Patient intelligent inclusion method and system

Abstract

The invention relates to an intelligent patient incorporation method and system, and aims to realize accurate screening and personalized treatment scheme generation of patients through intelligent data processing and a machine learning algorithm. The system integrates in-hospital and out-hospital multi-dimensional data, and a patient portrait is generated through data cleaning and feature extraction. A doctor can customize and optimize screening rules through an interface, and the system intelligently screens patient portraits according to preset rules. And the patient meeting the condition is automatically included in the corresponding treatment scheme. The system performs detailed health assessment and classification on the patient by using a health score and risk classification model, and generates a personalized treatment scheme based on a classification result. According to the system, the screening efficiency is effectively improved, manual errors are reduced, unification and flexible adjustment of screening standards are achieved, and early intervention and precise medical support can be provided for patients, so that the overall diagnosis and treatment effect is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data integration, and in particular to a method and system for intelligent patient inclusion. Background Art

[0002] In modern healthcare management, with the rapid development of medical informatization and the popularization of the Internet of Things (IoT), medical institutions are able to collect a large amount of multi-dimensional data from patients. These data include in-hospital test results, electronic medical records, treatment records, etc., and out-of-hospital health data generated by patients in their daily lives through wearable devices and health applications. However, the traditional medical system has significant limitations in processing this data, which are mainly reflected in the following aspects:

[0003] 1. Low data utilization and serious information island phenomenon: In the existing medical system, data between different departments often exist independently, forming information islands. For example, physicians and surgeons use different electronic medical record systems, which makes it difficult to integrate data. In addition, it is difficult to integrate data within the hospital (such as test data and imaging data) with data outside the hospital (such as exercise monitoring data and daily diet records). In this way, although the hospital can collect a large amount of patient data, this data is not effectively used, lacks a unified description of the patient's overall picture, and it is difficult to fully assess the patient's health status.

[0004] 2. Traditional patient screening relies on manual labor and is inefficient: Currently, hospitals’ screening and inclusion of patients mainly rely on doctors’ professional experience and manual operations. Doctors need to manually review medical records and analyze test results to determine whether patients meet the screening criteria for a specific disease. This method is not only time-consuming and labor-intensive, but also prone to human errors and judgment bias. For some complex or atypical cases, doctors may miss the opportunity for early intervention, resulting in delays in the disease. In addition, manual screening is difficult to achieve large-scale patient assessment and classification, especially when facing a large patient population in a hospital, the efficiency and accuracy of screening are low.

[0005] 3. The existing screening rules are single and lack personalization: In the traditional medical model, screening rules are often single and fixed. For example, the screening criteria for hypertension are usually based on a simple threshold judgment of blood pressure values, and the screening criteria for diabetes are based on fasting blood glucose levels or glycosylated hemoglobin levels. This standardized screening method cannot take into account individual differences in patients, such as age, gender, living habits, genetic background and other factors. Therefore, standardized screening rules often lead to the omission of some patients, especially for high-risk patients with atypical symptoms or no obvious medical history. In addition, due to the lack of personalized screening criteria, it is difficult for doctors to flexibly adjust the criteria during the screening process to meet the needs of specific patients.

[0006] 4. Increasing demand for automated and intelligent screening systems: In order to cope with the limitations of traditional patient screening methods, intelligent and automated screening systems are gradually being introduced into the medical field. These systems rely on big data analysis, machine learning, and artificial intelligence technologies. Through automated data collection and processing, they can quickly screen a large number of patient data and identify potential high-risk patients. However, existing intelligent screening systems also have problems. On the one hand, many systems lack sufficient data accumulation and it is difficult to form a reliable screening model; on the other hand, some systems rely on rules that are too fixed and difficult to adjust according to the latest medical research or the actual needs of doctors. This makes it difficult for current intelligent screening systems to be effectively popularized and promoted in practical applications.

[0007] 5. Lack of uniformity in screening standards makes it difficult to achieve accurate diagnosis: Existing medical screening standards are mostly formulated by each hospital or department independently, and lack uniform specifications. This leads to differences in screening standards between different hospitals and different doctors, making it difficult to achieve accurate diagnosis. For example, for screening of cardiovascular disease, different hospitals may have different blood pressure thresholds or risk assessment methods. This lack of uniformity in standards will affect the accuracy and comparability of screening, making it difficult for patients' screening results to be widely recognized. In addition, the existing standards are not flexible enough to be dynamically adjusted based on the latest clinical research and individualized patient data, resulting in some potential high-risk patients being overlooked.

