Conference
Salah Eddine El Herrag, Noria Harir, Soraya Moulessehoul
1st National Scientific Day, 2025 Nov 11
1st National Scientific Day, 2025 Nov 11
DOI:
https://osf.io/f3szu/files/wk8uf
Abstract
Cite
Cite
APA
Click to copy
Salah Eddine, E. H., Noria, H., & Soraya, Moulessehoul. (2025). Artificial intelligence for predicting chemotherapy and targeted therapy eligibility in cancer patients: a machine learning approach. https://doi.org/https://osf.io/f3szu/files/wk8uf
Chicago/Turabian
Click to copy
Salah Eddine, El Herrag, Harir Noria, and Soraya Moulessehoul. “Artificial Intelligence for Predicting Chemotherapy and Targeted Therapy Eligibility in Cancer Patients: a Machine Learning Approach.” 1st National Scientific Day, 2025.
MLA
Click to copy
Salah Eddine, El Herrag, et al. Artificial Intelligence for Predicting Chemotherapy and Targeted Therapy Eligibility in Cancer Patients: a Machine Learning Approach. 2025, doi:https://osf.io/f3szu/files/wk8uf.
BibTeX Click to copy
@conference{salah2025a,
title = {Artificial intelligence for predicting chemotherapy and targeted therapy eligibility in cancer patients: a machine learning approach},
year = {2025},
month = nov,
day = {11},
series = {1st National Scientific Day},
doi = {https://osf.io/f3szu/files/wk8uf},
author = {Salah Eddine, El Herrag and Noria, Harir and Soraya, Moulessehoul},
month_numeric = {11}
}
Abstract
Introduction: Artificial intelligence (AI) is increasingly used in oncology to analyze complex clinical and molecular data. Accurate prediction of therapy eligibility can guide personalized treatment. This study aimed to develop machine learning models to predict chemotherapy administration and targeted therapy eligibility in colorectal cancer patients.
Methods: Clinical and molecular data from 250 colorectal cancer patients were analyzed, including age, BMI, tumor site, KRAS/NRAS/BRAF mutation status, tumor stage, and serum CEA/CA19-9 levels. Random Forest classifiers were trained in R to predict chemotherapy administration and targeted therapy eligibility. The dataset was split into training (70%) and testing (30%) sets. Performance was evaluated using accuracy, sensitivity, specificity, Kappa, and area under the ROC curve (AUC).
Results: For chemotherapy prediction, the Random Forest model achieved an out-of-bag error rate of 13.9% and an accuracy of 86.3% (95% CI: 0.76–0.93), with Kappa 0.71, sensitivity 75.9%, specificity 93.2%, and AUC 0.873. Key predictors were tumor stage, KRAS/BRAF status, BMI, CEA level, and tumor site.
For targeted therapy eligibility, the model achieved an out-of-bag error rate of 15.6% and accuracy of 90.4% (95% CI: 0.81–0.96), with Kappa 0.70, sensitivity 96.5%, specificity 68.8%, and AUC 0.917. Important predictors were BRAF/KRAS status, tumor stage, CEA level, and tumor site.
Conclusions: AI-based models accurately predicted chemotherapy and targeted therapy eligibility. Tumor characteristics and molecular markers consistently guided treatment decisions, illustrating the potential of AI to support personalized oncology care.
Keywords: Artificial intelligence, machine learning, chemotherapy, targeted therapy.
