A disease prediction method and system based on disease history and personal habit data
Patent Information
- Application Number
- CN202611152545.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-31
- Publication Date
- 2026-10-16
AI Technical Summary
[0003]目前，对于患者的疾病风险预测，均是将患者的行为习惯直接输入风险预测模型中进行评估，但是忽略了在实际情况中，疾病均是因为日积月累而形成的，并且疾病和疾病之间又有较强的关联性，单一的行为习惯评估得出的疾病风险并不准确，实用性较低
本发明基于疾病史与个人习惯数据的疾病预测方法，在对神经网络模型训练的样本数据中，每个患者的特征事件沿生命时间轴按顺序分布，并且在医学层面上，将部分发生时间在后的疾病确诊事件和发生时间在前的特征事件中的一种或者多种组合定义因果关联，在形成疾病风险预测模型后，将用户的输入数据输入疾病风险预测模型，疾病风险预测模型对该用户历史数据分析，并且根据待预测年龄和待预测疾病处理出预测结果，预估该用户年龄下待预测疾病的患病风险，本申请的疾病风险预测模型考虑待预测年龄在生命时间轴中的位置，以及待预测疾病与历史数据中的特征事件的因果关联，优化疾病评估的可靠性以及实用性，为用户的健康风险评估提供辅助依据。
Smart Images
Figure CN122930508A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of disease risk prediction technology, and in particular to a disease prediction method and system based on medical history and personal habit data. Background Technology
[0002] With the rapid development of smart healthcare, disease risk prediction technology based on user health data has gradually become a core means of early disease screening and health risk management, which is of great significance for reducing the incidence of major diseases and improving the health of the whole population.
[0003] Currently, disease risk prediction for patients is based on directly inputting patients' behavioral habits into risk prediction models for assessment. However, this ignores the fact that in reality, diseases are formed over time and there is a strong correlation between diseases. Disease risk derived from a single behavioral habit assessment is inaccurate and has low practicality. Summary of the Invention
[0004] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a disease prediction method and system based on medical history and personal habit data, optimizing the reliability and practicality of disease assessment and providing auxiliary evidence for users' health risk assessment.
[0005] A disease prediction method based on medical history and personal habit data according to a first aspect of the present invention includes: acquiring multiple sets of sample data, wherein the sample data includes multiple feature events distributed along a life timeline, and the multiple feature events include one or more combinations of birth feature events, personal habit events, and disease diagnosis events, wherein some disease diagnosis events that occurred later are causally associated with one or more combinations of feature events that occurred earlier; training a neural network model with the multiple sets of sample data to construct a disease risk prediction model; acquiring input data to be predicted, wherein the input data includes historical data, the age to be predicted, and the disease to be predicted, and the historical data includes multiple feature events distributed along a life timeline; and inputting the input data into the disease risk prediction model to process and obtain a prediction result, wherein the prediction result includes at least the risk of the disease to be predicted at the user's age.
[0006] A disease prediction method based on medical history and personal habit data according to an embodiment of the present invention has at least the following beneficial effects: This invention relates to a disease prediction method based on medical history and personal habit data. In the sample data used to train the neural network model, each patient's characteristic events are sequentially distributed along the life timeline. At the medical level, causal relationships are defined between one or more combinations of later-occurring disease diagnosis events and earlier-occurring characteristic events. After forming a disease risk prediction model, the user's input data is fed into the model. The model analyzes the user's historical data and, based on the age and disease to be predicted, generates a prediction result, estimating the risk of the predicted disease at the user's age. This disease risk prediction model considers the position of the predicted age on the life timeline and the causal relationships between the predicted disease and characteristic events in historical data, optimizing the reliability and practicality of disease assessment and providing auxiliary basis for users' health risk assessment.
[0007] According to some embodiments of the present invention, the sample data further includes multiple featureless filling events, each with a time anchor point. The featureless filling events are used to characterize the occurrence of featureless events at the corresponding time anchor points, and the featureless filling events are arranged at corresponding positions on the life timeline according to the corresponding time anchor points.
