Preoperative prognosis prediction method and device for colorectal cancer

Abstract

The invention discloses a colorectal cancer preoperative prognosis prediction method and device, and the method comprises the steps: calculating a systematic inflammation index and a body composition index of a target based on obtained object preoperative baseline data, the baseline data comprising basic clinical data, blood indexes and abdominal CT images; carrying out importance evaluation on the target systemic inflammation index, the body composition index and the basic clinical data, carrying out dimensionality reduction and regularization processing on the evaluated indexes, and screening to obtain a target index; and constructing a comprehensive prognosis prediction model based on the target index, and outputting a target risk score through the comprehensive prognosis prediction model so as to carry out preoperative prognosis prediction. According to the application, the scientificity, accuracy and intelligent level of preoperative risk prediction of the colorectal cancer patient are improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence and big data analysis technology, and in particular to a method and device for predicting the preoperative prognosis of colorectal cancer. Background Technology

[0002] Colorectal cancer (CRC) is one of the most common malignant tumors of the digestive system worldwide, with its incidence and mortality rates continuing to rise. Although the TNM staging system (AJCC 8th edition) remains the primary basis for assessing the prognosis of colorectal cancer, clinical practice has revealed significant differences in recurrence and survival even within the same stage, suggesting that the traditional staging system fails to adequately reflect patient physiological differences and tumor biological characteristics. In recent years, numerous studies have demonstrated that host factors such as systemic inflammatory responses and body composition status play crucial roles in tumor development, postoperative recovery, and long-term prognosis.

[0003] However, existing research often focuses on the analysis of single inflammation or single body composition parameters, lacking systematic studies that organically combine the two. Furthermore, traditional statistical methods such as univariate or multivariate Cox regression struggle to handle multidimensional, nonlinear, and significantly interactive data structures, resulting in insufficient model generalization ability.

[0004] Therefore, it is crucial to integrate and analyze inflammation and body composition parameters to improve the scientific rigor, accuracy, and intelligence of preoperative risk prediction for colorectal cancer patients. Summary of the Invention

[0005] The main objective of this invention is to provide a method and device for predicting the preoperative prognosis of colorectal cancer, thereby improving the scientific rigor, accuracy, and intelligence of preoperative risk prediction for colorectal cancer patients.

[0006] In a first aspect, this application provides a method for predicting the preoperative prognosis of colorectal cancer, wherein the method includes the following steps: Based on the acquired preoperative baseline data of the subjects, the systemic inflammatory markers and body composition indicators of the targets were calculated. The baseline data included: basic clinical data, blood indicators and abdominal CT images. The importance of the systemic inflammatory markers, body composition markers and basic clinical data of the target were assessed, and the assessed markers were subjected to dimensionality reduction and regularization to screen out the target markers. A comprehensive prognostic prediction model is constructed based on the target indicators, and a target risk score is output through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

[0007] In conjunction with the first aspect mentioned above, as an optional implementation method, the Boruta Random Forest algorithm is used to assess the importance of the target's systemic inflammatory indicators, body composition indicators, and basic clinical data. The evaluated indicators are input into the LASSO-Cox regression model to screen non-zero coefficient indicators under ten-fold cross-validation, thereby reducing the dimensionality and regularizing the indicators, and finally selecting the target indicators, which include: TNM stage, age, SMD, SFI, SII and LMR.

[0008] In conjunction with the first aspect mentioned above, as an optional implementation method, the target indicator is input into the constructed Cox proportional hazards model, and the regression coefficients corresponding to each indicator are obtained by fitting. A comprehensive prognostic prediction model is constructed based on the regression coefficients and target indicators corresponding to each indicator.

[0009] In conjunction with the first aspect mentioned above, as an optional implementation method, the target indicator is input into the comprehensive prognostic prediction model to output a target risk score through the comprehensive prognostic prediction model; The integrated prognostic prediction model is: Risk Score = ,in The regression coefficients corresponding to the indicators. For the corresponding indicators.

