Liver cancer early diagnosis risk prediction model construction method

Abstract

The invention discloses a construction method of a risk prediction model for early diagnosis of liver cancer. The construction method of the risk prediction model comprises the following steps: 1, preparing clinical data; 2, processing, analyzing and learning clinical data, and constructing a project database of a hepatocellular carcinoma clinical diagnosis path based on big data; 3, determining main diagnosis key points in the disease timing sequence diagnosis scheme and related indexes influencing the operation, and formulating a diagnosis and treatment path extraction rule; 4, constructing a quality-efficiency evaluation index system for the diagnosis and treatment path set, and calculating the early diagnosis probability and postoperative recurrence probability of the liver cancer; and 5, establishing a Markov model for early screening and diagnosis of liver cancer and risk prediction of postoperative recurrence. According to the method, the ANN principle is utilized, an HCC early diagnosis risk prediction model and an HCC clinical diagnosis and treatment path evaluation system are established, HCC high-risk groups are subjected to early recognition, and the HCC early diagnosis rate is increased; meanwhile, an optimal clinical diagnosis and treatment decision is provided for HCC treatment.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and more particularly to the field of medical informatics technology, and specifically to a method for constructing a risk prediction model for early diagnosis of liver cancer. Background Art

[0002] Liver cancer is a general term for malignant tumors of the liver, which are divided into two categories: primary liver cancer (originating from hepatocytes or bile duct epithelial cells) and secondary liver cancer (metastasis from tumors in other organs to the liver). Among primary liver cancers, hepatocellular carcinoma (HCC) accounts for approximately 90%. Liver cancer has an insidious onset and progresses rapidly. Approximately 60%-70% of patients are already in the middle or late stages when diagnosed, missing the opportunity for radical treatment. However, the population of patients with chronic liver disease is large, but only a portion of them will progress to liver cancer. In order to screen for liver cancer more quickly and accurately, some early prediction models for liver cancer are currently being constructed to assist in screening, thereby optimizing the allocation of medical resources and cost-effectiveness. For example, the patent application with application number 202310109670.6 discloses a method for constructing a liver cancer risk prediction model and a network calculator thereof, which includes the following steps: S1. Acquiring the corresponding clinical and test data of the research subjects; S2. Screening of independent prediction features; S3. Construction of a prediction model; S4. Generation of a network calculator; S5. Calculation of the predicted probability of liver cancer occurrence. Integrating the familial hepatitis B-related liver cancer risk prediction model and the network calculator into the electronic medical record system can provide clinicians with electronic decision-making, better help clinicians assess the risk of familial hepatitis B-related liver cancer, and thus conduct regular follow-up tests on patients at high risk of liver cancer, so as to identify and intervene in the occurrence of liver cancer at an early stage, thereby improving the patient's prognosis; For example, the patent application with application number 202311779195.X discloses a method and system for predicting the risk of recurrence after surgery for early-stage liver cancer, which involves the fields of biological genes, bioinformatics, and medical diagnosis technology. It is based on establishing a model for predicting the risk of recurrence after surgery for early-stage liver cancer and predicting the risk level classification based on the model; including: obtaining clinical data and mutation data of early-stage liver cancer patients through public databases and performing preprocessing and data filtering to obtain initial modeling features; screening key features based on the initial modeling features, the univariate Cox proportional risk model, and the multivariate Cox regression model and establishing a primary Cox proportional risk model based on the key features, where the key features are related to survival risk; determining the primary Cox proportional risk model as a model for predicting the risk of recurrence after surgery for early-stage liver cancer, and predicting the risk of recurrence based on the model. The present invention makes full use of data resources, comprehensively considers multiple biological characteristics, and improves the accuracy of prediction; However, combined with some common liver cancer diagnosis and prediction models, we found that the current prediction models still have certain shortcomings. For example, the current prediction models use risk scores to predict liver cancer, which is not accurate and reliable enough, and lacks a clinical diagnosis and treatment evaluation system. This is not conducive to improving the early diagnosis rate of HCC, and cannot predict the clinical treatment effect of HCC in advance, and cannot predict the risk of postoperative recurrence. Therefore, we propose a method for constructing a liver cancer early diagnosis risk prediction model to address the above-mentioned issues. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for constructing a risk prediction model for the early diagnosis of liver cancer, so as to address the problems of the current prediction model proposed in the above-mentioned background art, which uses risk scoring to predict liver cancer, but is not accurate and reliable enough, and lacks a clinical diagnosis and treatment evaluation system, which is not conducive to improving the early diagnosis rate of HCC, and cannot predict the clinical treatment effect of HCC in advance, and cannot predict the risk of postoperative recurrence.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for constructing a risk prediction model for early diagnosis of liver cancer, the method comprising the following steps: Step 1: Prepare clinical data; Step 2: Process, analyze, and learn clinical data to build a project database for "Clinical Diagnostic Pathway for Hepatocellular Carcinoma Based on Big Data"; Step 3: Determine the main diagnostic points and relevant indicators affecting postoperative outcomes in the temporal diagnosis plan for each disease, and formulate rules for extracting diagnosis and treatment pathways; Step 4: Construct a quality and efficiency evaluation indicator system for the diagnosis and treatment pathway set to calculate the probability of early diagnosis of liver cancer and the probability of postoperative recurrence; Step 5: Establish a Markov model for early screening, diagnosis and risk prediction of postoperative recurrence of liver cancer.

