In addition, personalized appointment reminders are generated based on the patient's specific situation. These reminders can include notifying the patient in advance what materials to prepare, reminding the patient to take medication on time, and advising the patient to avoid certain activities. For example, for a patient who needs a blood test, he may be reminded not to eat the day before the appointment to ensure the accuracy of the test results.By taking into account the probability of appointment changes and patient compliance, a comprehensive reminder strategy is generated to ensure that each patient receives personalized reminder services. For example, suppose a patient, Mr. Li, has shown a high cancellation rate in the past few appointments and often postpones his appointments. By analyzing his appointment performance characteristics, it is predicted that he will have low compliance in future appointments and a high probability of change. Based on this prediction result, a detailed reminder schedule will be developed, sending reminders every day starting one week before the appointment and again on the morning of the appointment. In addition, specific appointment reminders will be generated, such as informing Mr. Li in advance what information he needs to bring and reminding him to stay fasting on the day of the appointment to ensure the smooth progress of chemotherapy and examinations. Through these personalized reminder strategies, Mr. Li's compliance can be greatly improved and the cancellation and rescheduling of appointments can be reduced.
[0093] In a specific embodiment, the process of executing step S105 may specifically include the following steps:
[0094] (1) Segment the target medical project into project nodes to obtain multiple project nodes;
[0095] (2) Perform intelligent agent matching on each project node to obtain the intelligent agent corresponding to each project node;
[0096] (3) Construct task data and resource data corresponding to the intelligent agent corresponding to each project node;
[0097] (4) Sort the nodes according to the task data and resource data corresponding to the intelligent agent corresponding to each project node to obtain the initial node sorting;
[0098] (5) The initial node ranking is input into the multi-agent algorithm for multiple rounds of iterations to obtain the target diagnosis and treatment node ranking table.
[0099] Specifically, the target medical project must first be broken down and analyzed in detail. The target medical project is a specific treatment plan for the patient, which includes multiple steps to be completed, such as preliminary consultation, physical examination, blood test, imaging examination, doctor evaluation and treatment plan confirmation. These steps will be divided into multiple specific project nodes, each of which represents an independent medical link. After the project node segmentation is completed, an intelligent agent will be matched for each project node. Intelligent agents are autonomous working units based on artificial intelligence technology that can independently handle and coordinate specific tasks. An intelligent agent will be assigned to each project node, and these agents will be responsible for the execution and management of specific project nodes. For example, the preliminary consultation node may be assigned a dedicated consultation agent to be responsible for the preliminary examination and information collection of the patient; the blood test node will be assigned a laboratory agent to arrange and manage the collection and analysis of blood samples. Next, task data and resource data need to be constructed for each intelligent agent. Task data includes specific tasks and operation steps that the intelligent agent needs to complete, such as the consultation agent needs to record the patient's medical history, symptom description and preliminary diagnosis information; the laboratory agent needs to arrange the collection, transportation and analysis of blood samples, etc. Resource data includes the resources required for the intelligent agent to perform tasks, such as the doctor's time, the use of treatment rooms, the availability of laboratory equipment, etc. Detailed task data and resource data will be constructed for each project node and the corresponding intelligent agent to ensure that each intelligent agent can efficiently complete the task it is responsible for. After the construction of task data and resource data is completed, each project node will be initially sorted. This step is to arrange all project nodes in a certain logical order based on the task and resource requirements of each intelligent agent. The initial sorting takes into account the dependencies between the various project nodes and the availability of resources to ensure that each node can be executed at the right time. For example, the initial consultation node may be ranked first because it is a key step to obtain basic patient information and preliminary diagnosis; followed by physical examinations and blood tests to ensure that doctors can obtain comprehensive diagnostic data when evaluating the patient's condition. After the initial node sorting, these sorting results will be input into Multi-Agent Systems (MAS) for multiple rounds of iterative optimization. Multi-Agent is an algorithm