Predicting liver metastasis in pancreatic neuroendocrine tumors with an interpretable machine learning algorithm: a...

Area of Science:

• Oncology
• Medical Informatics
• Machine Learning

Background:

• Liver metastasis is a common complication in pancreatic neuroendocrine tumors (PaNETs).
• Metastasis significantly impacts patient prognosis and treatment decisions.
• Accurate prediction of liver metastasis is crucial for personalized patient management.

Purpose of the Study:

• To develop and validate machine learning algorithms for predicting liver metastasis in PaNETs patients.
• To identify independent risk factors associated with liver metastasis.
• To create a clinical decision support tool for personalized treatment.

Main Methods:

• Utilized data from the Surveillance, Epidemiology, and End Results (SEER) database (2010-2021).
• Employed Boruta and LASSO for feature selection, identifying T-stage, N-stage, tumor size, grade, surgery, lymphadenectomy, chemotherapy, and bone metastasis as risk factors.
• Developed and compared 10 machine learning models, with Gradient Boosting Machine (GBM) showing superior performance.

Main Results:

• The GBM model achieved an AUC of 0.937 and an accuracy of 0.87.
• Decision curve analysis and calibration curves confirmed the GBM model's clinical utility and predictive accuracy.
• SHapley Additive exPlanations (SHAP) identified surgery, N-stage, and T-stage as primary predictive factors.

Conclusions:

• The GBM model demonstrates high performance in predicting liver metastasis in PaNETs.
• The developed web-based calculator, based on the GBM algorithm, offers a valuable tool for clinical decision-making.
• This predictive model supports the development of personalized medical strategies for PaNETs patients.