Related Experiment Video
Updated: Jul 15, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
Artificial intelligence-based models for colorectal cancer diagnosis using laboratory tests: an exploratory
Jian-Mei Lin1,2,3, Hai-Yan Huang1,2,3, Hui-Liu Tan1,2,3
1Department of Clinical Laboratory, The Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou, China.
Translational Cancer Research
|July 14, 2026
Summary
This study developed a diagnostic model for colorectal cancer (CRC) using routine lab tests and age. The Random Forest model achieved high accuracy, offering a potential new screening strategy for early CRC detection.
Area of Science:
- Oncology
- Biomarkers
- Machine Learning
Background:
- Colorectal cancer (CRC) is often diagnosed late due to subtle early symptoms and lack of effective screening.
- Delayed diagnosis complicates treatment and highlights the need for population-based screening strategies.
- Developing accurate diagnostic models is crucial for early detection and improved patient outcomes.
Purpose of the Study:
- To construct a diagnostic model for colorectal cancer (CRC) using readily available laboratory parameters and demographic data.
- To explore a potential screening strategy for population-based CRC tumor census.
- To evaluate the performance of machine learning algorithms in predicting CRC risk.
Main Methods:
- Retrospective case-control study analyzing laboratory parameters and clinical data from CRC patients and healthy controls.
- Utilized Mann-Whitney U test for intergroup comparisons and Receiver Operating Characteristic (ROC) curve analysis for biomarker evaluation.
- Employed Least Absolute Shrinkage and Selection Operator (LASSO) regression for variable selection and Random Forest model for classification, with SHapley Additive exPlanations (SHAP) for feature importance.
Main Results:
- Identified four key predictors for CRC diagnosis: total protein (TP), carcinoembryonic antigen (CEA), age, and apolipoprotein A1 (ApoA1).
- The Random Forest model demonstrated superior diagnostic performance with an area under the curve (AUC) of 0.946 and a Brier score of 0.101.
- Statistically significant differences were observed in age and serum biomarkers between CRC patients and controls (P<0.05).
Conclusions:
- Developed an effective risk stratification model for CRC screening utilizing routine clinical laboratory parameters and age.
- The model shows promise for population-based screening and early detection of colorectal cancer.
- Further multi-center, prospective validation is recommended to confirm generalizability and mitigate potential overfitting.