based on cooperative game theory that can find the optimal task execution order and resource allocation plan through coordination and cooperation between multiple intelligent agents. In a multi-agent system, each intelligent agent will continuously adjust its own strategy and try different task execution orders and resource allocation schemes to maximize the efficiency of the overall process and patient satisfaction. Through multiple rounds of iterations, the intelligent agents will gradually reach a consensus and find an optimal ranking table for diagnosis and treatment nodes. For example, suppose that a day chemotherapy center has developed a target medical program for a patient, including preliminary consultation, blood tests, imaging examinations, and treatment plan confirmation.First, divide this project into four project nodes, each of which corresponds to the above four steps. Then, assign intelligent agents to each project node: the consultation agent is responsible for the preliminary consultation, the laboratory agent is responsible for blood tests, the imaging agent is responsible for imaging examinations, and the treatment agent is responsible for treatment plan confirmation. Next, construct the task data and resource data of each intelligent agent. For example, the task data of the consultation agent includes recording medical history and symptom descriptions, and the resource data includes the doctor's time; the task data of the laboratory agent includes arranging the collection and analysis of blood samples, and the resource data includes the availability of laboratory equipment, etc. These project nodes are initially sorted to ensure that each node is executed in a reasonable order. Then, the initial sorting results are input into the multi-agent for iterative optimization. Through multiple rounds of iterations, each intelligent agent will adjust its own task execution order and resource allocation, and finally find an optimal diagnosis and treatment node sorting table. This sorting table may show that the preliminary consultation is scheduled at 8 am, followed by blood tests and imaging examinations, and finally the treatment plan is confirmed at 1 pm. This optimal sorting not only maximizes the utilization of resources, but also reduces the waiting time of patients and improves the efficiency of the overall process.
[0100] In a specific embodiment, the process of executing step S106 may specifically include the following steps:
[0101] (1) Extract data from the target diagnosis and treatment node ranking table to obtain the specific execution time;
[0102] (2) Modify the reminder schedule corresponding to the reminder strategy according to the specific execution time to obtain the target reminder schedule;
[0103] (3) Matching the reminder content of the target reminder schedule to obtain the target reminder content, wherein the target reminder content includes SMS content, phone notification content, and application notification content;
[0104] (4) Generate an appointment status report based on the target reminder content, and transmit the appointment status report to the target terminal corresponding to the patient's appointment request.
[0105] It should be noted that the specific execution time of each node needs to be extracted from the target diagnosis and treatment node sorting table. The target diagnosis and treatment node sorting table is generated through the optimization process of multi-agent and cooperative game theory, and contains the specific execution order and time arrangement of each diagnosis and treatment step. The execution time of each project node, including preliminary consultation, blood test, imaging examination and treatment plan confirmation, will be read from the sorting table. Once the specific execution time is extracted, this time information is applied to the reminder strategy. The reminder strategy is generated based on the prediction of patient behavior patterns and includes different types of reminders, such as SMS reminders, phone notifications, and application notifications. The original reminder schedule will be corrected according to the specific execution time of each project node to ensure that each reminder can be sent at the appropriate time. For example, if a project node is scheduled for a preliminary consultation at 8 am, reminders will be sent the night before and the morning of the same day to ensure that the patient does not forget the appointment. In this way, a target reminder schedule can be generated, accurately indicating the sending time of each reminder. Next, the target reminder schedule needs to be matched with reminder content. The reminder content is customized according to the specific medical project and patient needs to ensure that each reminder conveys accurate and useful information. For example, for patients who need a blood test, the reminder content may include specific preparations for the test, such as fasting requirements and bringing necessary medical records. The corresponding reminder content will be generated according to the specific requirements of each project node and the patient's situation. The reminder content can be sent through multiple channels, including SMS, phone calls, and application notifications. The diversity and flexibility of the reminder content