[0008] According to some embodiments of the present invention, a plurality of the featureless filling events are arranged sequentially in the lifetime timeline at sampling time intervals.
[0009] According to some embodiments of the present invention, in the disease risk prediction model, each disease diagnosis event has a causal association set and a risk benchmark set. The causal association set has a one-to-one causal weight value representing the degree of association with other characteristic events, and the risk benchmark set has a benchmark risk value associated with other characteristic events. In the disease risk prediction model, the causal weight value and benchmark risk value of the characteristic events that occurred earlier are obtained based on the age to be predicted, historical data, and the disease to be predicted. The prediction result is calculated at least based on the corresponding causal weight value and benchmark risk value.
[0010] According to some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and produce prediction results includes: setting prediction time nodes in the life timeline according to the age to be predicted; and setting the causal weight values of feature events in historical data whose occurrence time is equal to or greater than the prediction time node to zero to mask the corresponding feature events.
[0011] According to some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results further includes: the time interval between the corresponding prediction time node and the feature event that occurred earlier in the statistical historical data; the prediction result is calculated at least based on the corresponding causal weight value, the baseline risk value, and the time interval value.
[0012] According to some embodiments of the present invention, inputting input data into a disease risk prediction model to process and obtain prediction results includes: obtaining the current time node; inputting the current time node and various disease diagnosis events in a disease type set into the disease risk prediction model to process and obtain the disease risk of various diseases, wherein the disease type set includes multiple disease diagnosis events; selecting the disease diagnosis event corresponding to the highest disease risk as the most likely disease diagnosis event to occur in the future to constitute at least part of the prediction results.
[0013] According to some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and generate prediction results includes: obtaining a disease diagnosis event to be predicted; inputting the disease diagnosis event to be predicted and various time nodes in the life timeline into the disease risk prediction model to process and generate the disease risk of the disease diagnosis event to be predicted at different time nodes; when the disease risk corresponding to any time node is greater than a risk threshold, outputting warning information to constitute at least a partial prediction result, wherein the warning information indicates that the disease diagnosis event to be predicted may occur at that time node.
[0014] According to some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and generate prediction results includes: inputting various disease-diagnosed events from a disease category set and various time nodes from a life timeline into the disease risk prediction model to process the disease risk of various disease-diagnosed events at different time nodes, wherein the disease category set includes multiple disease-diagnosed events; when the disease risk of any disease-diagnosed event at any time node is greater than a risk threshold, the time node is marked as the onset time node of the disease-diagnosed event in the life timeline, and health trajectory data is formed to constitute at least a partial prediction result.
[0015] According to a second aspect of the present invention, a disease prediction system is used to perform the disease prediction method based on medical history and personal habit data disclosed in any of the above embodiments.
[0016] The disease prediction system according to embodiments of the present invention has at least the following beneficial effects: The disease prediction system of this invention applies the disease prediction method based on medical history and personal habit data disclosed in any of the above embodiments, optimizes the reliability and practicality of disease assessment, and provides auxiliary basis for users' health risk assessment.
[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0018] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of one embodiment of the disease prediction method based on medical history and personal habit data of the present invention. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0020] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.
[0021] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than", "less than", "exceeding" are understood to exclude the number itself, and "above", "below", "within" are understood to include the number itself.
[0022] 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 belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0023] like Figure 1 As shown, a disease prediction method based on medical history and personal habit data according to a first aspect embodiment of the present invention includes: S110. Obtain multiple sets of sample data, wherein the sample data includes multiple characteristic events distributed along the life timeline, and the multiple characteristic events include one or more combinations of birth characteristic events, personal habit events, and disease diagnosis events, wherein some disease diagnosis events that occur later are causally associated with one or more combinations of characteristic events that occur earlier. S120. Train the neural network model with multiple sets of sample data to construct a disease risk prediction model; S130. Obtain the input data to be predicted, wherein the input data includes historical data, the age to be predicted, and the disease to be predicted, and the historical data includes multiple feature events distributed along the life timeline. S140. Input the input data into the disease risk prediction model to process and produce prediction results, wherein the prediction results include at least the risk of the disease to be predicted for the user's age.