[0010] In conjunction with the first aspect above, as an optional implementation method, the systemic inflammatory markers of the target are calculated based on the blood indicators in the obtained preoperative baseline data of the subject. The systemic inflammatory markers include: NLR, PLR, SII and LMR. The abdominal CT images obtained from the preoperative baseline data of the subjects were segmented into skeletal muscle and adipose tissue. Based on the segmented skeletal muscle and adipose tissue, the target's body composition indices are calculated, including SMI, SMD, SFI, and IMFI.

[0011] In conjunction with the first aspect mentioned above, as an optional implementation method, the basic clinical data and preoperative blood indicators in the acquired preoperative baseline data of the target are uniformly formatted and standardized.

[0012] In conjunction with the first aspect mentioned above, as an optional implementation method, multiple imputation is used to fill in missing values ​​in the basic clinical data and preoperative blood indicators.

[0013] In conjunction with the first aspect mentioned above, as an optional implementation method, the acquired preoperative baseline data of the subjects is divided into a training queue and an independent external validation queue by utilizing spatiotemporal partitioning. The performance of the prognostic prediction model was quantitatively evaluated using clinical data from subjects in the independent external validation cohort. The performance evaluation of the prognostic prediction model included: Harrell's C-index, ROC curve, time-dependent AUC, Brier Score, Integrated Discriminant Improvement Index (IDI), calibration curve, and decision curve analysis (DCA).

[0014] In conjunction with the first aspect mentioned above, as an optional implementation method, the comprehensive prognostic model can be transformed into a visual nomogram and / or integrated into an interactive software tool to output individualized survival probability prediction results for the target.

[0015] Secondly, this application provides a device for predicting the preoperative prognosis of colorectal cancer, the device comprising: The calculation module is used to calculate the target's systemic inflammatory markers and body composition indicators based on the acquired preoperative baseline data of the subject, the baseline data including: basic clinical data, blood indicators and abdominal CT images; The processing module is used to assess the importance of the target's systemic inflammatory indicators, body composition indicators, and basic clinical data, and to perform dimensionality reduction and regularization on the assessed indicators to screen out the target indicators. The prediction module is used to construct a comprehensive prognostic prediction model based on the target indicators, and output a target risk score through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

[0016] This application provides a method and device for preoperative prognostic prediction of colorectal cancer. The method includes the following steps: calculating systemic inflammatory markers and body composition indicators of the target based on acquired preoperative baseline data of the subject. The baseline data includes: basic clinical data, blood indicators, and abdominal CT images; assessing the importance of the systemic inflammatory markers, body composition indicators, and basic clinical data of the target, and performing dimensionality reduction and regularization processing on the assessed indicators to screen out target indicators; constructing a comprehensive prognostic prediction model based on the target indicators, and outputting a target risk score through the comprehensive prognostic prediction model for preoperative prognosis prediction. This application improves the scientific rigor, accuracy, and intelligence of preoperative risk prediction for colorectal cancer patients.

[0017] It should be understood that the above general description and the following detailed description are merely exemplary and do not limit the invention. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0019] Figure 1This is a flowchart of a method for predicting the preoperative prognosis of colorectal cancer provided in the embodiments of this application; Figure 2 This is a schematic diagram of a colorectal cancer preoperative prognosis prediction device provided in an embodiment of this application. Detailed Implementation

[0020] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0021] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. Some of the block diagrams shown in the drawings represent functional entities and do not necessarily correspond to physically or logically independent entities.

[0022] The embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0023] Reference Figure 1 , Figure 1 The diagram shown is a flowchart of a preoperative prognostic prediction method for colorectal cancer provided by the present invention. Figure 1 As shown, the method includes the following steps: Step S101: Based on the acquired preoperative baseline data of the subject, calculate the target's systemic inflammatory markers and body composition indicators. The baseline data includes: basic clinical data, blood indicators, and abdominal CT images.

[0024] It should be explained that systemic inflammatory markers (such as NLR, PLR, SII, LMR, etc.) reflect the overall state of the body's immune response and inflammatory microenvironment, and can affect tumor immune escape and micrometastasis formation; while body composition parameters (such as SMI, SMD, SFI, IMFI) can objectively assess a patient's muscle mass and fat distribution through quantitative CT analysis, representing their nutritional metabolic reserves and stress resistance, and are closely related to postoperative complications, chemotherapy tolerance and overall survival.