will be ensured to provide patients with the most convenient reminder method. Based on the target reminder content, a detailed appointment status report will be generated. This report not only includes the specific execution time of each diagnosis and treatment step, but also includes all related reminder content and sending time. For example, the report may show that the preliminary consultation is scheduled at 8 am, and the reminder will be sent at 7 pm the previous night and 7 am on the same day; the blood test is scheduled at 9 am, and the reminder content includes fasting requirements and notifications to bring medical records, and the reminder will be sent at 9 pm the previous night and 8 am on the same day. In this way, the appointment status report can comprehensively display the entire diagnosis and treatment process and all related reminder information, providing patients with a clear schedule. Finally, the generated appointment status report will be transmitted to the patient's target terminal through a secure communication channel. The target terminal can be the patient's mobile phone, email, or the hospital's patient management application. The transmission process of the report will be ensured to be safe and reliable to avoid any information leakage or loss. For example, for patients who choose SMS notification, the encrypted SMS service will be used to ensure the security of the information; for patients who choose email notification, the report will be sent through a secure mail server; for patients who use the hospital app, the report will be sent directly to the patient's account through the notification function of the app.For example, suppose a 50-year-old male patient, Mr. Li, makes an appointment for a medical project that includes a preliminary consultation, blood test, imaging examination, and treatment plan confirmation. First, the specific execution time of each project node is extracted from the target diagnosis and treatment node sorting table: the preliminary consultation is scheduled at 8 am, the blood test is scheduled at 9 am, the imaging examination is scheduled at 10 am, and the treatment plan confirmation is scheduled at 1 pm. Then, these times are applied to the reminder strategy to generate a detailed target reminder schedule. For the preliminary consultation, reminders will be sent at 7 pm the previous night and 7 am the same day; for the blood test, reminders will be sent at 9 pm the previous night and 8 am the same day, including fasting requirements and notifications to bring medical records. Next, detailed reminder content is generated to ensure that each reminder conveys accurate and useful information. For example, the SMS reminder content may be: "Dear Mr. Li, please remember to go to the hospital for a preliminary consultation at 8 am tomorrow. Please bring your medical records." The phone notification content may be: "Hello, this is the hospital appointment reminder center, reminding you to have a blood test at 9 am tomorrow. Please fast after 9 pm tonight." The application notification content may be: "Mr. Li, you have an imaging examination appointment at 10 am tomorrow. Please make sure to arrive at the hospital in advance." Finally, a comprehensive appointment status report will be generated, including all execution times and reminder contents, and the report will be transmitted to Mr. Li's target terminal via encrypted SMS, email or hospital application. After receiving the report, Mr. Li can clearly understand the specific time and related preparations for each diagnosis and treatment step, ensuring that he can receive treatment on time and as required.
[0106] The embodiment of the present invention also provides a treatment reservation system for a day chemotherapy center, such as Figure 2 As shown, the appointment treatment system for a day chemotherapy center specifically includes:
[0107] The analysis module 201 is used to collect patient appointment requests and perform request scheduling analysis on the patient appointment requests to obtain appointment numbers and doctor schedules;
[0108] A matching module 202, for matching a target patient tag according to the patient appointment request, and extracting historical patient medical data through the target patient tag;
[0109] An extraction module 203 is used to extract the current diagnosis and treatment information corresponding to the patient appointment request, and match the target medical project based on the historical patient medical data and the current diagnosis and treatment information;
[0110] A prediction module 204, configured to predict a patient behavior pattern based on the historical patient medical data, obtain a predicted behavior pattern, and generate a reminder strategy based on the target medical item and the predicted behavior pattern;
[0111] The segmentation module 205 is used to segment the target medical project into project nodes to obtain multiple project nodes, input the multiple project nodes into a preset multi-agent algorithm to perform node sorting, and obtain a target diagnosis and treatment node sorting table;
[0112] The generation module 206 is used to generate an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmit the appointment status report to the target terminal corresponding to the patient appointment request.
