Cancer early warning management method and system based on physical examination report

Abstract

The invention discloses a cancer early warning management method and system based on a physical examination report, and belongs to the field of medicines.The method comprises the steps that health data of individuals are collected, and unstructured texts in the physical examination report are processed through combination of optical character recognition and a natural language processing technology; calculating a cancer risk based on a single physical examination report, evaluating a specific cancer by using a cancer risk scoring system, and analyzing whether an index exceeds a normal range or not through a rule engine; based on historical data trend prediction risks, index change rates are calculated, and an abnormal trend is analyzed and predicted in combination with a time sequence; multi-modal deep learning is utilized to predict individual cancer risks, and a Bayesian network is combined to perform joint analysis on multiple factors to generate a health management scheme; further examination is arranged according to the early warning level, and a screening strategy is provided for specific cancers. According to the method, the OCR + NLP technology is adopted, and the data quality is improved. And in combination with a multi-modal deep learning model, abnormal changes are found in advance, and early warning is realized.

Description

Technical Field

[0001] The present invention belongs to the field of medical technology, and more specifically, particularly relates to a cancer early warning management method and system based on a physical examination report. Background Art

[0002] Currently, cancer has become one of the main diseases causing death globally, and early detection and intervention are crucial for improving the survival rate. However, existing cancer screening methods mainly rely on regular physical examinations, imaging examinations (such as CT, MRI), and specific biomarker detections (such as tumor marker blood tests). These methods have the following deficiencies: Traditional physical examination data often only reflects the health status of an individual at a certain moment, unable to dynamically evaluate the long-term health trend of an individual, and prone to missing the early cancer risk. Physical examination reports are usually in unstructured text format, lacking standardized analysis tools and making it difficult to comprehensively evaluate multiple indicators. Existing cancer risk prediction methods usually rely on single indicators or expert experience, and fail to fully utilize multi-modal data. Traditional statistical analysis methods have limitations in processing high-dimensional, non-linear, biomedical data and are difficult to accurately predict cancer risks. Existing cancer screening processes mostly adopt fixed screening cycles and fail to adjust the screening strategy according to the individual's health status. For high-risk individuals, there is a lack of personalized intervention plans and no dynamic health management can be provided.

[0003] To overcome the above problems, the present invention proposes a cancer early warning management method based on a physical examination report, which extracts physical examination data through OCR + NLP technology, combines multi-modal deep learning and Bayesian network for cancer risk prediction, and uses time series analysis to evaluate the changing trend of indicators, realizes personalized cancer risk early warning, and automatically generates a health management plan. Summary of the Invention

[0004] In view of the above or existing problems of the cancer early warning management method and system based on a physical examination report, the present invention is proposed.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: An embodiment of the present invention provides a cancer early warning management method based on a physical examination report, including: collecting an individual's health data through a hospital, a physical examination center, or a wearable device, processing the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extracting the required indicators and storing them in a database; Calculate cancer risk based on a single physical examination report, evaluate specific cancers using a cancer risk scoring system, and analyze whether the indicators exceed the normal range through a rule engine; predict risks based on historical data trends, calculate the rate of change of indicators, and combine time series analysis to predict abnormal trends; use multi-modal deep learning to predict individual cancer risks, combine Bayesian networks for joint analysis of multiple factors, screen high-risk individuals, classify the early warnings, and generate a health management plan based on the analysis results; arrange further examinations according to the early warning level and provide screening strategies for specific cancers.

[0006] As a preferred solution of the cancer early warning management method based on the physical examination report described in the present invention, wherein: process the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extract the required indicators and store them in the database, including: Convert the color picture into a grayscale image and perform binarization processing. The calculation formula for the binarization processing is as follows: where T is the optimal binarization threshold, is the between-class variance; Detect the text direction, perform rotation correction, use Gaussian filtering to smooth the text edge, use PaddleOCR for text recognition, extract single characters, and then detect the complete text line; Use the BERT-BiLSTM-CRF named entity recognition architecture to recognize medical entities, and achieve the coherence of entity annotation through global optimization of CRF. The CRF calculation formula is as follows: where X is the input text, y is the recognized medical entity, and s(X,y) is the CRF score; Use the relational database MySQL and NoSQL to store the physical examination data.

[0007] As a preferred solution of the cancer early warning management method based on the physical examination report described in the present invention, wherein: calculate cancer risk based on a single physical examination report, and evaluate specific cancers using a cancer risk scoring system, including: Calculate the individual cancer risk CRS, which is calculated using a linear weighted model. The CRS calculation formula is as follows: where X i is the standardized health indicator, and W i is the weight; If CRS < 0.4, it is a low risk and no intervention is required; if 0.4 ≤ CRS < 0.7, it is a medium risk and a reexamination is required in 3 - 6 months; if CRS ≥ 0.7, it is a high risk and immediate medical treatment and further examination are required.

[0008] As a preferred solution of the cancer early warning management method based on physical examination reports according to the present invention, wherein: analyzing whether an indicator exceeds the normal range through a rule engine, including: Setting the normal range of the indicator, establishing an over-limit rule, and the rule engine adopts an if-then rule. IF the indicator value exceeds the normal range THEN mark it as abnormal. IF the indicator seriously exceeds the standard THEN trigger a health warning; setting the normal range of the physical examination indicator X as [X min , X max : wherein, X is a specific health indicator in the physical examination report, X min is the lower limit of the normal range of this indicator, X max is the upper limit of the normal range of this indicator, X norm is the normalized value; If X > X max , then it exceeds the standard; if X < X min , then it is on the low side.