[0025] Specifically, based on the blood indicators in the preoperative baseline data of the subject, the systemic inflammatory indicators of the target are calculated, including: NLR, PLR, SII, and LMR; the abdominal CT images in the preoperative baseline data of the subject are segmented into skeletal muscle and adipose tissue; based on the segmented skeletal muscle and adipose tissue, the body composition indicators of the target are calculated, including: SMI, SMD, SFI, and IMFI.

[0026] For ease of understanding and illustration, we collected baseline clinical data (age, sex, BMI, tumor stage (TNM stage), differentiation grade, etc.), preoperative blood parameters (neutrophils, lymphocytes, platelets, albumin, etc.), and abdominal CT images within three months prior to surgery from multicenter colorectal cancer patients. It should be noted that baseline data includes: basic clinical data, blood parameters, and abdominal CT images.

[0027] Systemic inflammatory markers were calculated using standard formulas, including NLR = neutrophils / lymphocytes, PLR = platelets / lymphocytes, LMR = lymphocytes / monocytes, SII = platelets × neutrophils / lymphocytes, and PNI = albumin + 5 × lymphocyte count.

[0028] The CT images were semi-automatically segmented into skeletal muscle and adipose tissue at the L3 level using SliceOmatic software, and parameters such as skeletal muscle index (SMI), skeletal muscle density (SMD), subcutaneous fat index (SFI), and intramuscular fat index (IMFI) were calculated.

[0029] In one embodiment, before calculating the target's systemic inflammatory markers and body composition indicators, the following steps are performed: the basic clinical data and preoperative blood indicators from the acquired preoperative baseline data of the subject are uniformly formatted and standardized. Missing values ​​in the basic clinical data and preoperative blood indicators are imputed using multiple imputation.

[0030] To illustrate this, during the data processing phase, clinical data and blood indicators from different centers underwent standardized formatting and feature standardization to ensure dimensional consistency across variables. Furthermore, for missing values, the Multiple Imputation, ICE (Multiple Interpolation, Multiple Execution, and Multiple Component Imputation) method was employed. All data underwent standardized formatting and missing value processing to ensure data quality and comparability.

[0031] Step S102: Assess the importance of the target's systemic inflammatory markers, body composition markers, and basic clinical data, and perform dimensionality reduction and regularization on the assessed markers to screen out the target markers.

[0032] Specifically, the Boruta Random Forest algorithm was used to assess the importance of the target’s systemic inflammatory markers, body composition markers, and basic clinical data. The evaluated indicators are input into the LASSO-Cox regression model to screen non-zero coefficient indicators under ten-fold cross-validation, thereby reducing the dimensionality and regularizing the indicators, and finally selecting the target indicators, which include: TNM stage, age, SMD, SFI, SII and LMR.

[0033] To illustrate this, to screen variables significantly related to overall survival, the Boruta algorithm based on random forest was first used to assess the importance of variables, and the Z-score threshold was used to determine the features to be retained. Then, the LASSO-Cox regression model was used with 10-fold cross-validation to further screen variables with non-zero coefficients, thereby achieving feature dimensionality reduction and regularization to prevent model overfitting. Finally, six core indicators—TNM stage, age, SMD, SFI, SII, and LMR—were selected as model input variables. Essentially, the Boruta algorithm was first used to assess the importance of all indicators (systemic inflammatory markers, body composition markers, and basic clinical data). This importance assessment can be understood as scoring indicators significantly related to patient survival to determine which indicators are important and which are not. After the assessment, the LASSO-Cox regression model was used to select the final six core indicators: TNM stage, age, SMD, SFI, SII, and LMR.

[0034] Step S103: Construct a comprehensive prognostic prediction model based on the target indicators, and output a target risk score through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

[0035] Specifically, the target indicator is input into the constructed Cox proportional hazards model to obtain the regression coefficient corresponding to each indicator; based on the regression coefficient corresponding to each indicator and the target indicator, a comprehensive prognostic prediction model is constructed.