[0024] Taking patient A's sample data as an example, multiple characteristic events distributed along the life timeline can be set in the form of [A, 0 days, male]. Here, 0 days is the marker on the life timeline, and male is the birth characteristic event. For example, in [A, 9125 days, current smoking], current smoking is a personal habit event, and [A, 9125 days, current smoking] indicates that patient A started smoking at 9125 days after birth. In [A, 16071 days, hypertension], hypertension is a disease diagnosis event, and [A, 16071 days, hypertension] indicates that patient A was diagnosed with hypertension at 16071 days after birth.
[0025] After constructing the disease risk prediction model, the user's input data contains historical data, which includes multiple characteristic events distributed along the life timeline. These characteristic events include birth characteristic events, personal habit events, and disease diagnosis events. By utilizing the existing personal habit events and disease diagnosis events in the user's historical data, the risk of the predicted disease at the user's age can be analyzed.
[0026] This invention relates to a disease prediction method based on medical history and personal habit data. In the sample data used to train the neural network model, each patient's characteristic events are sequentially distributed along the life timeline. At the medical level, causal relationships are defined between one or more combinations of later-occurring disease diagnosis events and earlier-occurring characteristic events. After forming a disease risk prediction model, the user's input data is fed into the model. The model analyzes the user's historical data and, based on the age and disease to be predicted, generates a prediction result, estimating the risk of the predicted disease at the user's age. This disease risk prediction model considers the position of the predicted age on the life timeline and the causal relationships between the predicted disease and characteristic events in historical data, optimizing the reliability and practicality of disease assessment and providing auxiliary basis for users' health risk assessment.
[0027] In some embodiments of the present invention, the sample data further includes multiple featureless filling events. Each featureless filling event is set with a time anchor point. The featureless filling event is used to characterize the occurrence of a featureless event at the corresponding time anchor point. The featureless filling events are arranged at the corresponding positions in the life timeline according to the corresponding time anchor points.
[0028] Specifically, the no-feature-filled event can be set in the form of [A, 1825 days, no event], with 1825 days as the time anchor point, used to represent that no feature event occurred at that time point.
[0029] Understandably, for younger populations or those with low disease burden, the sample data has fewer characteristic events, making it difficult for disease risk prediction models to accurately capture the risk characteristics of the sample data throughout its entire life cycle. Therefore, multiple featureless events are inserted into the life timeline and interleaved with the characteristic events of the original sample data to ensure that the disease risk prediction model can capture the dynamic changes in risk throughout the entire life cycle.
[0030] In some embodiments of the present invention, a plurality of the featureless filling events are arranged sequentially in the life time axis at sampling time intervals. For example, featureless filling events are added to the life event axis at 5-year time intervals between each pair, by uniformly sampling 20 time points in the range of (0, 36525 days) to place featureless filling events.
[0031] In some embodiments of the present invention, each disease diagnosis event in the disease risk prediction model has a causal association set and a risk benchmark set. The causal association set has a causal weight value that represents the degree of association with other feature events, and the risk benchmark set has a benchmark risk value that is associated with other feature events. In the disease risk prediction model, the causal weight values of the feature events that occurred earlier and the baseline risk values are obtained based on the age to be predicted, historical data, and the disease to be predicted. The prediction results are calculated based at least on the corresponding causal weight values and benchmark risk values.
[0032] Understandably, the risk of developing the disease can be calculated by multiplying and summing the causal weight values of each associated feature event and the baseline risk value, based on the user's historical data and the disease to be predicted and the age to be predicted.
[0033] In some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results includes: Set the prediction time node in the life timeline based on the age to be predicted; Set the causal weight of feature events in historical data whose occurrence time is equal to or greater than the prediction time node to zero to mask the corresponding feature events.
[0034] It is understandable that when using the prediction time point to obtain feature events from historical data, the disease risk prediction model may traverse the causal weight values of all feature events, leading to misjudgment. Therefore, an attention mask value can be inserted for each feature event. When the occurrence time of a feature event is equal to or greater than the prediction time point, the attention mask value is set to zero, so that the causal weight value is set to zero to mask the corresponding feature event. For feature events that occur less than the prediction time point, the attention mask value is set to one or ∞, so that the causal weight value can be used for risk calculation.
