A follow-up method, device, equipment, and storage medium based on obstetric diagnosis and treatment information.
By integrating multimodal obstetric data and using a specialized obstetric language model to generate personalized follow-up plans and provide early warnings of abnormalities, the system addresses the issues of low efficiency and insufficient personalization in existing obstetric follow-up and rehabilitation guidance, achieving personalized and dynamic maternal and infant health management throughout the entire lifecycle.
Patent Information
- Application Number
- CN202611267092.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-20
- Publication Date
- 2026-11-17
AI Technical Summary
Current obstetric follow-up and rehabilitation guidance suffers from low follow-up efficiency, untimely and non-standardized content, lack of personalization, failure to fully utilize data from the entire process, inability to accurately identify health problems, and lack of dynamism and personalization in rehabilitation guidance, failing to meet personalized needs throughout the entire cycle.
By collecting and integrating obstetric multimodal data, a personalized follow-up plan is generated using a specialized obstetric language model. An abnormality warning is then issued using an analytical model, and a personalized health management plan is generated. Furthermore, a multi-dimensional evaluation mechanism is used to dynamically adjust the follow-up and management plans.
It achieves precision, timeliness, and effectiveness in personalized follow-up and rehabilitation guidance throughout the entire life cycle, reduces the workload of medical staff, and ensures dynamic closed-loop management of maternal and infant health.
Smart Images

Figure CN123120184A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of medical information technology, and in particular to a follow-up method, device, equipment and storage medium based on obstetric diagnosis and treatment information. Background Technology
[0002] Obstetric care is a comprehensive process encompassing four core stages: preconception checkups, prenatal checkups, labor management, and postpartum rehabilitation. Follow-up and rehabilitation guidance are crucial for ensuring maternal and infant health, enabling the timely detection of pregnancy abnormalities, labor complications, and postpartum issues, thereby improving treatment outcomes. However, current obstetric follow-up and rehabilitation guidance systems have many shortcomings and cannot meet the full-cycle, personalized needs.
[0003] Current follow-up methods rely heavily on manual telephone follow-ups and in-person visits by medical staff, which suffers from low efficiency, untimely follow-ups, and non-standard follow-up content. The follow-up content is mostly based on a general template and does not take into account the individual circumstances of pregnant women (such as gestational age, pregnancy complications, delivery method, etc.) to develop personalized follow-up plans, resulting in insufficient targeting of follow-ups and an inability to accurately identify health problems of pregnant women and newborns.
[0004] In terms of rehabilitation guidance, existing technologies are mostly static text-based, lacking dynamism and interactivity, and cannot adjust the guidance plan in real time according to the pregnant woman's rehabilitation progress and physical condition. At the same time, the rehabilitation guidance content does not incorporate data from the entire obstetric diagnosis and treatment process (such as prenatal check-up data, delivery data, and postpartum examination data), and the guidance basis is insufficient, making it difficult to meet the personalized rehabilitation needs of pregnant women with different delivery methods and different complications.
[0005] In addition, the existing system does not connect to the entire process of obstetric diagnosis and treatment data (such as multimodal data, medical record data, etc.), resulting in a prominent data silo problem. Follow-up and rehabilitation guidance cannot make full use of past diagnosis and treatment data, leading to a lack of targeted guidance plans. At the same time, the lack of a rehabilitation effect evaluation mechanism makes it impossible to understand the effect of rehabilitation guidance in a timely manner, making it difficult to adjust the guidance plan and affecting the rehabilitation effect. Summary of the Invention
[0006] The purpose of this application is to propose a follow-up method, device, equipment, and storage medium based on obstetric diagnosis and treatment information. Its main purpose is to connect with multimodal data of the entire obstetric process, generate personalized follow-up plans using obstetric-specific large language models, combine analysis models for abnormal early warning, generate personalized guidance plans using health conditioning models, and achieve dynamic closed-loop adjustment of follow-up and conditioning plans through multi-dimensional evaluation mechanisms, thereby improving the accuracy, timeliness, and effectiveness of follow-up and rehabilitation guidance and ensuring the health of mothers and infants throughout the entire life cycle.
[0007] To address the aforementioned technical problems, this application provides a follow-up method based on obstetric diagnosis and treatment information, employing the following technical solution: A follow-up method based on obstetric diagnosis and treatment information includes: S1, collects and integrates obstetric multimodal data; S2, Based on the obstetric multimodal data, a personalized full-cycle follow-up plan is generated using a pre-trained obstetric-specific large language model; S3, execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data; S4. Based on the obstetric multimodal data and the follow-up data, generate a personalized health conditioning reference plan; S5. The health conditioning effect is evaluated from multiple dimensions, and the health conditioning reference plan and the full-cycle follow-up plan are dynamically adjusted based on the evaluation results.
[0008] Furthermore, in step S1, the obstetric multimodal data includes preconception data, pregnancy data, labor data, and postpartum data; The collection and integration of obstetric multimodal data specifically includes: automatically collecting the obstetric multimodal data from the hospital through an interface; and structuring and extracting key health indicators from the collected obstetric multimodal data.