[0005] Preferably, the clinical data preparation in step 1 includes importing case data from HIS and LIS systems, which includes general information, laboratory test data, imaging data, oncology indicators, pathogenic factors, clinical examination information, metabolomics, and environmental factors / lifestyle.

[0006] Preferably, the processing of clinical data in step 2 mainly includes: cleaning, integrating, transforming, and reducing the multi-source heterogeneous data of hospitalized patients with hepatocellular carcinoma, cleaning and fusing the data, building a database, and then completing the classification and standardization of analysis documents through a natural language processing system.

[0007] Preferably, the analysis of clinical data in step 2 includes: using a Cox regression model to explore the risk level and confidence interval of various biological, environmental, and lifestyle factors that contribute to the occurrence of HCC, which can be used to predict disease risk in high-risk populations; Among them, various clinical data related to HCC during the disease process were extracted to establish an HCC clinical diagnosis and treatment database. Through the Cox regression model, the risk level and confidence interval of each data indicator were mined as an evaluation model affecting prognosis and predicting recurrence.

[0008] Preferably, the learning of clinical data in step 2 includes: performing data fusion and standardization of multiple factors in the occurrence and development of HCC through deep learning convolutional neural networks, predicting the occurrence and progression of the disease, integrating and processing data information and image information of HCC and control samples, establishing a systematic prediction model and diagnosis and treatment evaluation system for HCC through a deep learning CNN framework, and combining it with the Cox disease risk prediction model to improve prediction sensitivity and specificity.

[0009] Preferably, the formulation of diagnosis and treatment pathway extraction rules in step 3 includes the following steps: Step 1: Based on 70% of the HCC case data in the database, a solution to the KSP problem for extracting the diagnosis and treatment pathway set for HCC was proposed through research on heuristic search strategies, application of wave propagation ideas, the impact of network complexity on algorithm applicability, and pulse-coupled neural network models and their applications. Step 2: Determine the rules for capturing the diagnosis and treatment pathway through literature review, Delphi method, and verification and correction based on the captured diagnosis and treatment pathway results; ultimately, the treatment pathway set for hepatocellular carcinoma is captured.

[0010] Preferably, the quality-efficiency evaluation of the diagnosis and treatment pathway set in step 4 includes the following steps: Step 3: Using the Apriori algorithm, we constructed a quality-efficacy evaluation index system based on 30% of the HCC case data in the database and clinical characteristics. Step 4: Determine the postoperative recurrence status of HCC through expert consultation and literature review, and calculate the recurrence probability using the MSM package model.

[0011] Compared with existing technologies, the present invention has at least the following beneficial effects: This method for constructing a risk prediction model for early diagnosis of liver cancer utilizes the ANN principle to establish an HCC early diagnosis risk prediction model and an HCC clinical diagnosis and treatment pathway evaluation system. This facilitates the development of HCC disease risk screening software, enables early identification of high-risk HCC populations, and improves the early diagnosis rate of HCC. It also provides optimized clinical diagnosis and treatment decisions for HCC treatment. Based on a large sample collection of HCC patients, combined with HCC susceptibility factors, metabolic and clinical biomarkers, relevant imaging and clinical indicators, a comprehensive evaluation system that can be used to predict HCC risk, evaluate disease diagnosis and treatment effects, and predict postoperative recurrence was established and systematically evaluated, laying the foundation for the early diagnosis, prevention, and treatment of HCC. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 Schematic diagram of the data processing flow of the present invention; Figure 3 This is a schematic diagram of the process structure for extracting and establishing the diagnosis and treatment pathway set of the present invention. DETAILED DESCRIPTION