[0009] As a preferred solution of the cancer early warning management method based on physical examination reports according to the present invention, wherein: predicting risks based on historical data trends, calculating the change rate of indicators, and combining time series analysis to predict abnormal trends, including: The physical examination data is represented by X t indicating a health indicator for the t-th physical examination, and calculating the change rate: wherein, X t is the indicator value at the current time point t, X t-1 is the indicator value of the previous physical examination, and V t is the change rate of this indicator between t - 1 and t; If V t continues to increase, it indicates that this indicator has an upward trend and further detection is required; Calculating abnormal fluctuation detection mutations. If the fluctuation of an indicator exceeds the set threshold, there is a health risk. By calculating the relative change rate: Dragging R t > 20%, it indicates that this indicator has an abnormal increase and further analysis is required; if R t < -20%, it is an abnormal decrease due to treatment or other factors; Using the exponential smoothing method to predict future indicator values, and then comprehensively calculating the cancer risk score based on multiple physical examination indicators: wherein, w i is the weight of each indicator, and X norm is the normalized value of each indicator; If S > 0.8, it is a high risk, and immediate reexamination and imaging examination should be carried out; if 0.5 < S ≤ 0.8, it is a medium risk, and reexamination is recommended after 3 months; if S ≤ 0.5, it is a low risk, and regular physical examinations should be continued.

[0010] As a preferred solution of the cancer early warning management method based on physical examination reports according to the present invention, wherein: multi-modal deep learning is used to predict an individual's cancer risk, combined with a Bayesian network for joint analysis of multiple factors to screen high-risk individuals, grade the early warnings, and generate a health management plan based on the analysis results, including: Use a convolutional neural network to extract features from medical images to obtain a feature representation F of the images image , use a fully connected neural network to process physical examination data to obtain a feature representation F of physical examination indicators health , use a graph neural network or a recurrent neural network to model gene data to obtain a gene feature representation F genetic , Use a recurrent neural network to model lifestyle data to obtain a lifestyle feature representation F lifestyle , fuse the features extracted from multiple data sources to generate a comprehensive feature representation: Through an attention mechanism, different modalities of data have different weights during feature fusion, and the fused feature F multi-modal is input into a deep neural network to predict the probability of cancer; Construct a Bayesian network containing various risk factors, and use the Bayesian inference method to estimate the conditional probabilities of each edge in the network according to historical data, and calculate the probability of cancer occurrence for an individual: where Evidence is the information observed in the physical examination report and imaging data, P(cancer) is the prior probability of cancer, and P(evidence|cancer) is the probability of observing the evidence in the case of cancer occurrence; According to the predicted probability of cancer occurrence P(cancer), carry out early warning level classification: P(cancer) < 0.3 is a low risk, and regular physical examinations should be carried out; 0.3 ≤ P(cancer) < 0.6 is a medium risk, and close monitoring and regular reexamination are recommended; P(cancer) ≥ 0.6 is a high risk, and it is recommended to immediately carry out further examinations and diagnoses; physical examinations should be carried out once every six months for medium-risk individuals, and cancer screening and personalized treatment interventions should be carried out for high-risk individuals.

[0011] As a preferred solution of the cancer early warning management method based on physical examination reports according to the present invention, wherein: Arrange further examinations according to the early warning level, and provide screening strategies for specific cancers, including: For low-risk individuals, maintain healthy habits. For medium-risk individuals, detect cancer-related markers and perform imaging examinations every six months. For high-risk individuals, conduct in-depth cancer screenings every three months, including multiple imaging examinations, molecular diagnosis, and genetic testing; Based on the screening results, develop personalized health intervention plans. The health management plan includes controlling weight through nutritional counseling and diet plans to reduce the risk of cancer, and providing psychological support to high-risk patients to relieve mental burdens.

[0012] A cancer early warning management system based on physical examination reports, including: a data collection module for collecting individual health data through hospitals, physical examination centers, or wearable devices, processing unstructured text in physical examination reports through optical character recognition combined with natural language processing technology, extracting required indicators, and storing them in a database; A static risk assessment module for calculating cancer risks based on single physical examination reports, evaluating specific cancers using a cancer risk scoring system, and analyzing whether indicators exceed the normal range through a rule engine; A dynamic trend analysis module for predicting risks based on historical data trends, calculating the change rate of indicators, and predicting abnormal trends by combining time series analysis; A prediction and early warning module for predicting individual cancer risks using multi-modal deep learning, jointly analyzing multiple factors by combining Bayesian networks, screening high-risk individuals, classifying early warnings, and generating a health management plan based on the analysis results; A decision support module for arranging further examinations according to the early warning level and providing screening strategies for specific cancers.

[0013] A computing device, the computing device includes: At least one processor, a memory, and an input / output unit; Wherein, the memory is used to store computer programs, and the processor is used to call the computer programs stored in the memory to execute the steps of the cancer early warning management method based on physical examination reports.

