Deep-Learning Model Drives Histopathology-Based Biomarker Detection in NSCLC

Objective:

To explore the use of deep-learning algorithms for detecting biomarkers in tissue samples from patients with non-small cell lung cancer (NSCLC).

Key Findings:

• AI algorithms achieved areas under the curve of 0.87 for EGFR, 0.96 for ALK, 0.88 for BRAF, and 0.83 for MET.
• The negative predictive value was 98.6% for EGFR, 99.7% for ALK, 100% for BRAF, and 99.6% for MET.
• False positives were more common than false negatives, with variability by gene.

Interpretation:

Integrating AI biomarker classifiers early in the diagnostic process could enhance the detection of targetable alterations, optimize testing, and improve clinical outcomes.

Limitations:

• The study recommends prospective clinical studies before real-world application of AI classifiers.
• False positive rates varied widely by gene, indicating potential challenges in clinical settings.

Conclusion:

AI tools for biomarker detection in NSCLC have the potential to streamline testing and improve treatment decisions, ultimately leading to better patient outcomes.