[0008] 6. Unbalanced distribution of medical resources increases the burden on doctors: In most medical institutions, professional doctors and medical resources are relatively limited, while the number of patients is large and their health conditions are complex. Especially in some grassroots hospitals or areas with insufficient medical resources, doctors need to process a large amount of patient data and perform complex screening and diagnosis work. Traditional manual screening methods not only increase the workload of doctors, but may also lead to a decline in screening quality. Faced with limited time and energy, doctors may not be able to conduct a comprehensive assessment of each patient, which affects the diagnosis and treatment effect.

[0009] 7. The traditional medical model is difficult to achieve early disease intervention and precision treatment: With the development of modern medicine, early intervention and precision treatment have gradually become important means to improve patient prognosis and reduce medical expenses. However, due to the low efficiency and lack of accuracy of traditional patient screening methods, it is difficult to achieve early identification of high-risk patients. This has led to many patients failing to receive timely intervention in the early stages of the disease, thus missing the best time for treatment. The traditional medical model is also difficult to formulate personalized treatment plans based on the patient's personalized health data, which to some extent limits the development and popularization of precision medicine.

[0010] To this end, we continue to design a patient intelligent inclusion method and system to solve the above problems. Summary of the invention

[0011] The purpose of the present invention is to provide a patient intelligent inclusion method and system to address the deficiencies of the prior art and to solve the problems raised in the background technology.

[0012] To achieve the above object, the present invention provides the following technical solution: a patient intelligent inclusion method, comprising the following steps:

[0013] Collect multi-dimensional data of patients, including in-hospital data in , out-of-hospital data out and related health imaging dataD image ;

[0014] The collected data are cleaned and normalized to form a structured data set; the feature extraction algorithm is used to generate the patient portrait P i , the formula is:

[0015] P i =f(D in ,D out ,D image )

[0016] Among them, P i represents the portrait of the i-th patient.

[0017] As a preferred technical solution of the present invention, the feature extraction algorithm includes principal component analysis and linear discriminant analysis, which are used to reduce the dimension and optimize the features of multidimensional data. The formula is:

[0018] P i = PCA(D in )+LDA(D out )

[0019] As a preferred technical solution of the present invention, through the patient portrait P i and custom screening rules R j The patients are screened according to the screening rule R j Contains multiple condition combinations, the formula is expressed as:

[0020] R j =C1∧C2∨C3

[0021] Among them, C n For screening conditions.

[0022] As a preferred technical solution of the present invention, the screening rule R j Doctors can make fine-tuning through the rule configuration interface to adapt to the specific needs of different specialist departments.

[0023] As the preferred technical solution of the present invention, the screening rule R jPatients will be included in a specific treatment plan T k , the formula is:

[0024]

[0025] Among them, T k represents the kth treatment plan, δ is the matching function, S i Score the patient's health.

[0026] As a preferred technical solution of the present invention, the health score S i It is calculated by weighted multi-index evaluation model, the formula is:

[0027]

[0028] Among them, w m is the weight of each health indicator, x m is the patient's performance value on this indicator.

[0029] The present invention also provides a patient intelligent inclusion system, comprising:

[0030] Data collection module: used to collect in-hospital and out-of-hospital data of patients;

[0031] Data preprocessing module: used to clean and standardize the collected data;

[0032] Patient portrait generation module: used to generate multi-dimensional portraits of patients;

[0033] Screening rule configuration module: used to set and adjust patient screening rules;

[0034] Intelligent inclusion module: used to screen patients according to screening rules and include eligible patients in the treatment plan.

[0035] As a preferred technical solution of the present invention, the patient portrait generation module uses a support vector machine and a random forest algorithm to perform nonlinear mapping and feature selection on the data to improve the accuracy of patient portrait generation.

[0036] As a preferred technical solution of the present invention, the intelligent inclusion module includes a deep learning model, which can adjust the screening rules in real time according to the dynamic changes of patient characteristics, thereby optimizing the accuracy of screening and treatment plans.

[0037] As a preferred technical solution of the present invention, the system includes a patient classification module, which divides patients into different risk categories through a clustering algorithm, and the formula is:

[0038]

[0039] Among them, Ci is the classification category of the i-th patient, μ k is the center point of the kth class, ||P j -μ k || represents Euclidean distance.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention greatly improves the efficiency of patient screening through intelligent and automated screening methods. The system can automatically collect and process multi-dimensional data of patients, and quickly screen potential high-risk patients through machine learning algorithms, reducing the time and energy invested by doctors in manual analysis. Doctors only need to perform secondary confirmation and fine-tuning on the screened results, which greatly reduces the workload.

[0042] The system allows doctors to customize and fine-tune screening rules according to specific department needs and individual patient differences, and achieve personalized screening standards. At the same time, the system can automatically generate personalized treatment plans based on the patient's health profile and risk classification, supporting precision medicine. This can effectively reduce missed diagnoses and misdiagnoses, especially provide early intervention for patients with atypical symptoms, and improve the overall diagnosis and treatment effect.