[0014] A computer-readable storage medium, which includes instructions that, when running on a computer, cause the computer to execute the steps of the cancer early warning management method based on physical examination reports.

[0015] The beneficial effects of the present invention are as follows: The present invention adopts OCR + NLP technology to automatically identify and structurally process the unstructured text in the physical examination report, avoiding the errors caused by manual input and improving the data quality. By combining with a multi-modal deep learning model, it comprehensively analyzes physical examination indicators, biomarkers, gene data, imaging data, etc., improving the accuracy of cancer risk prediction. It uses time series analysis to calculate the change rate of physical examination indicators, establishes an individual health trend model, discovers abnormal changes in advance, and realizes early warning. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0017] Figure 1 It is a flowchart of the cancer early warning management method based on the physical examination report provided by the embodiment of the present invention.

[0018] Figure 2 It is a schematic structural diagram of the cancer early warning management system based on the physical examination report provided by the embodiment of the present invention.

[0019] Figure 3 It schematically shows a structural diagram of a medium according to an embodiment of the present invention.

[0020] Figure 4 It schematically shows a structural diagram of a computing device according to an embodiment of the present invention.

[0021] In the drawings, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed Embodiments

[0022] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification.

[0023] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0024] Second, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is separate or selectively mutually exclusive with other embodiments.

[0025] Embodiment The following refers to Figures 1 - 4 , Figure 1 is a flowchart of a cancer early warning management method based on a physical examination report provided for an embodiment of the present invention. It should be noted that the implementation manner of the present invention can be applied to any applicable scenario.

[0026] Figure 1 The process of the cancer early warning management method based on a physical examination report provided for an embodiment of the present invention shown in the figure includes: S1: Collect the health data of an individual through a hospital, a physical examination center, or a wearable device, process the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extract the required indicators, and store them in a database.

[0027] Preferably, convert the color picture into a grayscale image and perform binarization processing. The calculation formula for the binarization processing is as follows: where T is the optimal binarization threshold, is the between-class variance; Detect the text direction, perform rotation correction, smooth the text edge using Gaussian filtering, perform text recognition using PaddleOCR, extract single characters, and then detect the complete text line; Utilize the BERT - BiLSTM - CRF named entity recognition architecture to recognize medical entities, and achieve the coherence of entity annotation through CRF global optimization. The CRF calculation formula is as follows: where X is the input text, y is the recognized medical entity, and s(X, y) is the CRF score; Use the relational database MySQL and NoSQL to store the physical examination data.

[0028] Furthermore, assume the input physical examination text: The blood glucose of the patient is 7.2 mmol / L, which is higher than the normal range (3.9 - 6.1 mmol / L); After being processed by BERT - BiLSTM, the following scores are obtained: Token B - Physical Examination Items I - Physical Examination Items B - Value I - Value B - Reference Range I - Reference Range Blood Glucose 0.9 0.2 0.1 0.0 0.0 0.0 7.2 0.1 0.0 0.9 0.3 0.0 0.0 mmol / L 0.0 0.0 0.2 0.9 0.0 0.0 3.9 0.0 0.0 0.0 0.0 0.9 0.3 - 0.0 0.0 0.0 0.0 0.5 0.7 6.1 0.0 0.0 0.0 0.0 0.2 0.9 mmol / L 0.0 0.0 0.0 0.0 0.1 0.8 Let the state transition score W B-体检项目,I-体检项目 = 1.2, and the state 转移分数 WB-数值,I-数值 = 0.8; Final calculation: s(X,y) = (0.9 + 1.2) + (0.9 + 0.8) + (0.9 + 0.8) + (0.9 + 0.8) = 8.2; Select the tag sequence with the highest score and finally output: { "Physical examination item": "Blood glucose", "Value": "7.2 mmol / L", "Reference range": "3.9 - 6.1 mmol / L", "Status": "High" }}.

[0029] S2: Calculate the cancer risk based on a single physical examination report, evaluate a specific cancer using a cancer risk scoring system, and analyze whether the indicators exceed the normal range through a rule engine.

[0030] Preferably, calculate the individual cancer risk CRS using a linear weighted model. The formula for calculating CRS is as follows: Where X i is the standardized health indicator, and W i is the weight; If CRS < 0.4, it is a low risk and no intervention is required; if 0.4 ≤ CRS < 0.7, it is a medium risk and a reexamination is required in 3 - 6 months; if CRS ≥ 0.7, it is a high risk and immediate medical attention for further examination is required.

[0031] Preferably, set the normal range of the indicators, establish an over - limit rule. The rule engine uses if - then rules. IF the indicator value exceeds the normal range THEN mark it as abnormal. IF the indicator seriously exceeds the standard THEN trigger a health warning. Set the normal range of the physical examination indicator X as [X min , X max : Where X is a specific health indicator in the physical examination report, X min is the lower limit of the normal range of this indicator, X max is the upper limit of the normal range of this indicator, and X norm is the normalized value; If X > X max , then it exceeds the standard; if X < X min , then it is on the low side.