[0009] Furthermore, in step S2, the method for constructing the obstetric-specific large language model includes: Based on obstetric medical materials, a general large language model is trained for obstetric specialties. During the training process, the model’s ability to recognize obstetric specialties and extract information correlation is optimized to ensure that the model’s accuracy and recall rate for extracting obstetric information are greater than or equal to the preset values.
[0010] Furthermore, in step S2, a personalized full-cycle follow-up plan is generated, specifically including: Based on the aforementioned obstetric multimodal data, a personalized follow-up plan is generated using the pre-trained obstetric-specific large language model, which includes follow-up frequency, follow-up content, follow-up method, and follow-up reminder time. The personalized follow-up plan will be dynamically updated based on the pregnant woman's treatment progress, changes in health status, and feedback from medical staff.
[0011] Further, step S3 specifically includes: Before the follow-up time arrives, follow-up reminders containing the follow-up time, content, and method are sent to pregnant women and medical staff through multiple channels. For the first type of pre-set follow-up content, follow-up data is automatically collected through online questionnaires or smart devices; for the second type of pre-set follow-up content, offline follow-up appointments are automatically scheduled. Record data during the follow-up process, call the analysis model to analyze whether there are any abnormalities in the follow-up data; if an abnormality is found, it will be automatically marked and medical staff will be alerted to intervene.
[0012] Further, step S4 specifically includes: By calling the obstetric rehabilitation guidance model and combining the obstetric multimodal data and follow-up data, a personalized guidance plan containing text, images and videos is generated. The system receives real-time feedback and follow-up data from pregnant women regarding their health management practices. If it detects that the treatment is not effective or that the woman's physical condition has changed, it automatically adjusts the health management reference plan.
[0013] Furthermore, step S5 specifically includes: For each stage of treatment, corresponding evaluation indicators are set. By collecting conditioning and follow-up data, and comparing them with evaluation indicators and standards, a preliminary evaluation result is generated. Medical staff then conduct a comprehensive evaluation of the health conditioning effect based on offline follow-up visits and feedback from pregnant women, and revise the preliminary evaluation result accordingly. The treatment effect level is determined based on the revised assessment results; if the treatment effect level is lower than the preset threshold, the guidance plan adjustment process is automatically triggered to remind medical staff to provide targeted intervention suggestions.
[0014] A follow-up device based on obstetric diagnosis and treatment information, characterized in that it includes: The end-to-end data integration module is used to collect and integrate obstetric multimodal data; The personalized follow-up plan generation module is used to generate a personalized full-cycle follow-up plan based on the obstetric multimodal data and using a pre-trained obstetric-specific large language model. The intelligent follow-up execution module is used to execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data; The personalized health management module is used to generate personalized health management reference plans based on the obstetric multimodal data and the follow-up data. The conditioning effect evaluation module is used to evaluate the health conditioning effect from multiple dimensions and dynamically adjust the health conditioning reference plan and the full-cycle follow-up plan based on the evaluation results.
[0015] A computer device includes a memory and a processor. The memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the follow-up method based on obstetric diagnosis and treatment information described above.
[0016] A computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of the follow-up method based on obstetric diagnosis and treatment information described above.
[0017] Compared with existing technologies, the embodiments of this application have the following main advantages: This application integrates multimodal obstetric data from preconception to postpartum, and uses a dedicated large language model to generate personalized full-cycle follow-up plans covering frequency, content, and methods, filling the gap in single-stage management. During the follow-up process, the system analyzes data in real time and automatically issues early warnings of abnormalities, reminding medical staff to intervene to ensure maternal and infant safety. Simultaneously, this application combines diagnosis and follow-up data to generate health conditioning reference plans, and evaluates the conditioning effects from multiple dimensions. When the effect falls below a threshold, the plan is automatically adjusted, forming a dynamic closed-loop mechanism of "follow-up-conditioning-evaluation-adjustment," ensuring continuous optimization of the guidance plan. Furthermore, the fully automated execution of follow-up tasks, reminders, and data collection significantly shortens the time for manual intervention, effectively reduces the burden on medical staff, and comprehensively improves the standardization and overall efficiency of follow-up. Attached Figure Description
[0018] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is an exemplary system architecture diagram to which this application can be applied; Figure 2 This is a flowchart of an embodiment of the follow-up method based on obstetric diagnosis and treatment information according to this application; Figure 3 This is a schematic diagram of a structure of an embodiment of the follow-up device based on obstetric diagnosis and treatment information according to this application; Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation
[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.
[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.
[0023] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.
[0024] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.
[0025] Terminal devices 101, 102, and 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 players (Moving Picture Experts Group Audio Layer IV), laptops, and desktop computers, etc.
[0026] Server 105 can be a server that provides various services, such as a backend server that supports the pages displayed on terminal devices 101, 102, and 103.
[0027] It should be noted that the follow-up method based on obstetric diagnosis and treatment information provided in this application embodiment is generally executed by a server, and correspondingly, the follow-up device based on obstetric diagnosis and treatment information is generally set in a terminal device.
