Machine learning model improves treatment selection in non-small cell lung cancer
MD Anderson Research Highlight July 25, 2025
A significant challenge in developing treatment plans for patients with metastatic non-small cell lung cancer (NSCLC) is deciding whether to use immune checkpoint inhibitors alone or in combination with chemotherapy. Monotherapy is effective for many patients and is associated with fewer adverse effects. However, for patients who do not respond, the crucial time lost could lead to tumor progression. Unfortunately, there is a lack of reliable data to guide clinicians in this area. A team led by , used over 2,300 cases from four different cancer centers to develop a machine learning model called A-STEP (Attention-based Scoring for Treatment Effect Prediction) that employs 34 different variables to calculate an individual¡¯s potential benefit from combination therapy. In a simulation on an external data set, the model recommended treatment changes in more than half of the patients, and those treated with its recommendation showed improved progression-free survival after two years. While this specific use needs continued validation, this study demonstrates the potential of this approach to inform complex clinical decisions. Learn more in .
This is a promising strategy for improving clinical decision-making for patients beyond the current standard of single biomarker-based decisions.¡±