[0036] The target indicator is input into the comprehensive prognostic prediction model to output a target risk score; wherein the comprehensive prognostic prediction model is: Risk Score = ,in The regression coefficients corresponding to the indicators. For the corresponding indicators.

[0037] For ease of understanding and illustration, the comprehensive prognostic prediction model is: Risk Score = The six core indicators (TNM stage, age, SMD, SFI, SII, and LMR) are used as inputs to the model to obtain a risk score.

[0038] In one embodiment, after constructing a comprehensive prognostic prediction model based on the target indicator and outputting a target risk score through the comprehensive prognostic prediction model, the process includes: Using spatiotemporal partitioning, the acquired objects were divided into a training queue and an independent external validation queue. The performance of the prognostic prediction model was quantitatively evaluated using the clinical data of the objects in the independent external validation queue. The performance evaluation of the prognostic prediction model included: Harrell's C-index, ROC curve, time-dependent AUC, BrierScore, Integrated Discriminant Improvement Index (IDI), calibration curve, and decision curve analysis (DCA). It should be noted that the clinical data included baseline data and blood parameters, etc.

[0039] Understandably, after completing model construction, this invention evaluates model robustness through a multi-level, multi-faceted validation process. First, the data is divided into training queues and independent external validation queues based on different centers. Second, the predictive performance of the model is comprehensively evaluated in the external validation queues, including Harrell's C-index, ROC curve, time-dependent AUC, BrierScore, and Integrated Discriminant Improvement Index (IDI). Simultaneously, calibration curves are plotted to assess predictive consistency, and decision curve analysis (DCA) is used to determine the model's net clinical benefit at different thresholds. Furthermore, this invention further confirms the model's generalization ability and clinical usability across different populations and parameter settings through subgroup validation, sensitivity analysis, and threshold robustness analysis.

[0040] In one embodiment, after constructing a comprehensive prognostic prediction model based on the target indicators and outputting a target risk score through the comprehensive prognostic prediction model, the method further includes: converting the comprehensive prognostic model into a visual nomogram and / or integrating it into an interactive software tool to output individualized survival probability prediction results. That is, individualized prediction is achieved using nomograms, and an interactive web tool is built using Python Streamlit for real-time clinical use.

[0041] In one embodiment, a 60-year-old male patient with rectal cancer is used as an example: 1. Data Acquisition: ① Clinical Data: Extract patient gender (male), age (60 years), height (1.70 m), weight (68 kg), TNM stage (stage III), and preoperative laboratory test indicators, including neutrophil count, lymphocyte count, platelet count, serum albumin level, etc.; ② Imaging Data: Extract abdominal CT plain scan images within three months before surgery from the hospital PACS system, with a slice thickness of 5 mm, located at the L3 vertebral level.

[0042] 2. Data preprocessing: ① Image screening: Manually confirm the single-layer axial images of L3 level; ② Image normalization: Standardize the pixel values ​​to [-150, 250] HU to ensure consistency in tissue density calculation; ③ Body region segmentation: Use SliceOmatic 5.0 software to semi-automatically segment skeletal muscle, subcutaneous fat and visceral fat regions.

[0043] 3. Feature Extraction and Calculation: Based on the segmentation results, the skeletal muscle area (SMA = 112.5 cm²), subcutaneous fat area (SFA = 138.7 cm²), and intramuscular fat area (IMFA = 1.824 cm²) were calculated. Furthermore, the skeletal muscle index (SMI) was calculated as 38.9 cm² / m², skeletal muscle density (SMD) as 36.4 HU, subcutaneous fat index (SFI) as 48.0 cm² / m², and intramuscular fat index (IMFI) as 0.631 cm² / m². Systemic inflammatory parameters were calculated based on hematological indicators: NLR = 3.4, PLR = 178, SII = 560, and LMR = 2.9.

[0044] 4. Results Output and Clinical Application: Based on the interactive webpage toolbar generated by the model, after inputting patient data, the system automatically calculates the 1-year, 3-year, and 5-year survival probabilities as 0.975, 0.900, and 0.845, respectively, classifying the patient as a medium- to high-risk patient. These results can guide preoperative intensive nutritional intervention and precise postoperative follow-up, enabling individualized prognostic management.

