Validation of AI Lung Cancer Risk Model in Black-Dominant Population Shows Promising Results

A recent breakthrough in lung cancer detection has been demonstrated through the validation of the AI-powered risk assessment tool, Sybil, in a primarily Black patient population at UI Health, part of the University of Illinois Chicago. Presented at the 2025 World Conference on Lung Cancer by the International Association for the Study of Lung Cancer, the study highlights how Sybil's deep learning algorithm effectively predicts future lung cancer risk over a period of up to six years following a single low-dose CT (LDCT) scan.

Traditionally, the validation studies for Sybil involved cohorts predominantly composed of white individuals, with over 90% of participants being white. However, this new analysis focused on a diverse group where 62% identified as Non-Hispanic Black, 13% as Hispanic, and 4% as Asian, making it a significant step towards addressing racial disparities in lung cancer screening.

The research, led by the University of Illinois Hospital & Clinics, involved evaluating 2,092 baseline LDCTs collected from 2014 to 2024. Of these, 68 patients developed lung cancer, with follow-up durations extending up to 10.2 years. The study's results demonstrated high predictive accuracy across different racial groups, with Sybil achieving an Area Under the Curve (AUC) of 0.94 after one year and maintaining strong performance (AUC of 0.79) up to six years.

This means the model is highly capable of distinguishing between patients who will and will not develop lung cancer within this timeframe. The findings also remained consistent when analyzing Black participants specifically and after excluding cancers diagnosed shortly after screening.

The cohort's diversity and the model’s robust performance suggest that Sybil may be unbiased regarding race and ethnicity, making it a promising tool for improving early detection and reducing disparities in lung cancer outcomes, especially among underserved populations.

The Sybil Implementation Consortium, comprising institutions like Mass General Brigham, WellStar Health System, and MIT, plans to move forward with prospective clinical trials to incorporate Sybil into real-world clinical workflows, enhancing lung cancer screening practices.

This development offers hope for a more equitable approach to early lung cancer detection, potentially saving lives through more tailored and accessible risk assessment tools.