[0113] Through the collaborative work of the above modules, by collecting patient appointment requests and performing request scheduling analysis with deep reinforcement learning algorithms, optimized appointment numbers and doctor schedules can be intelligently generated. This approach can not only efficiently handle a large number of appointment requests and maximize the resource utilization of doctors and treatment rooms, but also effectively reduce appointment conflicts and waiting time, greatly improving the overall efficiency and accuracy of appointments. Secondly, based on the patient appointment request, the target patient label is matched and historical patient medical data is extracted, so that the patient's key medical information can be accurately extracted through natural language processing and deep learning algorithms to generate detailed patient medical records. This not only improves the efficiency of information collection, but also ensures that doctors have comprehensive patient information as a reference when formulating treatment plans, providing a solid data foundation for personalized treatment. By matching the current diagnosis and treatment information and historical medical data corresponding to the patient's appointment request with the target medical project, the generative adversarial network (GANs) algorithm is used to generate high-quality personalized treatment plans, and through continuous optimization and evaluation, it is ensured that each treatment plan can fully consider the patient's specific situation and needs, thereby improving the accuracy and effectiveness of treatment. Machine learning and user behavior prediction algorithms are used to conduct in-depth analysis of historical patient medical data, extract appointment performance feature sets, and predict patient compliance and appointment changes. Based on these prediction results, a personalized reminder strategy is generated, including reminder frequency, content and method, to ensure that patients can receive diagnosis and treatment on time and as required. In this way, not only the accuracy of reminders is improved, but also personalized services can be provided according to individual differences, appointment cancellations and delays are reduced, and patient compliance and overall satisfaction are significantly improved. For the preparation and coordination before treatment, an algorithm based on multi-agent and cooperative game theory is introduced. Each preparation link is responsible for an independent intelligent agent to ensure seamless connection between each link and efficient use of resources. Through collaboration and optimization, the intelligent agent dynamically adjusts the resource allocation and process sequence of each link, improves the efficiency of the overall process, ensures that all preparations before treatment can be carried out efficiently and orderly, and lays a solid foundation for the chemotherapy process. The patient's real-time physiological data is efficiently analyzed in time series to predict future changes in physiological state. According to these prediction results, the DRL algorithm dynamically adjusts the monitoring and treatment parameters during chemotherapy. Once an abnormality is detected, the DRL algorithm will immediately calculate the optimal response strategy and notify the doctor to take corresponding measures. This method ensures the safety and effectiveness of the chemotherapy process and provides patients with comprehensive medical protection. In addition, by generating an appointment status report and transmitting it to the patient's target terminal, the patient can be informed of the appointment status and specific treatment arrangements in a timely manner, which improves the transparency of information and the patient's peace of mind. By integrating the target diagnosis and treatment node sorting table and reminder strategy, a detailed appointment status report is generated, including the execution time of each medical step, the items that the patient needs to prepare, the reminder time point and the specific reminder content.The appointment status report not only improves patients' understanding of the treatment process, but also ensures that patients can receive treatment on time and as required, further improving the success rate of treatment and patient satisfaction. Overall, the appointment treatment method of the day chemotherapy center has introduced advanced artificial intelligence algorithms and management methods, from appointments, information collection, personalized treatment plan formulation, appointment confirmation and reminders, pre-treatment preparation and coordination, to real-time monitoring and optimization during chemotherapy, which has comprehensively improved the operational efficiency, medical service quality and patient experience of the day chemotherapy center, and has significant innovation and practical application value.
[0114] The above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the embodiments, a person skilled in the art should understand that the specific implementation modes of the present invention can still be modified or replaced by equivalents, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.