[0043] The system can unify and optimize screening standards through standardized data processing and flexible rule configuration. The intelligent screening model can continuously adjust and optimize based on new data and feedback to ensure the scientificity and accuracy of screening standards. This not only improves the credibility of screening results, but also facilitates the sharing and comparison of screening standards between different hospitals and departments, helping to build a more accurate and consistent diagnosis and treatment system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A method flow chart of a patient intelligent inclusion method and system proposed by the present invention;

[0045] Figure 2 This is a system block diagram of a patient intelligent inclusion method and system proposed by the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] The following is combined with Figure 1 and Figure 2And multiple embodiments are described in detail for specific implementations of the present invention.

[0048] Example 1: Data collection and patient portrait generation

[0049] In this embodiment, it is described in detail how to collect and process patient data to generate an accurate patient portrait.

[0050] Data collection: Integrate the hospital information system (HIS) with the laboratory information system (LIS), electronic medical record system (EMR), and picture acquisition and imaging data management system (PACS) to collect the patient's in-hospital data. in , including test data (such as blood test results), diagnostic information (such as medical history), and medication use (such as frequency and dosage).

[0051] Collect out-of-hospital data through mobile medical devices (such as wearable devices) and patient self-reported data collection systems out , including daily exercise steps, diet records, sleep quality and other lifestyle information.

[0052] Image data D image The imaging system from the hospital covers the results of CT, MRI, X-ray, etc. Data preprocessing: The system uses an outlier detection algorithm to remove abnormal points in the data, such as using the interquartile range (IQR) method to detect abnormal values ​​in the test indicators.

[0053] Normalize the numerical data and map it to the interval [0,1] for easy processing by machine learning algorithms.

[0054] The normalization formula is:

[0055]

[0056] One-Hot Encoding is performed on categorical variables (such as medical history and medication usage) to ensure data availability and consistency.

[0057] Patient portrait generation:

[0058] Principal component analysis (PCA) is used to reduce the dimensionality of high-dimensional data and extract key features.

[0059] The PCA algorithm selects the top k most important features based on the variance. The formula is:

[0060] X pca =X·W

[0061] Among them, X is the original data matrix and W is the eigenvector matrix.

[0062] Linear discriminant analysis (LDA) is used to optimize features and improve inter-class differences. The formula is:

[0063]

[0064] Among them, S b is the class scatter matrix, S w is the within-class scatter matrix.

[0065] Generated patient portrait P i It is saved in the database and displayed to doctors through a data visualization interface to quickly understand the patient's overall health status.

[0066] Example 2: Configuration and optimization of patient screening rules

[0067] This embodiment provides a detailed description of how to configure and optimize screening rules according to specific needs.

[0068] Configuration of screening rules: Doctors can select preset screening templates through the system's rule configuration interface. Taking diabetes screening as an example, doctors can select the following standard screening rules:

[0069] R diabetes =(FPG>7.0)∧(BMI>25)∨(HbA1c>6.5)

[0070] Among them, FPG is fasting blood sugar, BMI is body mass index, and HbA1c is glycated hemoglobin level. Doctors can fine-tune the thresholds in the rules according to specific circumstances. For example, for high-risk patient groups, the screening standard for BMII can be adjusted to 20.

[0071] Rule optimization and automatic adjustment: The system has a built-in model based on LSTM (Long Short-Term Memory) neural network to analyze historical screening data. Through this model, the system can automatically adjust the threshold of the rule. For example, when the system detects that the sensitivity of certain screening rules is insufficient, it will automatically adjust the threshold to reduce the missed diagnosis rate. The adjusted rules will generate recommendations for doctors to manually confirm and fine-tune.

[0072] Example 3: Intelligent screening of patients and generation of treatment plans

[0073] In this embodiment, the system describes in detail how to use screening rules to screen patients and generate personalized treatment plans.

[0074] Intelligent screening process: The system uses the configured screening rules R j , for patient portrait P i The matching algorithm is a combination of multiple conditions based on Boolean logic. For example, the rules for cardiovascular disease screening are:

[0075] R cardio =(SBP>140)∧(DBP>90)∨(Age>60andSmoker=True)

[0076] Among them, SBP is systolic blood pressure and DBP is diastolic blood pressure.

[0077] The system calls the matching function δ to determine whether the patient meets the screening criteria. The matching formula is:

[0078]

[0079] Generation of treatment plans: For patients who meet the screening criteria, the system generates personalized treatment plans.

[0080] For example, for patients with cardiovascular disease, the system generates the following multi-layered treatment plan:

[0081] T cardio ={drug treatment (antihypertensive drugs), exercise intervention (30 minutes of brisk walking every day), diet control (low-salt diet)}

[0082] The system automatically adjusts the specific content of each intervention based on the patient's age, gender, weight and other parameters. For example, for patients over 65 years old, the system may recommend a more gentle exercise program.