[0028] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.
[0029] Continue to refer to Figure 2 This paper illustrates a follow-up method based on obstetric diagnosis and treatment information according to the present application. This method can run on a local terminal device or a server. When the method runs on a local terminal device, the local terminal device can execute the follow-up method based on obstetric diagnosis and treatment information provided in the embodiments of this application; when the method runs on a server, the method can be implemented through data interaction between the server and the local terminal device.
[0030] The aforementioned follow-up method based on obstetric diagnosis and treatment information includes the following steps: S1. Collect and integrate obstetric multimodal data; In this step, multimodal data from the entire obstetric diagnosis and treatment process are collected. This obstetric multimodal data includes pre-pregnancy data, pregnancy data, labor data, and postpartum data, as detailed below: Preconception data can include preconception test results such as blood routine, urine routine, blood sugar and thyroid function, medical history such as chronic diseases such as hypertension and diabetes, and reproductive history such as previous number of pregnancies, delivery method and miscarriage history.
[0031] Pregnancy data can include prenatal examination data such as blood pressure, weight, fundal height and abdominal circumference; ultrasound images such as fetal structural screening images and growth assessment images; fetal heart rate monitoring data; information on pregnancy complications such as gestational diabetes, gestational hypertension and placenta previa; as well as doctor-patient dialogue data and clinical decision-making recommendations.
[0032] Labor data can include delivery methods such as vaginal delivery, cesarean section, and forceps delivery, duration of labor, uterine contractions, fetal heart rate monitoring data, and labor complications such as postpartum hemorrhage and shoulder dystocia.
[0033] Postpartum data can include postpartum examination data such as wound healing and uterine involution, newborn health data such as Apgar scores and weight changes, breastfeeding status, and postpartum rehabilitation indicators such as pelvic floor muscle function assessment, weight recovery and emotional state assessment.
[0034] The collection and integration of obstetric multimodal data specifically includes: automatically collecting obstetric multimodal data from hospitals via API interfaces. Data sources include internal medical institutions' systems such as HIS, LIS, PACS, and electronic medical record systems, as well as external data interfaces. Encrypted transmission methods are used to ensure data security and comply with medical data privacy protection requirements. The collected obstetric multimodal data is structured and organized to establish a full-cycle health record for pregnant women. New diagnostic and treatment data is received in real time, and the health record is automatically updated to ensure the timeliness and completeness of the data. Simultaneously, key health indicators such as blood pressure, blood sugar, fetal heart rate, and uterine contraction frequency are extracted using information extraction algorithms for the generation of subsequent follow-up plans and health management reference plans.
[0035] S2. Based on the obstetric multimodal data, a personalized full-cycle follow-up plan is generated using a pre-trained obstetric-specific large language model. In this step, based on the multimodal obstetric data collected and integrated in step S1, a pre-trained obstetric-specific large language model is used to generate a personalized full-cycle follow-up plan.
[0036] The method for constructing the obstetric-specific large language model specifically includes: General-purpose large language models such as LLaMA, ChatGLM, or GPT series can be used as base models for optimized training in obstetric scenarios.
[0037] By designing obstetric-specific prompt word templates, the extraction targets, extraction formats, and professional terminology standards for information such as gestational age, fetal heart rate, chief complaints, and complications are clearly defined. This guides the large language model to accurately extract obstetric-related information. The prompt word templates can be dynamically adjusted to adapt to different types of medical materials.
[0038] We collected a massive amount of obstetric medical materials, including text medical records, ultrasound reports, laboratory reports, and speech-to-text transcripts, as training data to fine-tune the base model. During training, we focused on optimizing the recognition and extraction capabilities of obstetric specialty terminology, as well as the extraction of information correlations such as the association between complications and symptoms, and the association between laboratory indicators and diseases. We also adopted a combination of supervised and semi-supervised learning to improve the model's generalization ability and avoid overfitting.
[0039] A test set containing various obstetric multimodal materials was used to validate and optimize the fine-tuned large language model, ensuring that the model's accuracy and recall in extracting obstetric information were greater than or equal to preset performance thresholds. These preset performance thresholds were set according to the needs of obstetric clinical applications, and were exemplarily set to an accuracy threshold of no less than 96% and a recall threshold of no less than 95%. When the model's accuracy in extracting obstetric information reached 96% or higher and its recall reached 95% or higher, the model's performance was deemed to meet the requirements. If the above thresholds were not met, fine-tuning training and parameter optimization continued, and a combination of test set and cross-validation was used for evaluation to ensure the reliability of the validation results.
[0040] The generation of a personalized, full-cycle follow-up plan specifically includes: By dividing follow-up into four core stages, covering the entire process of obstetric diagnosis and treatment, we include preconception follow-up, pregnancy follow-up, labor follow-up, and postpartum follow-up. Preconception follow-up is for women preparing for pregnancy and includes preconception check-up, guidance on healthy lifestyles, and suggestions for disease intervention. Pregnancy follow-up is tailored to different frequencies and contents based on gestational age and pregnancy complications. Labor follow-up includes real-time follow-up during labor to monitor the progress of labor, maternal and infant status, and timely feedback on abnormalities. Postpartum follow-up includes key follow-up within 42 days postpartum, including maternal recovery, newborn health, and breastfeeding status, and long-term postpartum follow-up (1 year) to monitor the long-term health of mother and child.