[0035] In some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results further includes: The time interval between the predicted time point and the characteristic event that occurred earlier in the statistical historical data; The prediction result is calculated based on at least the corresponding causal weight value, the baseline risk value, and the time interval value.
[0036] Understandably, considering that diseases are formed due to the accumulation of behavioral habits or other illnesses over time, when estimating the risk of disease, we first statistically analyze the time interval between the corresponding prediction time point and the characteristic event that occurred earlier in the historical data. We then use the time interval to calculate the risk gain value corresponding to each characteristic time. It is understood that the larger the time interval, the larger the risk gain value. By multiplying and summing the causal weight values of each related characteristic event, the baseline risk value, and the risk gain value, we can calculate the risk of disease, thereby improving the accuracy and reliability of disease risk prediction.
[0037] In some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results includes: Get the current time point; The disease risk prediction model is input based on the current time point and various disease diagnosis events in the disease category set to process the risk of various diseases. The disease category set includes multiple disease diagnosis events. The disease diagnosis event corresponding to the highest risk of developing the disease is selected as the most likely disease diagnosis event to constitute at least part of the prediction results.
[0038] The disease category set can include 1256 disease or death diagnosis events. Based on the user's historical data, the risk of contracting various disease diagnoses at different time points can be analyzed. Then, the disease diagnosis event with the highest risk of contracting the disease after the current time point is taken as the most likely disease diagnosis event to occur in the future and output as a prediction result for the user's reference and prevention.
[0039] In some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results includes: Obtain the estimated disease diagnosis events; The disease risk prediction model is processed by inputting the disease diagnosis event to be predicted and the various time points in the life timeline into the model to calculate the disease risk of the disease diagnosis event at different time points. When the risk of disease at any given time point is greater than the risk threshold, an alert message is output to form at least a partial prediction result. The alert message indicates that a disease diagnosis event to be predicted may occur at that time point.
[0040] Understandably, based on current historical data, users can also input the disease to be predicted. The disease risk prediction model analyzes the changes in the risk of the disease at different time points, and outputs a warning message when the risk of the disease is greater than the risk threshold. The risk threshold can be set by doctors or designers according to the actual situation, such as 70%, 80%, etc.
[0041] In some embodiments of the present invention, the process of inputting input data into a disease risk prediction model to process and obtain prediction results includes: The disease risk prediction model is input into the disease type set, which includes various disease diagnosis events and time nodes in the life timeline, in order to process the disease risk of various disease diagnosis events at different time nodes. The disease type set includes multiple disease diagnosis events. If the risk of contracting any disease at any given time point is greater than the risk threshold, then that time point is marked as the onset time point of the disease diagnosis event on the life timeline, and health trajectory data is formed to constitute at least part of the prediction results.
[0042] Understandably, based on current historical data, disease risk prediction models can analyze the changes in the risk of various disease diagnoses at different time points within a set of disease types. When the risk of any disease diagnosis at any time point exceeds a risk threshold, that time point is marked as the onset time point of the disease diagnosis on the life timeline, forming health trajectory data. In the health trajectory data, there can be multiple onset time points, and each onset time point presents a corresponding disease diagnosis event and disease risk.
[0043] According to a second aspect of the present invention, a disease prediction system is used to perform the disease prediction method based on medical history and personal habit data disclosed in any of the above embodiments.
[0044] The disease prediction system of this invention applies the disease prediction method based on medical history and personal habit data disclosed in any of the above embodiments, optimizes the reliability and practicality of disease assessment, and provides auxiliary basis for users' health risk assessment.
[0045] A computer-readable storage medium stores a computer program that, when executed by a processor, implements the chronic disease assessment method disclosed in any of the above embodiments.
[0046] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0047] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.
[0048] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.
[0049] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0050] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.
[0051] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0052] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.