[0083] Example 4: Health score calculation and patient risk classification

[0084] This embodiment demonstrates how to calculate a patient's health score and classify the patient's risk based on the score.

[0085] Detailed calculation of health score: The system uses a multi-index weighted model to calculate the health score S of each patient. i The weight of each indicator is obtained by doctors or through machine learning model training. The formula is:

[0086]

[0087] Among them, w m is the indicator weight, x m The score of the patient on this indicator. For example, fasting blood sugar accounts for 0.5, blood pressure accounts for 0.3, and body mass index accounts for 0.2.

[0088] Patient risk classification: The system uses the K-means clustering algorithm to classify patients into different risk categories. The algorithm calculates the patient profile P i With various center points μ k The distance to determine the category:

[0089]

[0090] Among them, C i The risk category of the patient.

[0091] The system divides patients into low-risk, medium-risk and high-risk categories and provides corresponding treatment recommendations.

[0092] Example 5: Automated Adjustment and Feedback Mechanism

[0093] In this embodiment, it is described in detail how to continuously optimize the screening and treatment plans through the system's feedback mechanism and deep learning model.

[0094] Automatic adjustment of screening rules:

[0095] The system automatically adjusts the threshold of the screening rule based on the accuracy of the screening results. For example, if the system detects that a certain condition (such as BMI threshold) has a high misdiagnosis rate in a certain age group, it will automatically lower the weight or threshold of the condition.

[0096] This automated adjustment process is performed through deep learning models (such as CNN) to enable real-time optimization of screening rules.

[0097] Doctor feedback mechanism: The system records the results of each screening and treatment, and allows doctors to provide feedback on each case, including whether the screening criteria are met and how effective the treatment is.

[0098] The system continuously adjusts and optimizes screening rules based on doctors’ feedback. The feedback data will also be used to train deep learning models to better predict future patient health risks.

[0099] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in the field. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A patient intelligent inclusion method, characterized in that: The following steps are involved: Collect multi-dimensional data of patients, including in-hospital data in , out-of-hospital data out and related health imaging dataD image ; Clean and normalize the collected data to form a structured data set; Apply feature extraction algorithm to generate patient portrait P i , the formula is: P i =f(D in ,D out ,D image ) Among them, P i represents the portrait of the i-th patient.

2. A patient intelligent inclusion method according to claim 1, characterized in that: The feature extraction algorithm includes principal component analysis and linear discriminant analysis, which are used to reduce the dimension and optimize the features of multidimensional data. The formula is: P i =PCA(D in )+LDA(D out )。 3. A patient intelligent inclusion method according to claim 1, characterized in that: Through patient portrait i and custom screening rules R j The patients are screened according to the screening rule R j Contains multiple condition combinations, the formula is expressed as: R j =C1∧C2VC3 Among them, C n For screening conditions.

4. A patient intelligent inclusion method according to claim 3, characterized in that: The screening rule R j Doctors can make fine-tuning through the rule configuration interface to adapt to the specific needs of different specialist departments.

5. A patient intelligent inclusion method according to claim 3, characterized in that: Meet the screening rule R j Patients will be included in a specific treatment plan T k , the formula is: Among them, T k represents the kth treatment plan, δ is the matching function, S i Score the patient's health.

6. A patient intelligent inclusion method according to claim 5, characterized in that: The health score S i It is calculated by weighted multi-index evaluation model, the formula is: Among them, w m is the weight of each health indicator, x m is the patient's performance value on this indicator.

7. A patient intelligent inclusion system, characterized in that: include: Data collection module: used to collect in-hospital and out-of-hospital data of patients; Data preprocessing module: used to clean and standardize the collected data; Patient portrait generation module: used to generate multi-dimensional portraits of patients; Screening rule configuration module: used to set and adjust patient screening rules; Intelligent inclusion module: used to screen patients according to screening rules and include eligible patients in the treatment plan.

8. A patient intelligent inclusion system according to claim 7, characterized in that: The patient portrait generation module uses support vector machines and random forest algorithms to perform nonlinear mapping and feature selection on data to improve the accuracy of patient portrait generation.

9. A patient intelligent inclusion system according to claim 7, characterized in that: The intelligent inclusion module includes a deep learning model that can adjust the screening rules in real time according to the dynamic changes in patient characteristics, thereby optimizing the accuracy of screening and treatment plans.

10. A patient intelligent inclusion system according to claim 7, characterized in that: The system includes a patient classification module, which divides patients into different risk categories through a clustering algorithm, and the formula is: Among them, C i is the classification category of the i-th patient, μ k is the center point of the kth class, ||P j -μ k || represents the Euclidean distance.

Images