[0045] In summary, this method collects clinical characteristics, laboratory inflammatory markers, and abdominal CT images within three months preoperatively from multiple centers of colorectal cancer patients. It extracts multidimensional feature variables, including blood inflammatory parameters (NLR, PLR, SII, LMR, PNI, etc.) and CT-derived body composition parameters (SMI, SMD, SFI, IMFI, etc.). The Boruta random forest algorithm is used for preliminary feature screening, and the LASSO-Cox regression model is combined to complete variable dimensionality reduction and risk coefficient fitting, thereby establishing a comprehensive prognostic model integrating multidimensional physiological features. The model has been validated in multiple centers, achieving a C-index of 0.748 (0.714 on the validation set), significantly outperforming individual inflammatory or body composition models. Based on this invention, the model can achieve individualized prediction of 1-, 3-, and 5-year survival probabilities for preoperative colorectal cancer patients. The results are output through nomograms and an intelligent system, providing a scientific basis for precise stratified management and preoperative intervention decisions. This method exhibits good stability, interpretability, and generalizability, significantly improving the accuracy of preoperative risk assessment for colorectal cancer, and has high clinical application and research translational value. This method analyzes routine preoperative blood inflammatory data (such as white blood cell and platelet counts) and CT scan images (to calculate quantitative parameters of muscle and fat), combining these data with machine learning algorithms to build a predictive model. After doctors input patient data, the model outputs individualized survival probabilities for 1, 3, and 5 years, displaying key influencing factors through simple and intuitive charts. This method uses existing routine hospital examination data (blood tests and CT scans), eliminating the need for additional examinations, and can continuously optimize the model based on data from different hospitals, helping doctors to more accurately formulate preoperative treatment and follow-up plans.

[0046] The beneficial effects of this application include: by deeply integrating systemic inflammatory indicators with quantitative body composition features from CT scans, and combining this with machine learning algorithms to establish a colorectal cancer prognostic prediction system, a comprehensive assessment of patients' multidimensional biological characteristics is achieved. Compared with traditional prognostic assessment methods that rely on TNM staging, this invention, for the first time, incorporates inflammatory status and body composition parameters as complementary features into the same intelligent modeling framework, capturing the complex nonlinear correlation between the two, thus more comprehensively reflecting the dynamic balance between the body's immunity, metabolism, and tumor progression. Through the combined screening mechanism of the Boruta algorithm and LASSO-Cox regression, redundant variables are effectively reduced and overfitting is avoided, enabling the model to achieve high predictive accuracy while maintaining stability. The constructed comprehensive model performs excellently in multi-center validation, with a C-index higher than 0.74, significantly outperforming single clinical, inflammatory, or body composition models. The system uses visualized nomogram output and stratified risk reports, providing physicians with clear individualized prognostic probabilities and the weight distribution of key influencing factors, facilitating clinical understanding and decision-making. The input data for this method all come from routine preoperative examinations (peripheral blood routine and CT images), requiring no additional testing costs, and possessing strong scalability and practicality. Meanwhile, the system possesses self-learning and modular features, enabling dynamic updates and algorithm retraining based on data characteristics from different hospitals or populations, achieving continuous performance optimization and generalization. Furthermore, this framework can be extended to other digestive system tumors and multi-omics data integration analysis, providing a feasible path for building AI-driven precision medicine decision support systems. In summary, this invention achieves breakthroughs in data fusion dimensions, algorithm optimization depth, and clinical application feasibility, significantly improving the scientific rigor, accuracy, and intelligence of preoperative risk prediction for colorectal cancer patients, and is of great significance for promoting personalized cancer treatment and precise prognostic assessment.

[0047] Reference Figure 2 , Figure 2 The diagram shown is a schematic of a preoperative prognostic prediction device for colorectal cancer provided by the present invention. Figure 2 As shown, the device includes: Calculation module 201: It is used to calculate the target's systemic inflammatory markers and body composition indicators based on the acquired preoperative baseline data of the subject, the baseline data including: basic clinical data, blood indicators and abdominal CT images.