[0041] Based on the aforementioned obstetric multimodal data, a pre-trained obstetric-specific large language model is used to generate a plan that includes follow-up frequency, follow-up content, follow-up methods, and follow-up reminder times. For example, for high-risk pregnant women such as those with gestational hypertension or gestational diabetes, the follow-up frequency is higher than that of ordinary pregnant women, with a focus on blood pressure and blood sugar-related indicators, and follow-up methods including telephone follow-up, online questionnaire follow-up, and offline follow-up visits.
[0042] The plan is automatically updated based on the progress of the pregnant woman's treatment and changes in her health status, and manually adjusted in conjunction with feedback from medical staff. For example, if the pregnant woman's blood pressure remains high, the follow-up frequency is automatically adjusted from once a week to once every three days. Medical staff can also manually add follow-up items based on clinical judgment to ensure timeliness.
[0043] S3. Execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data; This step allows for automated execution of follow-up tasks, improving follow-up efficiency, ensuring timely and standardized follow-ups, and reducing the workload of medical staff.
[0044] The implementation of the full-cycle follow-up plan specifically includes: Before the scheduled follow-up time, reminders are sent to pregnant women and medical staff through multiple channels, including SMS, app push notifications, and phone calls, clearly indicating the follow-up time, content, and method to avoid omissions. For the first type of simple follow-up content, such as pre-set fetal movement monitoring during pregnancy, postpartum temperature monitoring, and blood pressure measurement, follow-up data is automatically collected through online questionnaires or smart devices such as fetal heart monitors, thermometers, and smart blood pressure monitors, without manual intervention from medical staff. For the second type of complex follow-up content, such as postpartum wound recovery examination and newborn development assessment, offline follow-up appointments are automatically scheduled, and pregnant women are reminded to attend on time.
[0045] The system automatically records data during the follow-up process, such as symptoms reported by pregnant women, examination results, and data collected by smart devices, and calls the analysis model to analyze whether there are any abnormalities in the follow-up data.
[0046] The analytical model described is an obstetric follow-up anomaly detection model built based on machine learning algorithms. Its construction method includes: collecting historical obstetric follow-up data and distinguishing between normal and abnormal samples; extracting structured features (such as blood pressure trends, fetal heart rate fluctuation patterns, weight gain rates, and blood glucose fluctuation amplitudes); training a classification model (such as random forests, gradient boosting trees, and deep neural networks); and outputting a judgment result indicating whether abnormalities exist in the follow-up data and the type of abnormality (such as abnormal fetal movement, postpartum hypertension, and abnormal blood glucose). During training, abnormal samples are oversampled or normal samples are undersampled to balance the training set and ensure the model's sensitivity to abnormal situations. The model is validated using a test set not used in training, with performance standards including an anomaly detection sensitivity of no less than 90% and a specificity of no less than 85%.
[0047] If the analysis model detects anomalies, it automatically marks the abnormal data and promptly alerts medical staff for intervention. For example, when the blood pressure data collected by the smart blood pressure monitor exceeds a preset threshold (such as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg), it is automatically marked as abnormal blood pressure and a warning notification is sent to medical staff. When fetal movement monitoring data is below the normal range (such as the system identifying that the total number of fetal movements in the pregnant woman is only 15 times in 12 hours, or if the fetal movement frequency in a specific time period of the day decreases by more than 50% compared to the average fetal movement frequency in the same period of the previous 3 days), it is automatically marked as abnormal fetal movement and medical staff are reminded to conduct further examination.
[0048] S4. Based on the obstetric multimodal data and the follow-up data, generate a personalized health conditioning reference plan; In this step, based on the obstetric multimodal data collected in step S1 and the follow-up data collected in step S3, the obstetric rehabilitation guidance model is invoked to generate a personalized health conditioning reference plan, covering pre-pregnancy conditioning, prenatal care, and postpartum rehabilitation, to achieve dynamic guidance.
[0049] The obstetric rehabilitation guidance model is a specialized guidance model built upon a large language model. Its construction method includes: using an obstetric-specific large language model as a foundation, collecting professional knowledge in the field of obstetric rehabilitation guidance (including guidelines and literature on dietary management, exercise guidance, postpartum rehabilitation, breastfeeding, and emotional regulation), cleaning and organizing the content, and then constructing an obstetric rehabilitation guidance knowledge base; fine-tuning the large language model based on this knowledge base, enabling the model to generate structured, personalized health management reference plans based on the pregnant woman's full-cycle data (such as delivery method, complications, postpartum rehabilitation indicators, etc.) and follow-up data; during the fine-tuning process, a human feedback reinforcement learning method is used, with obstetric experts providing quality assessment and feedback on the guidance plans generated by the model, further optimizing the model's output quality and ensuring the scientific validity and practicality of the guidance plans.
