مجلة إقتصاد المال والأعمال
Volume 11, Numéro 1, Pages 647-666
2026-05-31
Decision Tree Models For Managing And Predicting Cancer Patient Outcomes
Authors : Medini Atman .
Abstract
This study aimed to develop and evaluate decision tree classifiers for predicting cancer patient outcomes using demographic, clinical, and socioeconomic variables, focusing on model interpretability and generalizability. A dataset of 10,000 cancer patient records, including factors such as age, gender, cancer type, treatment type, comorbidities, and anthropometric features, was analyzed. Four models were developed: one unconstrained, two pruned with different complexity levels, and one trained and validated with an 80/20 split. Performance was assessed through accuracy, precision, recall, and F1-score, with feature importance identifying key predictors. The unconstrained model overfitted despite perfect training accuracy, while pruned models showed better generalization, and the optimal pruned model balanced accuracy and interpretability. Height, weight, and age were consistent primary predictors. Overall, pruned and validated decision tree models proved transparent, reliable, and suitable for supporting personalized clinical decision-making in oncology
Keywords
Decision trees ; Cancer outcome prediction ; Artificial intelligence in healthcare ; ; Interpretability ; Predictive modeling
Les articles similaires
مسعودي محمد رضا
.
دراز خليدة
.
ص 449-480.
الغامدي سعيد
.
pages 32-60.
Daoudi Nadia
.
Youcefi Nour Elhouda
.
Aribi Noureddine
.
Belaoudmou Ramzeddine
.
pages 11-19.
Mahamdi Yassine
.
Boubakeur Ahmed
.
Mekhaldi Abdelouahab
.
Benmahamed Youcef
.
pages 1-5.