[0048] Processing module 202: It is used to assess the importance of the target's systemic inflammatory indicators, body composition indicators and basic clinical data, and to perform dimensionality reduction and regularization on the assessed indicators to screen out the target indicators.

[0049] Prediction module 203: It is used to construct a comprehensive prognostic prediction model based on the target indicators, and output a target risk score through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

[0050] Furthermore, in one possible implementation, the processing module is also used to assess the importance of the target's systemic inflammatory indicators, body composition indicators, and basic clinical data using the Boruta Random Forest algorithm. The evaluated indicators are input into the LASSO-Cox regression model to screen non-zero coefficient indicators under ten-fold cross-validation, thereby reducing the dimensionality and regularizing the indicators, and finally selecting the target indicators, which include: TNM stage, age, SMD, SFI, SII and LMR.

[0051] Furthermore, in one possible implementation, a construction module is used to input the target indicator into the constructed Cox proportional hazards model and fit the regression coefficient corresponding to each indicator. A comprehensive prognostic prediction model is constructed based on the regression coefficients and target indicators corresponding to each indicator.

[0052] Furthermore, in one possible implementation, the prediction module is also used to input the target indicator into the comprehensive prognostic prediction model, so as to output a target risk score through the comprehensive prognostic prediction model; The integrated prognostic prediction model is: Risk Score = , where β is the regression coefficient corresponding to the index, and X is the corresponding index.

[0053] Furthermore, in one possible implementation, the calculation module is also configured to calculate the target's systemic inflammatory markers based on blood indicators in the acquired preoperative baseline data of the subject, wherein the systemic inflammatory markers include: NLR, PLR, SII, and LMR; The abdominal CT images obtained from the preoperative baseline data of the subjects were segmented into skeletal muscle and adipose tissue. Based on the segmented skeletal muscle and adipose tissue, the target's body composition indices are calculated, including SMI, SMD, SFI, and IMFI.

[0054] Furthermore, in one possible implementation, the processing module is also used to perform unified formatting and standardization processing on the basic clinical data and preoperative blood indicators in the acquired preoperative baseline data of the subject.

[0055] Furthermore, in one possible implementation, the processing module is also used to perform missing value imputation on the basic clinical data and preoperative blood indicators using multiple interpolation.

[0056] Furthermore, in one possible implementation, the processing module is also used to divide the acquired preoperative baseline data of the object into a training queue and an independent external validation queue by utilizing spatiotemporal partitioning. The performance of the prognostic prediction model was quantitatively evaluated using clinical data from subjects in the independent external validation cohort. The performance evaluation of the prognostic prediction model included: Harrell's C-index, ROC curve, time-dependent AUC, Brier Score, Integrated Discriminant Improvement Index (IDI), calibration curve, and decision curve analysis (DCA).

[0057] Furthermore, in one possible implementation, the prediction module is also used to transform the comprehensive prognostic model into a visual nomogram and / or integrate it into an interactive software tool to output individualized survival probability prediction results for the target.

[0058] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements all or part of the method steps of the above method.

[0059] This application implements all or part of the processes in the above methods, which can also be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0060] Based on the same inventive concept, embodiments of this application also provide an electronic device, including a memory and a processor. The memory stores a computer program that runs on the processor. When the processor executes the computer program, it implements all or part of the method steps described above.

[0061] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device through various interfaces and lines.

[0062] Memory can be used to store computer programs and / or modules. The processor performs various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function (e.g., sound playback, image playback, etc.); the data storage area can store data created based on the use of the mobile phone (e.g., audio data, video data, etc.). Furthermore, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMedia Cards (SMC), Secure Digital (SD) cards, Flash Cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0063] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, servers, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0064] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), servers, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for predicting the preoperative prognosis of colorectal cancer, characterized in that, include: Based on the acquired preoperative baseline data of the subjects, the systemic inflammatory markers and body composition indicators of the targets were calculated. The baseline data included: basic clinical data, blood indicators and abdominal CT images. The importance of the systemic inflammatory markers, body composition markers and basic clinical data of the target were assessed, and the assessed markers were subjected to dimensionality reduction and regularization to screen out the target markers. A comprehensive prognostic prediction model is constructed based on the target indicators, and a target risk score is output through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