[0050] The dynamic guidance may include preconception conditioning guidance, prenatal care guidance, and postpartum recovery guidance; among which... Preconception conditioning guidance can combine preconception examination data to provide suggestions on diet conditioning, exercise guidance, work and rest adjustment, disease intervention, etc., to help women preparing for pregnancy improve their physical fitness; Pregnancy care guidance can provide dietary and nutritional guidance, fetal movement monitoring guidance, pregnancy exercise guidance, and complication prevention guidance, based on the gestational age and the pregnant woman's health status. Postpartum rehabilitation guidance can be combined with delivery method (vaginal delivery, cesarean section) and postpartum examination data to provide guidance on wound recovery, pelvic floor muscle rehabilitation, breastfeeding, dietary conditioning, and emotional regulation.
[0051] The specific method for generating a personalized health conditioning reference plan is as follows: call the obstetric rehabilitation guidance model, combine the pregnant woman's full-cycle data and follow-up data, and generate a personalized guidance plan that includes pictures, text and videos and is easy to understand, so that pregnant women can understand and implement it. The focus is adjusted according to different rehabilitation stages, such as focusing on wound care in the early postpartum period and focusing on pelvic floor muscle rehabilitation in the later period.
[0052] In this embodiment, the system can also receive real-time feedback and follow-up data from pregnant women. If it detects poor treatment results such as slow progress in pelvic floor muscle recovery or changes in physical condition such as postpartum depression, the system will automatically adjust the treatment plan and optimize the content in conjunction with the guidance of medical staff to ensure effectiveness.
[0053] S5. Conduct multi-dimensional evaluation of the health conditioning effect, and dynamically adjust the health conditioning reference plan and the full-cycle follow-up plan based on the evaluation results.
[0054] In this step, a multi-dimensional rehabilitation effectiveness evaluation mechanism is established to assess the effectiveness of rehabilitation guidance in real time and provide a basis for adjusting the plan.
[0055] The aforementioned multi-dimensional evaluation of the health conditioning effects specifically includes the setting of evaluation indicators, specific evaluation methods, and processing of evaluation results; among which... The assessment indicators are set according to each stage of the treatment, such as pre-pregnancy conditioning assessment, prenatal health care assessment, and postpartum recovery assessment. Among them, the pre-pregnancy conditioning assessment includes physical indicators (such as BMI, blood pressure, blood sugar) and disease improvement; the prenatal health care assessment includes fetal development (such as ultrasound indicators such as biparietal diameter, femur length, etc.), maternal health indicators (blood pressure, blood sugar, etc.), and complication control; the postpartum recovery assessment includes wound healing, pelvic floor muscle function, breastfeeding effect, emotional state, and weight recovery.
[0056] The assessment method can combine automated and manual evaluation. Automated evaluation involves collecting treatment data and follow-up data (such as rehabilitation data collected by smart devices and data recorded during follow-up visits), comparing them with evaluation indicator standards, and generating preliminary assessment results. Manual evaluation involves medical staff conducting a comprehensive assessment of the health treatment effect based on offline follow-up visits and feedback from pregnant women, correcting the automated assessment results, and ensuring the accuracy and comprehensiveness of the evaluation.
[0057] The assessment results are processed by classifying the treatment effect into four levels, such as "excellent," "good," "average," and "poor," based on the revised assessment results. If the treatment effect level is lower than a preset assessment threshold, the guidance plan adjustment process is automatically triggered, reminding medical staff to provide targeted intervention suggestions.
[0058] The preset assessment thresholds are set based on obstetric clinical guidelines and rehabilitation management experience. Specifically, the treatment effect level is quantified into a score system, for example: Excellent (90-100 points), Good (75-89 points), Average (60-74 points), and Poor (below 60 points). The preset assessment threshold is the lower limit of the "Good" level, i.e., 75 points. When the treatment effect score is below 75 points (i.e., the assessment level is "Average" or "Poor"), the treatment effect level is determined to be below the preset assessment threshold, and the system automatically triggers the guidance plan adjustment process, including: adjusting the content of the health treatment reference plan (such as adding rehabilitation training programs, adjusting the diet plan, etc.), adjusting the follow-up frequency and content of the full-cycle follow-up plan (such as increasing the number of follow-ups, adding key monitoring indicators), and reminding medical staff to pay special attention to the pregnant woman and provide targeted intervention suggestions.
[0059] In addition, regular rehabilitation effectiveness assessment reports are generated and provided to pregnant women and medical staff to facilitate understanding of rehabilitation progress. A closed-loop management mechanism of "follow-up-treatment-assessment-adjustment" ensures continuous optimization of the follow-up and treatment plans.
[0060] See Figure 3As shown, in order to solve the above-mentioned technical problems, this application embodiment also provides a follow-up system 300 based on obstetric diagnosis and treatment information, including a full-process data docking module 310, a personalized follow-up plan generation module 320, an intelligent follow-up execution module 330, a personalized health conditioning module 340, and a conditioning effect evaluation module 350.
