Extreme gradient boost machine learning model accurately predicts mortality risk in primary appendiceal cancer patients

Researchers developed a novel machine learning algorithm using extreme gradient boost to predict 1-, 5-, and 10-year mortality risk in patients with primary appendiceal cancer. The model exhibited high prediction accuracy, with a maximum area under the curve of 0.909 for the 10-year model. Variables such as disease grade, histology, lymph node involvement, and distant disease significantly influenced predictive ability. The model outperformed previous nomograms, incorporating a wide range of patient, surgical, and pathologic variables for improved prognosis prediction.