[0032] Furthermore, assume that the physical examination data of an individual is as follows: Health Index Actual Value Mean Value Standard Deviation Normalized Value Blood Glucose 7.8 5.6 15 1.47 Blood Pressure 150 120 15 2.00 Body Weight 85 70 10 1.50 Years of Smoking 10 5 3 1.67 According to the above - mentioned standardized health indicators and their corresponding weights, calculate the cancer risk score CRS of the individual: CRS = (X1 ⋅ W1) + (X2 ⋅ W2) + (X3 ⋅ W3) + (X4 ⋅ W4) CRS = 0.3675 + 0.60 + 0.30 + 0.4175 = 1.685 Since the calculated result is CRS = 1.685, this individual belongs to the high-risk category and needs to seek immediate medical attention for further examination; High-risk individuals (CRS ≥ 0.7) Health check suggestions: Conduct a detailed medical examination as soon as possible. It is recommended to perform imaging examinations (such as CT, MRI) and tumor marker tests. For high-risk cancer types (such as lung cancer, breast cancer, etc.), corresponding screenings (such as lung CT scans, mammograms, etc.) are recommended. For risk factors that are long-term controlled, such as blood sugar and blood pressure, long-term monitoring is recommended to ensure no deterioration. Maintain a healthy weight, increase exercise, and maintain a balanced diet.

[0033] For concomitant diseases such as diabetes and cardiovascular diseases, comprehensive management is required to avoid further deterioration of the condition.

[0034] Medium-risk individuals (0.4 ≤ CRS < 0.7) Health check suggestions: It is recommended to have a recheck every 3 - 6 months and pay attention to changes in physical examination indicators. Conduct routine screenings (such as blood tests, chest X-rays, etc.), and pay special attention to high-risk factors such as diabetes and cardiovascular diseases. Control weight, maintain moderate physical exercise, and reduce high-fat and high-sugar diets. For smokers, it is recommended to gradually quit smoking to reduce the health risks brought by tobacco.

[0035] Low-risk individuals (CRS < 0.4) Health check suggestions: Have a routine physical examination every year and pay attention to changes in health indicators. If there are no obvious health problems, no further screening is required, but it is recommended to maintain a healthy lifestyle. Continue to maintain a healthy diet, regular exercise, and avoid bad habits such as smoking and excessive drinking.

[0036] Furthermore, extract various health indicators X1, X2,..., X from the physical examination report n , set the normal range [X min , X max for each health indicator, compare the indicator values. For each indicator X, first perform normalization calculation to obtain X norm , if X > X max , then trigger an over-standard mark; if X < X min , then trigger a low mark; for over-standard situations, if it exceeds X max by a large amount (for example, exceeding the upper limit by 10%), then trigger a health warning.

[0037] Suppose blood glucose monitoring is carried out. The following are the physical examination data: Physical examination report: Blood glucose index X = 9.5 mmol / L Normal range: X min = 3.9, X max = 6.1 mmol / L Rule judgment: Step 1: Check whether the blood glucose value exceeds the normal range.

[0038] X = 9.5 mmol / L > X max = 6.1 mmol / L, meeting Rule 1 (exceeding the standard).

[0039] Step 2: Check whether a health warning is triggered.

[0040] The difference exceeding the maximum value of 6.1 is 9.5 - 6.1 = 3.4, exceeding the 10% threshold, so it meets Rule 3 (health warning); Output that the blood glucose value of this examinee exceeds the standard and the difference is large, so the system triggers a health warning and recommends further examination.

[0041] To improve the accuracy and intelligence of the system, the rule engine can be extended and optimized in the following ways: Dynamic threshold adjustment: Adjust the normal range [Xmin, Xmax] of health indicators according to factors such as the age, gender, and historical health records of different individuals; Combination of multiple rules: For complex situations, the rule engine can combine multiple health indicators for judgment. For example, if both blood glucose and weight exceed the standard, the system can comprehensively evaluate whether it belongs to a high-risk group; Historical data comparison: Combine the individual's historical health data for trend analysis. If a certain indicator rises or falls sharply in the short term, it represents a certain health risk; Intelligent warning: When a health warning is triggered, the system can not only give a warning, but also automatically generate specific health management suggestions, such as adjusting the diet and increasing exercise.

[0042] S3: Predict risks based on historical data trends, calculate the index change rate, and combine time series analysis to predict abnormal trends.

[0043] Preferably, the physical examination data is represented by X t indicating a health indicator for the t-th physical examination, and calculate the change rate: where X t is the indicator value at the current time point t, and X t-1是上一次体检的指标值，Vt is the change rate of this indicator between t - 1 and t; If Vt Continuously increasing indicates an upward trend in this indicator and further detection is required; Calculate the abnormal fluctuation to detect mutations. If the fluctuation of an indicator exceeds the set threshold, there is a health risk. By calculating the relative change rate: Drag R t > 20%, it indicates that this indicator has an abnormal increase and further analysis is required; if R t < -20%, it is an abnormal decrease due to treatment or other factors; Use the exponential smoothing method to predict future indicator values, and then comprehensively calculate the cancer risk score based on multiple physical examination indicators: Among them, w i is the weight of each indicator, X norm is the normalized value of each indicator; If S > 0.8, it is a high risk, and immediate reexamination and imaging examination are required; if 0.5 < S ≤ 0.8, it is a medium risk and reexamination is recommended after 3 months; if S ≤ 0.5, it is a low risk, and continue with regular physical examinations.