[0061] The end-to-end data integration module 310 is used to collect and integrate obstetric multimodal data; the obstetric multimodal data includes preconception data, pregnancy data, labor data and postpartum data; it automatically collects obstetric multimodal data from hospitals through API interface, structures and organizes the collected data, extracts key health indicators, and establishes a full-cycle health record for pregnant women.
[0062] The personalized follow-up plan generation module 320 is used to generate a personalized full-cycle follow-up plan based on the obstetric multimodal data and a pre-trained obstetric-specific large language model. The obstetric-specific large language model is constructed by training a general large language model base on obstetric medical materials to ensure that the model's accuracy and recall rate for extracting obstetric information are greater than or equal to a preset performance index threshold. The threshold can be set to an accuracy rate of not less than 96% and a recall rate of not less than 95%. The generated personalized follow-up plan includes follow-up frequency, follow-up content, follow-up method, and follow-up reminder time. The follow-up plan is dynamically updated according to the pregnant woman's treatment progress, changes in health status, and feedback from medical staff.
[0063] The intelligent follow-up execution module 330 is used to execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data. Specifically, it includes: sending follow-up reminders to pregnant women and medical staff through multiple channels before the follow-up time arrives; automatically collecting follow-up data through online questionnaires or smart devices for the preset first type of follow-up content; automatically scheduling offline follow-up appointments for the preset second type of follow-up content; recording data during the follow-up process; calling the analysis model to analyze whether there are any anomalies in the follow-up data; and automatically marking and reminding medical staff to intervene if an anomaly is found.
[0064] The personalized health conditioning module 340 is used to generate a personalized health conditioning reference plan based on the obstetric multimodal data and the follow-up data; call the obstetric rehabilitation guidance model, combine the full-cycle data and follow-up data to generate a personalized guidance plan including pictures, text and videos; receive the pregnant woman's conditioning feedback and follow-up data in real time, and automatically adjust the health conditioning reference plan if it is found that the conditioning effect is not good or the physical condition changes.
[0065] The conditioning effect evaluation module 350 is used to evaluate the health conditioning effect from multiple dimensions and dynamically adjust the health conditioning reference plan and the full-cycle follow-up plan based on the evaluation results. It sets corresponding evaluation indicators for each conditioning stage, collects conditioning data and follow-up data, compares the evaluation indicator standards to generate preliminary evaluation results, which are then corrected by medical staff. The conditioning effect level is then classified based on the corrected evaluation results. If the conditioning effect level is lower than a preset evaluation threshold (e.g., a score below 75), the guidance plan adjustment process is automatically triggered, reminding medical staff to provide targeted intervention suggestions.
[0066] The follow-up system based on obstetric diagnosis and treatment information in this embodiment achieves the collection and integration of data throughout the entire obstetric diagnosis and treatment process, intelligent generation and dynamic updating of personalized follow-up plans, automatic early warning of abnormal follow-up data, generation and dynamic adjustment of personalized health conditioning plans, and multi-dimensional rehabilitation effect evaluation and closed-loop plan adjustment through the collaborative work of the above modules. Under the premise of ensuring maternal and infant safety, it improves the accuracy, timeliness and effectiveness of follow-up and rehabilitation guidance.
[0067] This application also provides specific implementation examples of a follow-up method based on obstetric diagnosis and treatment information;
[0068] Ms. Li, a pregnant woman, had no history of hypertension before pregnancy. She was diagnosed with gestational hypertension at 28 weeks of pregnancy. The system connected her full-process data: pre-pregnancy examination data (normal blood pressure), prenatal examination data (blood pressure 150 / 100 mmHg, urine protein +), and clinical decision-making recommendations (regular blood pressure monitoring and oral antihypertensive drugs are required).
[0069] The personalized follow-up plan generation module 320 combines the above data and uses a large language model specifically designed for obstetrics to generate follow-up plans: 1. Pregnancy follow-up: Weekly follow-up, focusing on monitoring blood pressure and urine protein, using online questionnaires and smart blood pressure monitors to collect data, and monthly offline follow-up visits; 2. Labor follow-up: Follow-up every 30 minutes during labor, monitoring blood pressure and fetal heart rate data, and timely feedback on abnormalities; 3. Postpartum follow-up: Daily follow-up within the first week postpartum, monitoring blood pressure and postpartum wound recovery, every 2 weeks from 2 to 6 weeks postpartum, and monthly follow-up from 7 to 12 months postpartum, monitoring blood pressure and long-term health.
[0070] The intelligent follow-up execution module 330 sent follow-up reminders on time and automatically collected Ms. Li's blood pressure data through the intelligent blood pressure monitor. Two weeks postpartum, the analysis model detected that her blood pressure was still high (145 / 95 mmHg), with a systolic blood pressure ≥140 mmHg, exceeding the preset abnormal threshold. It was automatically marked as abnormal blood pressure, prompting medical staff to intervene.
[0071] The personalized health management module 340 utilizes the obstetric health management model, combining Ms. Li's full-cycle and follow-up data to generate a guidance plan: a low-salt, low-fat diet, appropriate daily exercise (such as brisk walking), timely administration of antihypertensive medication, and emotional regulation. The guidance content is dynamically adjusted based on blood pressure changes (such as increasing exercise duration). The guidance is presented in text, images, and videos for easy understanding and implementation by the pregnant woman.
