Mortality risk prediction for primary appendiceal cancer

Area of Science:

• Oncology
• Medical Informatics
• Computational Biology

Background:

• Accurate cancer survival prediction is vital for clinical decisions and patient guidance.
• Appendiceal neoplasms require precise prognostic tools for effective management.

Purpose of the Study:

• To develop the first machine learning algorithm for predicting mortality risk in appendiceal neoplasm patients.
• To create a patient-specific, web-based tool for risk assessment.

Main Methods:

• Utilized the Surveillance, Epidemiology, and End Results database (2000-2019) for patient data.
• Employed machine learning models including extreme gradient boost, random forest, neural network, and logistic regression.
• Validated algorithm performance and identified key predictive variables.

Main Results:

• Included 16,579 patients; extreme gradient boost showed highest accuracy for 1-, 5-, and 10-year mortality prediction.
• The 10-year model achieved an AUC of 0.909 (±0.006) post-cross-validation.
• Key predictors included disease grade, histology, lymph node status, and distant disease presence.

Conclusions:

• Developed and validated a novel prognostic model for appendiceal neoplasms using machine learning.
• The model incorporates diverse patient, surgical, and pathological variables.
• Achieved excellent predictive accuracy, outperforming existing nomograms.