Claims
1. A treatment method that is scheduled at a day chemotherapy center. It is characterized in that include: Collecting patient appointment requests, and performing request scheduling analysis on the patient appointment requests to obtain an appointment number and a doctor's schedule, specifically including: collecting the patient appointment requests, and performing data analysis on the patient appointment requests to obtain analysis data, wherein the analysis data includes: patient name, age and expected consultation time; collecting the current appointment data set in real time, wherein the appointment data set includes appointment requests of multiple patients, and at the same time, collecting doctor working hours; inputting the appointment requests of the multiple patients and the doctor working hours into a preset deep reinforcement learning algorithm to construct a simulation environment to obtain a simulated consultation environment; extracting parameter information of the simulated consultation environment, wherein the parameter information includes: state space, action space and coordinable space, wherein the state space includes appointment status, doctor's free time and free treatment rooms, the action space includes appointment arrangement, doctor scheduling and treatment room allocation, and the coordinable space includes appointment coordination, doctor coordination and treatment room coordination; according to the parameter information, performing request scheduling analysis on the analysis data in the simulated consultation environment to obtain the appointment number and the doctor's schedule; Matching the target patient label according to the patient appointment request, and extracting the historical patient medical data through the target patient label, specifically including: predicting the patient label of the parsed data to obtain the initial patient label; extracting similar case data based on the initial patient label, extracting common features of the similar case data to obtain the target feature; modifying the label of the initial patient label based on the target feature to obtain the target patient label; extracting the historical patient medical data from a preset historical patient database based on the target patient label; Extracting the current diagnosis and treatment information corresponding to the patient appointment request, and matching the target medical project based on the historical patient medical data and the current diagnosis and treatment information, specifically comprising: extracting the current diagnosis and treatment information corresponding to the patient appointment request, wherein the current diagnosis and treatment information includes: condition description, symptoms and diagnosis results; constructing a patient condition portrait for the condition description, the symptoms and the diagnosis results to obtain the patient condition portrait; extracting similar data from the historical patient medical data based on the patient condition portrait to obtain target similar data; matching the current diagnosis and treatment information with medical projects based on the target similar data to obtain the target medical project; Predicting patient behavior patterns of the historical patient medical data to obtain predicted behavior patterns, and generating reminder strategies based on the target medical items and the predicted behavior patterns, specifically including: extracting appointment performance features of the historical patient medical data to obtain an appointment performance feature set, wherein the appointment performance feature set includes: appointment time, number of cancellations, treatment frequency, visit rate, patient feedback, and daily activity patterns; predicting patient behavior patterns of the historical patient medical data based on the appointment performance feature set to obtain predicted behavior patterns; predicting patient status of the predicted behavior patterns to obtain predicted patient status data, wherein the predicted patient status data includes: patient compliance and appointment change probability; generating a reminder schedule based on the appointment change probability, and generating patient visit reminders according to the patient compliance; generating the reminder strategy through the visit reminders and the reminder schedule; The target medical project is segmented into project nodes to obtain a plurality of project nodes, and the plurality of project nodes are input into a preset multi-agent algorithm for node sorting to obtain a target diagnosis and treatment node sorting table, specifically comprising: segmenting the target medical project into project nodes to obtain a plurality of project nodes; performing intelligent agent matching on each of the project nodes respectively to obtain an intelligent agent corresponding to each of the project nodes; constructing task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes; performing node sorting according to the task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes to obtain an initial node sorting; inputting the initial node sorting into the multi-agent algorithm for multiple rounds of iterations to obtain the target diagnosis and treatment node sorting table; An appointment status report is generated based on the target diagnosis and treatment node ranking table and the reminder strategy, and the appointment status report is transmitted to the target terminal corresponding to the patient appointment request.
2. The day chemotherapy center appointment treatment method according to claim 1, It is characterized in that The step of generating an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmitting the appointment status report to the target terminal corresponding to the patient appointment request includes: Extracting data from the target diagnosis and treatment node sorting table to obtain specific execution time; Modify the reminder schedule corresponding to the reminder strategy according to the specific execution time to obtain a target reminder schedule; Matching the target reminder schedule with reminder content to obtain target reminder content, wherein the target reminder content includes text message content, phone notification content, and application notification content; An appointment status report is generated based on the target reminder content, and the appointment status report is transmitted to a target terminal corresponding to the patient appointment request.