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention, so that the implementation process of how this application applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0014] In order to solve the problems existing in the prior art, please refer to Figure 1-3 The present invention provides the following technical solutions: a method for constructing a risk prediction model for early diagnosis of liver cancer, the risk prediction model construction method comprising the following steps: step one: preparing clinical data; step two: processing, analyzing, and learning the clinical data, and constructing a project database of "clinical diagnostic pathway for hepatocellular carcinoma based on big data"; step three: determining the main diagnostic points and related indicators affecting postoperative care in the temporal diagnostic scheme of the disease, and formulating rules for extracting the diagnosis and treatment pathway; step four: constructing a quality and efficiency evaluation index system for the diagnosis and treatment pathway set, and calculating the probability of early diagnosis of liver cancer and the probability of postoperative recurrence; step five: establishing a Markov model for risk prediction of early screening, diagnosis, and postoperative recurrence of liver cancer.

[0015] Furthermore, the clinical data preparation in step 1 includes importing case data from HIS and LIS systems, including general information, laboratory test data, imaging data, oncology indicators, pathogenic factors, clinical examination information, metabolomics, and environmental factors / lifestyle; In a specific application scenario, all clinical data of multiple cases of hepatocellular carcinoma are collected, and a corresponding case capture program is designed. The capture is combined with manual entry to collect case information and establish an information database.

[0016] Furthermore, the processing of clinical data in the second step mainly includes: cleaning, integrating, transforming, and reducing the multi-source heterogeneous data of hospitalized patients with hepatocellular carcinoma, cleaning and fusing the data, building a database, and then completing the classification and standardization of analysis documents through the natural language processing system. The analysis of clinical data in the second step includes: mining the risk level and confidence interval of each factor for the biological, environmental and lifestyle factors that lead to the occurrence of multiple HCC diseases through the Cox regression model, which can be used for disease risk prediction in high-risk groups, extracting various clinical data related to HCC during the disease process, and establishing The HCC clinical diagnosis and treatment database uses a Cox regression model to mine the risk level and confidence interval of each data indicator as an evaluation model that affects prognosis and predicts recurrence. The learning of clinical data in step 2 includes: using a deep learning convolutional neural network to fuse and standardize data on multiple factors involved in the occurrence and development of HCC, predicting the occurrence and progression of the disease, integrating and processing data and image information of HCC and control samples, and establishing a systematic prediction model and diagnosis and treatment evaluation system for HCC using a deep learning CNN framework. This is combined with the Cox disease risk prediction model to improve prediction sensitivity and specificity. In specific application scenarios, deep learning (DL) convolutional neural networks (CNNs) are used to fuse and standardize data on numerous factors involved in the occurrence and development of HCC, predicting the onset and progression of the disease. Combined with the Cox disease risk prediction model, this method improves prediction sensitivity and specificity, achieving an AUC (Area Under Curve) of over 85%. By integrating and processing data and image information from HCC and control samples, and leveraging the deep learning CNN framework, a systematic prediction model and diagnosis and treatment evaluation system for HCC is established, improving prediction sensitivity and specificity to an AUC exceeding 90%.

[0017] Furthermore, the formulation of the diagnosis and treatment pathway extraction rules in the step three includes the following steps: Step 1: Based on 70% of the hepatocellular carcinoma case data in the database, through the research on heuristic search strategies, the application of wave transmission ideas, the research on the impact of network complexity on algorithm applicability, and the research on pulse coupled neural network models and applications, a solution to the KSP problem of capturing the diagnosis and treatment pathway set for hepatocellular carcinoma is proposed; Step 2: The diagnosis and treatment pathway capture rules are determined through literature review, Delphi method and verification and correction based on the diagnosis and treatment pathway capture results; finally, the treatment pathway set for hepatocellular carcinoma is captured, and the quality-efficiency evaluation of the diagnosis and treatment pathway set in the step four includes the following steps: Step 3: Through the Apriori algorithm, based on 30% of the hepatocellular carcinoma case data in the database combined with clinical characteristics, an evaluation index system for quality-efficiency evaluation is constructed; Step 4: The postoperative recurrence status of hepatocellular carcinoma is determined through expert consultation and literature review, and the recurrence probability is calculated using the MSM package model.