[0053] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0054] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
Claims
1. A disease prediction method based on medical history and personal habit data, characterized in that, include: Acquire multiple sets of sample data, which include multiple characteristic events distributed along the life timeline. The multiple characteristic events include one or more combinations of birth characteristic events, personal habit events, and disease diagnosis events. Among them, some disease diagnosis events that occurred later are causally associated with one or more combinations of characteristic events that occurred earlier. Multiple sets of sample data are used to train a neural network model to construct a disease risk prediction model; Obtain the input data to be predicted, which includes historical data, the age to be predicted, and the disease to be predicted. The historical data includes multiple feature events distributed along the life timeline. Input data is fed into a disease risk prediction model to generate prediction results, which include at least the risk of the disease to be predicted for the user's age.
2. The disease prediction method based on medical history and personal habit data according to claim 1, characterized in that, The sample data also includes multiple featureless filling events. Each featureless filling event is set with a time anchor point. The featureless filling event is used to characterize the occurrence of a featureless event at the corresponding time anchor point. The featureless filling events are arranged at the corresponding positions in the life timeline according to the corresponding time anchor points.
3. The disease prediction method based on medical history and personal habit data according to claim 2, characterized in that, Multiple featureless filling events are arranged sequentially on the lifetime timeline at sampling time intervals.
4. The disease prediction method based on medical history and personal habit data according to claim 2, characterized in that, In the disease risk prediction model, each disease diagnosis event has a causal association set and a risk benchmark set. The causal association set has a one-to-one causal weight value that represents the degree of association with other characteristic events, and the risk benchmark set has a benchmark risk value that is associated with other characteristic events. In the disease risk prediction model, the causal weight values of the feature events that occurred earlier and the baseline risk values are obtained based on the age to be predicted, historical data, and the disease to be predicted. The prediction results are calculated based at least on the corresponding causal weight values and benchmark risk values.
5. The disease prediction method based on medical history and personal habit data according to claim 4, characterized in that, The process of inputting data into a disease risk prediction model to generate prediction results includes: Set the prediction time node in the life timeline based on the age to be predicted; Set the causal weight of feature events in historical data whose occurrence time is equal to or greater than the prediction time node to zero to mask the corresponding feature events.
6. The disease prediction method based on medical history and personal habit data according to claim 5, characterized in that, The process of inputting data into a disease risk prediction model to produce prediction results also includes: The time interval between the predicted time point and the characteristic event that occurred earlier in the statistical historical data; The prediction result is calculated based on at least the corresponding causal weight value, the baseline risk value, and the time interval value.
7. The disease prediction method based on medical history and personal habit data according to claim 4, characterized in that, The process of inputting data into a disease risk prediction model to generate prediction results includes: Get the current time point; The disease risk prediction model is input based on the current time point and various disease diagnosis events in the disease category set to process the risk of various diseases. The disease category set includes multiple disease diagnosis events. The disease diagnosis event corresponding to the highest risk of developing the disease is selected as the most likely disease diagnosis event to constitute at least part of the prediction results.
8. The disease prediction method based on medical history and personal habit data according to claim 4, characterized in that, The process of inputting data into a disease risk prediction model to generate prediction results includes: Obtain the estimated disease diagnosis events; The disease risk prediction model is processed by inputting the disease diagnosis event to be predicted and the various time points in the life timeline into the model to calculate the disease risk of the disease diagnosis event at different time points. When the risk of disease at any given time point is greater than the risk threshold, an alert message is output to form at least a partial prediction result. The alert message indicates that a disease diagnosis event to be predicted may occur at that time point.
9. A disease prediction method based on medical history and personal habit data according to claim 4, characterized in that, The process of inputting data into a disease risk prediction model to generate prediction results includes: The disease risk prediction model is input into the disease type set, which includes various disease diagnosis events and time nodes in the life timeline, in order to process the disease risk of various disease diagnosis events at different time nodes. The disease type set includes multiple disease diagnosis events. If the risk of contracting any disease at any given time point is greater than the risk threshold, then that time point is marked as the onset time point of the disease diagnosis event on the life timeline, and health trajectory data is formed to constitute at least part of the prediction results.
10. A disease prediction system, characterized in that, Used to perform the disease prediction method based on medical history and personal habit data as described in any one of claims 1 to 8.