3. A day chemotherapy center appointment treatment system, used to implement the day chemotherapy center appointment treatment method as claimed in any one of claims 1 to 2, It is characterized in that include: The analysis module is used to collect patient appointment requests and perform request scheduling analysis on the patient appointment requests to obtain an appointment number and a doctor's schedule, specifically including: collecting the patient appointment requests and performing data analysis on the patient appointment requests to obtain analysis data, wherein the analysis data includes: patient name, age and expected consultation time; collecting the current appointment data set in real time, wherein the appointment data set includes appointment requests of multiple patients, and at the same time, collecting doctor working hours; inputting the appointment requests of the multiple patients and the doctor working hours into a preset deep reinforcement learning algorithm to construct a simulation environment to obtain a simulated consultation environment; extracting parameter information of the simulated consultation environment, wherein the parameter information includes: state space, action space and coordinable space, wherein the state space includes appointment status, doctor's free time and idle treatment rooms, the action space includes appointment arrangement, doctor scheduling and treatment room allocation, and the coordinable space includes appointment coordination, doctor coordination and treatment room coordination; according to the parameter information, performing request scheduling analysis on the analysis data in the simulated consultation environment to obtain the appointment number and the doctor's schedule; A matching module is used to match the target patient label according to the patient appointment request, and extract the historical patient medical data through the target patient label, specifically including: predicting the patient label of the parsed data to obtain the initial patient label; extracting similar case data based on the initial patient label, extracting common features of the similar case data to obtain the target feature; modifying the label of the initial patient label based on the target feature to obtain the target patient label; extracting the historical patient medical data from a preset historical patient database based on the target patient label; An extraction module is used to extract the current diagnosis and treatment information corresponding to the patient appointment request, and match the target medical project based on the historical patient medical data and the current diagnosis and treatment information, specifically comprising: extracting the current diagnosis and treatment information corresponding to the patient appointment request, wherein the current diagnosis and treatment information includes: a description of the condition, symptoms, and a diagnosis result; constructing a patient condition portrait for the description of the condition, the symptoms, and the diagnosis result to obtain a patient condition portrait; extracting similar data from the historical patient medical data based on the patient condition portrait to obtain target similar data; matching the current diagnosis and treatment information with medical projects based on the target similar data to obtain the target medical project; A prediction module, used for predicting patient behavior patterns of the historical patient medical data to obtain predicted behavior patterns, and generating reminder strategies based on the target medical items and the predicted behavior patterns, specifically comprising: extracting appointment performance features of the historical patient medical data to obtain an appointment performance feature set, wherein the appointment performance feature set includes: appointment time, number of cancellations, treatment frequency, visit rate, patient feedback, and daily activity patterns; predicting patient behavior patterns of the historical patient medical data based on the appointment performance feature set to obtain predicted behavior patterns; predicting patient status of the predicted behavior patterns to obtain predicted patient status data, wherein the predicted patient status data includes: patient compliance and appointment change probability; generating a reminder schedule based on the appointment change probability, and generating patient visit reminders according to the patient compliance; generating the reminder strategy through the visit reminders and the reminder schedule; The segmentation module is used to segment the project nodes of the target medical project to obtain multiple project nodes, input the multiple project nodes into a preset multi-agent algorithm for node sorting, and obtain a target diagnosis and treatment node sorting table, specifically including: segmenting the project nodes of the target medical project to obtain multiple project nodes; performing intelligent agent matching on each of the project nodes respectively to obtain an intelligent agent corresponding to each of the project nodes; constructing task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes; sorting nodes according to the task data and resource data corresponding to the intelligent agent corresponding to each of the project nodes to obtain an initial node sorting; inputting the initial node sorting into the multi-agent algorithm for multiple rounds of iterations to obtain the target diagnosis and treatment node sorting table; A generation module is used to generate an appointment status report based on the target diagnosis and treatment node ranking table and the reminder strategy, and transmit the appointment status report to the target terminal corresponding to the patient appointment request.
Citation Information
Patent Citations
Hemodialysis patient self-service dialysis treatment reservation system
CN109411061A
Dynamic intervention method and device for treatment strategy, electronic equipment and storage medium
CN112927812A
Health information generation method and device, medium and equipment
CN116844725A