[0072] The treatment effect assessment module evaluates Ms. Li's recovery progress monthly. Three months postpartum, the automatic assessment unit compares collected blood pressure and follow-up data with assessment standards to generate a preliminary assessment result. The manual assessment correction unit, staffed by medical personnel, refines the result based on in-person follow-up visits. The revised assessment result shows: blood pressure reduced to 135 / 85 mmHg, treatment effect score of 82 (good level, above the preset assessment threshold of 75), and the guidance plan adjusted to reduce the dosage of antihypertensive medication and change the follow-up frequency to once every 3 weeks. Ms. Li consults on dietary issues through the doctor-patient interaction platform, receiving real-time responses from medical staff, improving her adherence to rehabilitation.
[0073] Pregnant woman Zhang underwent a cesarean section due to breech presentation. The system connected her full-process data: pregnancy data (fetus in breech position, no other complications), labor data (cesarean section, successful surgery), and postpartum data (good wound healing, difficulty in breastfeeding, pelvic floor muscle function assessment showed mild damage).
[0074] The personalized follow-up plan generation module 320 generates the following follow-up plans: 1. Follow up once a day within the first week postpartum, focusing on wound healing and breastfeeding; 2. Follow up once every two weeks from the second to the sixth week postpartum, focusing on the progress of pelvic floor muscle rehabilitation; 3. Follow up once a month from the seventh to the twelfth month postpartum, monitoring long-term rehabilitation indicators.
[0075] The intelligent follow-up execution module 330 sent follow-up reminders on time. On the 3rd day postpartum, Ms. Zhang reported mild pain in her wound through an online questionnaire. The analysis model judged this as a normal postoperative reaction and did not trigger an abnormal warning. On the 10th day postpartum, Ms. Zhang reported low mood and insomnia. The analysis model detected postpartum depression tendencies, automatically marked it as abnormal, and reminded medical staff to intervene.
[0076] The personalized health management module 340 generates a guidance plan: ① Wound care guidance (text and video, focusing on cleaning and dressing of cesarean section wounds); ② Pelvic floor muscle rehabilitation guidance (Kegel exercise video tutorials, with training intensity customized according to the degree of pelvic floor muscle injury); ③ Breastfeeding guidance (including videos on correct breastfeeding postures and lactation-promoting recipes); ④ Emotional regulation guidance (relaxation training and psychological counseling suggestions). Based on Ms. Zhang's feedback, the plan is dynamically adjusted: postpartum depression psychological counseling content is added in the second week postpartum.
[0077] The postpartum care effectiveness assessment module 350 conducts a comprehensive evaluation at 6 weeks postpartum: good wound healing (score 90), improved pelvic floor muscle function (score 78), moderate breastfeeding effectiveness (score 68, below the preset assessment threshold of 75), and improved emotional state (score 80). Because the breastfeeding effectiveness score is below the preset assessment threshold of 75, the system automatically triggers the guidance plan adjustment process: increasing the frequency of lactation guidance, adjusting the dietary plan, increasing the follow-up frequency to once a week, and reminding medical staff to pay close attention to breastfeeding and provide targeted intervention suggestions.
[0078] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.
[0079] The computer device 40 includes a memory 401, a processor 402, and a network interface 403 that are interconnected via a system bus. It should be noted that only the computer device 40 with components 401-403 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.
[0080] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.
[0081] The memory 401 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 401 may be an internal storage unit of the computer device 40, such as the hard disk or memory of the computer device 40. In other embodiments, the memory 401 may also be an external storage device of the computer device 40, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device 40. Of course, the memory 401 may also include both the internal storage unit and the external storage device of the computer device 40. In this embodiment, the memory 401 is typically used to store the operating system and various application software installed on the computer device 40, such as computer-readable instructions for follow-up methods based on obstetric diagnosis and treatment information. Furthermore, the memory 401 can also be used to temporarily store various types of data that have been output or will be output.
[0082] In some embodiments, the processor 402 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 402 is typically used to control the overall operation of the computer device 40. In this embodiment, the processor 402 is used to execute computer-readable instructions stored in the memory 401 or to process data, for example, to execute computer-readable instructions for the follow-up method based on obstetric diagnostic information.
[0083] The network interface 403 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 40 and other electronic devices.
[0084] This application connects with multimodal data from the entire obstetric diagnosis and treatment process, uses a specialized obstetric language model to generate personalized follow-up plans, combines analytical models for abnormal early warning, and health conditioning models to generate personalized guidance plans. Through a multi-dimensional evaluation mechanism, it achieves dynamic closed-loop adjustment of follow-up and conditioning plans. Under the premise of ensuring maternal and infant safety, it can realize intelligent follow-up and rehabilitation guidance throughout the entire cycle of preconception, pregnancy, labor, and postpartum, improving the accuracy, timeliness, and effectiveness of follow-up and rehabilitation guidance.