2. The method according to claim 1, characterized in that, The importance of the systemic inflammatory markers, body composition indicators, and basic clinical data of the target is assessed, and the assessed indicators are then subjected to dimensionality reduction and regularization to screen out the target indicators, including: The Boruta Random Forest algorithm was used to assess the importance of the target's systemic inflammatory markers, body composition indicators, and basic clinical data. The evaluated indicators are input into the LASSO-Cox regression model to screen non-zero coefficient indicators under ten-fold cross-validation, thereby reducing the dimensionality and regularizing the indicators, and finally selecting the target indicators, which include: TNM stage, age, SMD, SFI, SII and LMR.

3. The method according to claim 1, characterized in that, The construction of a comprehensive prognostic prediction model based on the target index includes: The target indicators are input into the constructed Cox proportional hazards model, and the regression coefficients corresponding to each indicator are obtained by fitting. A comprehensive prognostic prediction model is constructed based on the regression coefficients and target indicators corresponding to each indicator.

4. The method according to claim 1, characterized in that, The comprehensive prognostic prediction model outputs a target risk score for preoperative prognostic prediction, including: The target indicator is input into the comprehensive prognostic prediction model to output a target risk score through the comprehensive prognostic prediction model; The integrated prognostic prediction model is: Risk Score = ,in The regression coefficients corresponding to the indicators. For the corresponding indicators.

5. The method according to claim 1, characterized in that, Based on the acquired preoperative baseline data of the subject, the systemic inflammatory markers and body composition indicators of the target are calculated, including: Based on the blood parameters obtained from the preoperative baseline data of the target, the systemic inflammatory markers of the target are calculated, including: NLR, PLR, SII and LMR; The abdominal CT images obtained from the preoperative baseline data of the subjects were segmented into skeletal muscle and adipose tissue. Based on the segmented skeletal muscle and adipose tissue, the target's body composition indices are calculated, including SMI, SMD, SFI, and IMFI.

6. The method according to claim 1, characterized in that, Prior to the systemic inflammatory markers and body composition markers of the calculation targets, the following are included: The basic clinical data and preoperative blood indicators in the obtained preoperative baseline data of the subjects were uniformly formatted and standardized.

7. The method according to claim 6, characterized in that, Also includes: Multiple imputation was used to fill in missing values ​​in the basic clinical data and preoperative blood indicators.

8. The method according to claim 1, characterized in that, The process of constructing a comprehensive prognostic prediction model based on the target indicators and outputting a target risk score through the comprehensive prognostic prediction model includes: Using spatiotemporal partitioning, the acquired objects are divided into a training queue and an independent external validation queue; The performance of the prognostic prediction model was quantitatively evaluated using clinical data from subjects in the independent external validation cohort. The performance evaluation of the prognostic prediction model included: Harrell's C-index, ROC curve, time-dependent AUC, Brier Score, Integrated Discriminant Improvement Index (IDI), calibration curve, and decision curve analysis (DCA).

9. The method according to claim 1, characterized in that, After constructing a comprehensive prognostic prediction model based on the target indicator and outputting a target risk score through the comprehensive prognostic prediction model, the process further includes: The comprehensive prognostic model is transformed into a visual nomogram and / or integrated into an interactive software tool to output individualized survival probability predictions for the target.

10. A device for predicting the preoperative prognosis of colorectal cancer, characterized in that, include: The calculation module is used to calculate the target's systemic inflammatory markers and body composition indicators based on the acquired preoperative baseline data of the subject, the baseline data including: basic clinical data, blood indicators and abdominal CT images; The processing module is used to assess the importance of the target's systemic inflammatory indicators, body composition indicators, and basic clinical data, and to perform dimensionality reduction and regularization on the assessed indicators to screen out the target indicators. The prediction module is used to construct a comprehensive prognostic prediction model based on the target indicators, and output a target risk score through the comprehensive prognostic prediction model to perform preoperative prognostic prediction.

Images