[0044] Furthermore, first, by calculating the change rate of the physical examination data, capture the degree of change of the health indicators within a certain period of time.

[0045] Blood glucose change rate: X t = 8.0 mmol / L, X t-1 = 7.2 mmol / L, V t = (8.0 - 7.2) / 7.2 × 100% = 11.11%; if the change rate V t continuously increases and exceeds the set threshold (such as 5%), it indicates that the blood glucose level shows a continuous upward trend, suggesting a risk of diabetes.

[0046] In addition to the change rate, the system also detects the trend of indicator changes. Assume that the change rates of consecutive n physical examination records are V1, V2,..., V n , if the change rates V t of these n times have a continuous upward or downward trend, and this trend exceeds the preset threshold (such as 5%), the system will trigger a health warning. For example, if the blood glucose values for three consecutive times are 6.0, 6.5, 7.0 mmol / L, and the calculated change rates are 8.33% and 7.69%, it can be inferred that the blood glucose has a gradually increasing trend, and further monitoring is required and diabetes screening may be recommended.

[0047] By calculating the relative change rate, further detect whether the index fluctuation exceeds the normal range. If the change amplitude of the index exceeds the set threshold (such as 20%), it indicates that the index has abnormal fluctuations, and this index needs to be concerned and the reasons behind it need to be further analyzed. For example, if in a physical examination report, a person's cholesterol level suddenly rises from 5.0 mmol / L to 6.3 mmol / L in one physical examination, then the relative change rate is: R t =6.3 - 5.05.0×100% = 26% Suppose in two consecutive physical examinations, a physical examination index such as blood pressure of an individual is 130 / 80 mmHg in the first physical examination and 160 / 100 mmHg in the second physical examination. The calculation of the relative change rate is as follows: Rt = 23.08% The fluctuation amplitude is greater than 20%, which belongs to abnormal fluctuation, and the system triggers a warning and prompts the user to seek medical examination as soon as possible.

[0048] S4: Use multi-modal deep learning to predict an individual's cancer risk, combine Bayesian network to conduct joint analysis on multiple factors, screen high-risk individuals, classify the early warnings, and generate a health management plan based on the analysis results.

[0049] Preferably, use a convolutional neural network to extract features from medical images to obtain the feature representation F of the images image , use a fully connected neural network to process the physical examination data to obtain the feature representation F of the physical examination indicators health , use a graph neural network or a recurrent neural network to model the gene data to obtain the gene feature representation F genetic , use a recurrent neural network to model the lifestyle data to obtain the lifestyle feature representation F lifestyle , fuse the features extracted from multiple data sources to generate a comprehensive feature representation: Through the attention mechanism, different modalities of data have different weights during feature fusion, and the fused feature F multi-modal is input into a deep neural network to predict the probability of cancer; Construct a Bayesian network containing various risk factors, and according to historical data, use the Bayesian inference method to estimate the conditional probabilities of each edge in the network and calculate the cancer occurrence probability of an individual: Among them, Evidence is the information observed in the physical examination report and image data, P(cancer) is the prior probability of cancer, and P(evidence|cancer) is the probability of observing the evidence in the case of cancer occurrence; According to the predicted probability of cancer occurrence P(cancer), the early warning level is divided as follows: When P(cancer) < 0.3, it is a low risk, and regular physical examinations are required; when 0.3 ≤ P(cancer) < 0.6, it is a medium risk, and close monitoring and regular reexaminations are recommended; when P(cancer) ≥ 0.6, it is a high risk, and further examinations and diagnoses are recommended immediately. Physical examinations are conducted on medium-risk individuals once every six months, and cancer screening and personalized treatment interventions are carried out on high-risk individuals.

[0050] Furthermore, by using convolutional layers, pooling layers, and fully connected layers, CNN can effectively extract local features in images and gradually synthesize high-level features. Through the stacking of multiple convolutional layers, the network can learn features at more abstract levels, thereby accurately identifying important information such as tumors and lesion areas. Physical examination data is usually unstructured tabular data, such as blood pressure, cholesterol, weight, etc. Through the connection of multiple layers of neurons, the fully connected network can learn the complex relationships between different physical examination indicators. Each physical examination data is regarded as an input feature, and the network outputs the contribution degree of this feature through weighted calculation. To avoid overfitting, L2 regularization or dropout techniques can be used to improve the generalization ability of the model, especially when the dimension of the input data is relatively high. Gene data often shows the relationship of mutual dependence between genes. GNN can process graph-structured data and learn the connections between gene nodes. For example, in a gene network, the interaction or regulatory relationship between genes can be represented by edges. GNN captures the complex associations between genes through an information propagation mechanism and generates a feature representation containing network topological relationships. Gene expression data has the property of time series. Lifestyle data, especially changes in behavior patterns, has time correlation. LSTM can effectively model this time-series data, capture the individual's behavior patterns changing over time, and thus provide a more accurate basis for cancer prediction.

[0051] S5: Arrange further examinations according to the early warning level and provide a screening strategy for specific cancers.