[0085] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the follow-up method based on obstetric diagnosis and treatment information as described above.
[0086] This application connects with multimodal data from the entire obstetric diagnosis and treatment process, uses a specialized obstetric language model to generate personalized follow-up plans, combines analytical models for abnormal early warning, and health conditioning models to generate personalized guidance plans. Through a multi-dimensional evaluation mechanism, it achieves dynamic closed-loop adjustment of follow-up and conditioning plans. Under the premise of ensuring maternal and infant safety, it can realize intelligent follow-up and rehabilitation guidance throughout the entire cycle of preconception, pregnancy, labor, and postpartum, improving the accuracy, timeliness, and effectiveness of follow-up and rehabilitation guidance.
[0087] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0088] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.
Claims
1. A follow-up method based on obstetric diagnosis and treatment information, characterized in that, include: S1. Collect and integrate obstetric multimodal data; S2. Based on the obstetric multimodal data, a personalized full-cycle follow-up plan is generated using a pre-trained obstetric-specific large language model. S3. Execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data; S4. Based on the obstetric multimodal data and the follow-up data, generate a personalized health conditioning reference plan; S5. Conduct multi-dimensional evaluation of the health conditioning effect, and dynamically adjust the health conditioning reference plan and the full-cycle follow-up plan based on the evaluation results.
2. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, In step S1, the obstetric multimodal data includes preconception data, pregnancy data, labor data, and postpartum data. The collection and integration of obstetric multimodal data specifically includes: automatically collecting the obstetric multimodal data from the hospital through an interface; and structuring and extracting key health indicators from the collected obstetric multimodal data.
3. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, In step S2, the method for constructing the obstetric-specific large language model includes: Based on obstetric medical materials, a general large language model is trained for obstetric specialties. During the training process, the model’s ability to recognize obstetric specialties and extract information correlation is optimized to ensure that the model’s accuracy and recall rate for extracting obstetric information are greater than or equal to the preset values.
4. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, In step S2, a personalized full-cycle follow-up plan is generated, specifically including: Based on the aforementioned obstetric multimodal data, a personalized follow-up plan is generated using the pre-trained obstetric-specific large language model, which includes follow-up frequency, follow-up content, follow-up method, and follow-up reminder time. The personalized follow-up plan will be dynamically updated based on the pregnant woman's treatment progress, changes in health status, and feedback from medical staff.
5. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, Step S3 specifically includes: Before the follow-up time arrives, follow-up reminders containing the follow-up time, content, and method are sent to pregnant women and medical staff through multiple channels. For the first type of pre-set follow-up content, follow-up data is automatically collected through online questionnaires or smart devices; for the second type of pre-set follow-up content, offline follow-up appointments are automatically scheduled. The system records data during the follow-up process and uses an analytical model built on machine learning algorithms to analyze whether there are any abnormalities in the follow-up data. If an abnormality is found, it is automatically marked and medical staff are alerted to intervene.
6. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, Step S4 specifically includes: The obstetric rehabilitation guidance model, built on a large language model, is invoked, and combined with the obstetric multimodal data and follow-up data, to generate a personalized guidance plan that includes text, images, and videos. The system receives real-time feedback and follow-up data from pregnant women regarding their health management practices. If it detects that the treatment is not effective or that the woman's physical condition has changed, it automatically adjusts the health management reference plan.
7. The follow-up method based on obstetric diagnosis and treatment information according to claim 1, characterized in that, Step S5 specifically includes: For each stage of treatment, corresponding evaluation indicators are set. By collecting conditioning and follow-up data, and comparing them with evaluation indicators and standards, a preliminary evaluation result is generated. Medical staff then conduct a comprehensive evaluation of the health conditioning effect based on offline follow-up visits and feedback from pregnant women, and revise the preliminary evaluation result accordingly. The treatment effect level is determined based on the revised assessment results; if the treatment effect level is lower than the preset threshold, the guidance plan adjustment process is automatically triggered to remind medical staff to provide targeted intervention suggestions.
8. A follow-up device based on obstetric diagnosis and treatment information, characterized in that, include: The end-to-end data integration module is used to collect and integrate obstetric multimodal data; The personalized follow-up plan generation module is used to generate a personalized full-cycle follow-up plan based on the obstetric multimodal data and using a pre-trained obstetric-specific large language model. The intelligent follow-up execution module is used to execute the full-cycle follow-up plan, collect follow-up data, and perform anomaly analysis on the follow-up data; The personalized health management module is used to generate personalized health management reference plans based on the obstetric multimodal data and the follow-up data. The conditioning effect evaluation module is used to evaluate the health conditioning effect from multiple dimensions and dynamically adjust the health conditioning reference plan and the full-cycle follow-up plan based on the evaluation results.
9. A computer device, comprising a memory and a processor, characterized in that, The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, it implements the steps of the follow-up method based on obstetric diagnosis and treatment information as described in any one of claims 1-7.
10. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When the computer-readable instructions are executed by a processor, they implement the steps of the follow-up method based on obstetric diagnosis and treatment information as described in any one of claims 1-7.
