.png){kind=icon}
.png){kind=icon}
Return
Integrated interpretability machine learning and clinlabomics analysis for the risk prediction of hepatocellular carcinoma
.png){kind=icon}
OA
AI
W
Y
Z
Y
F
L
D
J
C
DOI：10.1186/s12885-026-16709-5.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This study aimed to develop a hepatocellular carcinoma (HCC) risk prediction model based on clinlabomics and develop an online prediction tool to provide a novel approach for early HCC diagnosis. We retrospectively collected clinical and laboratory data from 1,017 patients (576 HCC cases, 358 cirrhosis cases, and 83 chronic viral hepatitis cases), randomly dividing them into a training set (798 cases) and a validation set (219 cases). Key variables were selected using LASSO logistic regression and variable importance scoring, followed by the construction of seven machine learning models. Model performance was comprehensively evaluated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA). The optimal model was interpreted using SHapley Additive exPlanations (SHAP). Subsequently, we developed an online prediction tool for real-time risk prediction and validated its predictive performance. A total of 16 variables significantly related to HCC were selected for model construction, including: age, liver nodules (≥ 1 cm), CRP, RBC, PDW, AFP, PIVKA-II, IBIL, TBA, GASR, AAAR, ALBI, CK, INR, FIB, and FDP. Among the seven machine learning models developed, the SVM model demonstrated optimal performance, achieving AUC values of 0.962 (training set) and 0.877 (validation set). In prospective validation, the SVM-based online prediction tool demonstrated 85% concordance with clinical diagnosis and achieved an AUC of 0.735. The SVM model developed using clinlabomics demonstrates high predictive performance for HCC. The implemented online prediction tool (https://pl-riskmodel.shinyapps.io/workrun10/) provides a convenient means for clinical screening. • Successfully established and validated a risk prediction model for HCC. • Maintained high diagnostic performance even in patients negative for traditional serum biomarkers. • Developed and validated an online prediction tool, enhancing the model’s feasibility for real-world clinical application. • The model could serve as a valuable supplement to existing HCC diagnostic systems.
Keywords:
Hepatocellular carcinoma
Clinlabomics
Machine learning
Risk prediction model
Journal
IF:
3.4
Papers:
2.0W
Citations:
4.7W
Organization
Cited Papers