[0052] Preferably, for low-risk individuals, maintain healthy habits; for medium-risk individuals, detect cancer-related markers and conduct imaging examinations every six months; for high-risk individuals, conduct in-depth cancer screening every three months, and the screening content includes multiple imaging examinations, molecular diagnosis, and gene testing. Based on the screening results, develop a personalized health intervention plan. The health management plan includes controlling weight through nutritional counseling and diet plans to reduce the risk of cancer occurrence, and providing psychological support to high-risk patients to relieve mental burdens.

[0053] Furthermore, based on cancer screening results and an individual's genetic, physical examination, and imaging data, a personalized diet and nutrition plan is developed to help the individual control their weight and reduce the risk of cancer: Provide nutritional counseling for each high-risk individual and customize a daily diet plan through a smart APP. The system recommends foods that combat cancer and avoids foods that may trigger cancer based on genetic testing, weight management, and blood indicators. Real-time monitoring of an individual's diet and exercise data is carried out through smart health devices (such as smart kitchen devices, wearable devices, etc.) to help them achieve the goals of controlling weight and improving diet structure. For example, monitor the exercise volume through a smart watch and link it with the smart catering system to adjust the individual's nutritional intake.

[0054] For high-risk individuals, the system not only provides guidance on a healthy lifestyle but also pays attention to the patient's mental state to relieve their mental burden: Provide regular psychological counseling services for high-risk patients through a mental health management platform. This can include virtual psychiatrists (AI mental assistants based on natural language processing), which understand the patient's emotional fluctuations through regular conversations and help them reduce anxiety and stress. Help patients cope with cancer-related anxiety through online psychological counseling, emotional counseling, mental health education courses, etc. The system will also connect patients with others facing similar health challenges through a community support platform to provide mental support and emotional communication. Develop a personalized exercise plan for high-risk individuals to promote physical health and reduce the risk of cancer: Generate a suitable exercise plan based on individual health data (such as weight, heart rate, exercise ability, etc.) to help patients improve their immunity and physical fitness. The exercise plan will be adjusted at any time to ensure that patients exercise reasonably while ensuring safety. The diet plan can be automatically adjusted according to the exercise volume and dietary intake. For example, when an individual engages in high-intensity exercise, the system will adjust the energy intake in a timely manner to maximize the exercise effect.

[0055] After introducing the method of the exemplary embodiment of the present invention, next, reference is made to Figure 2 An explanation is given of the cancer early warning management system based on a physical examination report according to the exemplary embodiment of the present invention. The system includes: A data collection module for collecting an individual's health data through a hospital, a physical examination center, or a wearable device, processing the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extracting the required indicators, and storing them in a database; A static risk assessment module for calculating the cancer risk based on a single physical examination report, evaluating a specific cancer using a cancer risk scoring system, and analyzing whether the indicators exceed the normal range through a rule engine; A dynamic trend analysis module for predicting risks based on historical data trends, calculating the rate of change of indicators, and predicting abnormal trends by combining time series analysis; A prediction and early warning module, which is used to predict an individual's cancer risk using multimodal deep learning, combine a Bayesian network for joint analysis of multiple factors, screen high-risk individuals, classify the levels of early warnings, and generate a health management plan based on the analysis results; A decision support module, which is used to arrange further examinations according to the early warning levels and provide screening strategies for specific cancers.

[0056] After introducing the methods and devices of the exemplary embodiments of the present invention, next, refer to Figure 3 A computer-readable storage medium of the exemplary embodiments of the present invention will be described. Please refer to Figure 3 , which shows that the computer-readable storage medium is an optical disc 30, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it will implement the steps recorded in the above method embodiments. For example, collect an individual's health data through a hospital, a physical examination center or a wearable device, process the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extract the required indicators and store them in a database; calculate the cancer risk based on a single physical examination report, evaluate a specific cancer using a cancer risk scoring system, and analyze whether the indicators exceed the normal range through a rule engine; predict the risk based on the historical data trend, calculate the change rate of the indicators, and combine time series analysis to predict abnormal trends; predict an individual's cancer risk using multimodal deep learning, combine a Bayesian network for joint analysis of multiple factors, screen high-risk individuals, classify the levels of early warnings, and generate a health management plan based on the analysis results; arrange further examinations according to the early warning levels and provide screening strategies for specific cancers; the specific implementation manners of each step will not be repeated here.

[0057] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated one by one here.

[0058] After introducing the methods, devices and media of the exemplary embodiments of the present invention, next, refer to Figure 4 A computing device for cancer early warning management based on a physical examination report of the exemplary embodiments of the present invention.

[0059] Figure 4 The block diagram of an exemplary computing device 40 suitable for implementing the embodiments of the present invention is shown. The computing device 40 may be a computer system or a server. Figure 4The illustrated computing device 40 is only an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0060] As Figure 4 shown, the components of the computing device 40 may include, but are not limited to: one or more processors or processing units 401, a system memory 402, and a bus 403 that connects different system components (including the system memory 402 and the processing unit 401).

[0061] The computing device 40 typically includes a variety of computer system-readable media. These media can be any available media accessible by the computing device 40, including volatile and non-volatile media, removable and non-removable media.