[0018] In specific application scenarios, data fusion and standardization are carried out, 70% of all cases are used for dynamic adaptation of the pathway system, diagnosis and treatment pathway sets are extracted, and a clinical diagnosis and treatment pathway system for hepatocellular carcinoma is established; the remaining 30% of cases are used for quality-efficiency evaluation and analysis of the diagnosis and treatment plan sets, providing a theoretical basis, data support and application tools for the disease clinical pathway optimization research system and application based on medical big data.

[0019] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of the present invention; the contents not described in detail in this specification belong to the existing technology known to professional and technical personnel in this field.

[0020] 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 scope of protection of the present invention.

Claims

1. A method for constructing a risk prediction model for early diagnosis of liver cancer, characterized in that: The risk prediction model construction method comprises the following steps: Step 1: Prepare clinical data; Step 2: Process, analyze, and learn clinical data to build a project database for "Clinical Diagnostic Pathway for Hepatocellular Carcinoma Based on Big Data"; Step 3: Determine the main diagnostic points and relevant indicators affecting postoperative outcomes in the temporal diagnosis plan for each disease, and formulate rules for extracting diagnosis and treatment pathways; Step 4: Construct a quality and efficiency evaluation indicator system for the diagnosis and treatment pathway set to calculate the probability of early diagnosis of liver cancer and the probability of postoperative recurrence; Step 5: Establish a Markov model for early screening, diagnosis and risk prediction of postoperative recurrence of liver cancer.

2. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 1, characterized in that: The clinical data preparation in step 1 includes importing case data from the HIS and LIS systems, including general information, laboratory test data, imaging data, oncology indicators, pathogenic factors, clinical examination information, metabolomics, and environmental factors / lifestyle.

3. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 1, characterized in that: The processing of clinical data in step 2 mainly includes: cleaning, integrating, transforming, and reducing the multi-source heterogeneous data of hospitalized patients with hepatocellular carcinoma, cleaning and fusing the data, building a database, and then completing the classification and standardization of analysis documents through a natural language processing system.

4. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 3, characterized in that: The analysis of clinical data in step 2 includes: using a Cox regression model to identify the risk levels and confidence intervals of various biological, environmental, and lifestyle factors that contribute to the development of HCC, which can be used to predict disease risk in high-risk populations; Among them, various clinical data related to HCC during the disease process were extracted to establish an HCC clinical diagnosis and treatment database. Through the Cox regression model, the risk level and confidence interval of each data indicator were mined as an evaluation model affecting prognosis and predicting recurrence.

5. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 4, characterized in that: The learning of clinical data in step 2 includes: using a deep learning convolutional neural network to fuse and standardize data on various factors involved in the occurrence and development of HCC, predicting the occurrence and progression of the disease, integrating and processing data and image information of HCC and control samples, and establishing a systematic prediction model and diagnosis and treatment evaluation system for HCC using a deep learning CNN framework, combined with the Cox disease risk prediction model to improve prediction sensitivity and specificity.

6. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 1, characterized in that: The formulation of the diagnosis and treatment pathway extraction rules in step 3 includes the following steps: Step 1: Based on 70% of the HCC case data in the database, a solution to the KSP problem for extracting the diagnosis and treatment pathway set for HCC was proposed through research on heuristic search strategies, application of wave propagation ideas, the impact of network complexity on algorithm applicability, and pulse-coupled neural network models and their applications. Step 2: Determine the rules for capturing the diagnosis and treatment pathway through literature review, Delphi method, and verification and correction based on the captured diagnosis and treatment pathway results; ultimately, the treatment pathway set for hepatocellular carcinoma is captured.

7. The method for constructing a risk prediction model for early diagnosis of liver cancer according to claim 1, characterized in that: The quality-efficiency evaluation of the diagnosis and treatment pathway set in step 4 includes the following steps: Step 3: Using the Apriori algorithm, we constructed a quality-efficacy evaluation index system based on 30% of the HCC case data in the database and clinical characteristics. Step 4: Determine the postoperative recurrence status of HCC through expert consultation and literature review, and calculate the recurrence probability using the msmpackage model.

Images