[0062] The system memory 402 may include computer system-readable media in the form of volatile memory, such as random access memory (RAM) 4021 and / or cache memory 4022. The computing device 40 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, ROM 4023 can be used to read and write on a non-removable, non-volatile magnetic medium ( Figure 4 not shown in the figure and commonly referred to as a "hard disk drive"). Although not shown in Figure 4 the figure, a disk drive for reading and writing on a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing on a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM, or other optical medium) can be provided. In these cases, each drive can be connected to the bus 403 through one or more data medium interfaces. The system memory 402 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present invention.

[0063] A program / utility 4025 having a set (at least one) of program modules 4024 can be stored, for example, in the system memory 402, and such program modules 4024 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data, and the implementation of a network environment may be included in each or some combination of these examples. The program modules 4024 generally perform the functions and / or methods described in the embodiments of the present invention.

[0064] The computing device 40 may also communicate with one or more external devices 404 (such as a keyboard, a pointing device, a display, etc.). Such communication may be carried out through an input / output (I / O) interface 405. Moreover, the computing device 40 may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 406. As Figure 4 shown, the network adapter 406 communicates with other modules (such as the processing unit 401, etc.) of the computing device 40 through a bus 403. It should be understood that although Figure 4 not shown in the figure, other hardware and / or software modules may be used in combination with the computing device 40.

[0065] The processing unit 401 executes various functional applications and data processing by running programs stored in the system memory 402. For example, it collects the health data of an individual through a hospital, a physical examination center, or a wearable device, processes the unstructured text in the physical examination report through optical character recognition combined with natural language processing technology, extracts the required indicators and stores them in a database; calculates the cancer risk based on a single physical examination report, uses a cancer risk scoring system to evaluate a specific cancer, and analyzes whether the indicators exceed the normal range through a rule engine; predicts the risk based on the historical data trend, calculates the rate of change of the indicators, and combines time series analysis to predict abnormal trends; uses multi-modal deep learning to predict an individual's cancer risk, combines Bayesian networks to conduct a joint analysis of multiple factors, screens high-risk individuals, classifies the early warnings, and generates a health management plan based on the analysis results; arranges further examinations according to the early warning level and provides a screening strategy for a specific cancer.

[0066] The specific implementation manners of each step will not be repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the synchronous escape routing device based on multi-commodity flow are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present invention, the features and functions of the two or more units / modules described above may be embodied in one unit / module. Conversely, the features and functions of one unit / module described above may be further divided and embodied by multiple units / modules.

[0067] In the description of the present invention, it should be noted that the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0068] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above may refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0069] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0070] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0071] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0072] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical disks, etc., which can store program codes.

[0073] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions recorded in the foregoing embodiments, or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0074] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution.

Claims

1. A cancer early warning management method based on a physical examination report, characterized in that Including: Collecting the health data of individuals through hospitals, physical examination centers or wearable devices, processing the unstructured text in the physical examination reports by combining optical character recognition with natural language processing technology, extracting the required indicators and storing them in the database; Calculating the cancer risk based on a single physical examination report, evaluating specific cancers using a cancer risk scoring system, and analyzing whether the indicators exceed the normal range through a rule engine; Predicting the risk based on the historical data trend, calculating the change rate of the indicators, and combining time series analysis to predict abnormal trends; Using multi-modal deep learning to predict the individual cancer risk, combining Bayesian networks to conduct joint analysis on multiple factors, screening high-risk individuals, classifying the early warnings, and generating a health management plan based on the analysis results; Arranging further examinations according to the warning levels and providing screening strategies for specific cancers.

2. The cancer early warning management method based on a physical examination report according to claim 1, wherein The processing of the unstructured text in the physical examination reports by combining optical character recognition with natural language processing technology, extracting the required indicators and storing them in the database includes: Convert the color picture into a grayscale image and perform binarization processing. The calculation formula for the binarization processing is as follows: where T is the optimal binarization threshold, is the between-class variance; Detecting the text direction, performing rotation correction, smoothing the text edges using Gaussian filtering, performing text recognition using PaddleOCR, extracting individual characters, and then detecting the complete text lines; Using the BERT-BiLSTM-CRF named entity recognition architecture to identify medical entities, and realizing the coherence of entity annotation through the global optimization of CRF. The CRF calculation formula is as follows: Among them, X is the input text, y is the recognized medical entity, and s(X, y) is the CRF score; Using the relational database MySQL and NoSQL to store the physical examination data.

3. The cancer early warning management method based on a physical examination report according to claim 1, characterized in that The calculating of the cancer risk based on a single physical examination report and the evaluation of specific cancers using a cancer risk scoring system include: Calculating the individual cancer risk CRS using a linear weighted model, and the formula for CRS is as follows: Among them, X i is a standardized health indicator, and W i is the weight; If CRS < 0.4, it is a low risk and no intervention is required; if 0.4 ≤ CRS < 0.7, it is a medium risk and a reexamination is required in 3 - 6 months; if CRS ≥ 0.7, it is a high risk and immediate medical treatment and further examinations are required.

4. The cancer early warning management method based on a physical examination report according to claim 1, characterized in that The analysis of whether the indicators exceed the normal range through a rule engine includes: Set the normal range of indicators, establish over-limit rules. The rule engine uses if-then rules. IF the indicator value exceeds the normal range THEN mark it as abnormal. IF the indicator seriously exceeds the standard THEN trigger a health warning. Set the normal range of the physical examination indicator X as [X min , X max : Among them, X is a specific health indicator in the physical examination report, and X min is the lower limit of the normal range of this indicator, and X max is the upper limit of the normal range of this indicator, and X norm is the value after normalization; If X > X max , then it exceeds the standard; if X < X min , then it is on the low side.

5. The cancer early warning management method based on a physical examination report according to claim 1, wherein The predicting of the risk based on the historical data trend, calculating the change rate of the indicators, and combining time series analysis to predict abnormal trends include: The physical examination data is represented by X t indicating a health indicator for the t-th physical examination, and calculating the rate of change: where X t is the index value at the current time point t, X t-1 is the index value of the previous physical examination, and V t is the rate of change of this index between t - 1 and t; If V t continues to increase, it indicates that there is an upward trend in this indicator and further detection is required; Calculate abnormal fluctuations to detect mutations. If the fluctuation of an indicator exceeds the set threshold, there is a health risk. By calculating the relative change rate: Drag R t > 20%, it indicates that the indicator has an abnormal increase and further analysis is required; if R t < -20%, it is an abnormal decrease caused by treatment or other factors; Predict future metric values using the exponential smoothing method, and then comprehensively calculate the cancer risk score based on multiple physical examination metrics: Among them, w i is the weight of each index, and X norm is the normalized value of each index; If S > 0.8, it is a high risk and an immediate reexamination and imaging examination are required; if 0.5 < S ≤ 0.8, it is a medium risk and it is recommended to have a reexamination after 3 months; if S ≤ 0.5, it is a low risk and continue with regular physical examinations.

6. The cancer early warning management method based on a physical examination report according to claim 1, wherein The using of multi-modal deep learning to predict the individual cancer risk, combining Bayesian networks to conduct joint analysis on multiple factors, screening high-risk individuals, classifying the early warnings, and generating a health management plan based on the analysis results includes: Use a convolutional neural network to extract features from medical images and obtain the feature representation F of the images image , use a fully connected neural network to process the physical examination data and obtain the feature representation F of the physical examination indicators health , use a graph neural network or a recurrent neural network to model the gene data and obtain the gene feature representation F genetic , Modeling lifestyle data using a recurrent neural network to obtain a lifestyle feature representation F lifestyle , fusing features extracted from multiple data sources to generate a comprehensive feature representation: The attention mechanism is used to give different weights to data of different modalities during feature fusion, and the fused feature F multi-modal is input into a deep neural network to predict the probability of cancer; Construct a Bayesian network that includes various risk factors. Based on historical data, use the Bayesian inference method to estimate the conditional probabilities of each edge in the network and calculate the probability of cancer occurrence for an individual: Among them, Evidence is the information observed in the physical examination reports and imaging data, P(cancer) is the prior probability of cancer, and P(evidence|cancer) is the probability of observing the evidence in the case of cancer occurrence; According to the predicted cancer occurrence probability P(cancer), conduct early warning level classification: P(cancer) < 0.3 indicates low risk, and regular physical examinations are required; 0.3 ≤ P(cancer) < 0.6 indicates medium risk, and close monitoring and regular reexaminations are recommended; P(cancer) ≥ 0.6 indicates high risk, and further examinations and diagnoses are recommended immediately; physical examinations are conducted on medium-risk individuals every six months, and cancer screenings and personalized treatment interventions are performed on high-risk individuals.

7. The cancer early warning management method based on a physical examination report according to claim 1, characterized in that Based on the warning levels, further examinations are arranged, and screening strategies for specific cancers are provided, including: For low-risk individuals, healthy habits are maintained; for medium-risk individuals, cancer-related markers and imaging examinations are performed every six months; for high-risk individuals, in-depth cancer screenings are conducted every three months, including multiple imaging examinations, molecular diagnoses, and genetic tests; Based on the screening results, personalized health intervention plans are developed. The health management plan includes controlling weight through nutritional counseling and diet plans to reduce the risk of cancer, and providing psychological support to high-risk patients to relieve mental burdens.

8. A cancer early warning management system based on a physical examination report, characterized in that, Including: A data collection module for collecting an individual's health data through hospitals, physical examination centers, or wearable devices, processing unstructured text in physical examination reports through optical character recognition combined with natural language processing technology, extracting required indicators, and storing them in a database; A static risk assessment module for calculating cancer risks based on single physical examination reports, evaluating specific cancers using a cancer risk scoring system, and analyzing whether indicators exceed the normal range through a rule engine; A dynamic trend analysis module for predicting risks based on historical data trends, calculating the rate of change of indicators, and predicting abnormal trends in combination with time series analysis; A prediction and warning module for predicting an individual's cancer risk using multi-modal deep learning, jointly analyzing multiple factors in combination with a Bayesian network, screening high-risk individuals, classifying the warnings, and generating a health management plan based on the analysis results; A decision support module for arranging further examinations based on the warning levels and providing screening strategies for specific cancers.

9. A computing device, the computing device includes: At least one processor, a memory, and an input / output unit; Wherein, the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the steps of the cancer warning management method based on physical examination reports as described in any one of claims 1 to 7.

10. A computer-readable storage medium, which includes instructions that, when running on a computer, cause the computer to execute the steps of the cancer warning management method based on physical examination reports as described in any one of claims 1 to 7.

Images