Published medical AI research on evaluation bias as a 9th grader
Neil examined how standard evaluation protocols inflate reported accuracy in nail abnormality detection models. He showed that random data splits allow leakage between training and test sets, and that models with near-perfect accuracy under random splits decline sharply when evaluated across different data sources. Region-of-interest training focused on the nail plate improved external F1-score and reduced background bias.
Medical AI & Dermatology
Elsevier journal submission
Elsevier, 2026
Neil Voore, YRI Fellow
A 9th grader at Mountain House High School who wanted to apply AI to real medical problems and publish his findings.
Published research in medical image analysis with region-of-interest based bias mitigation.
“The best part of YRI Fellowship was the positive attitude of the mentors, as they guided me to finish my research paper. The YRI Fellowship fosters a positive and inclusive environment where participants feel respected and supported. I was encouraged to share ideas and collaborate, which helps build strong connections and teamwork skills. Mentors are approachable and provide guidance that boosts confidence and growth. Overall, the program creates a space where individuals can learn, improve, and succeed together.”
Elsevier journal submission in medical AI as a 9th grader
Showed that standard random splits leak between training and test sets, inflating metrics
Demonstrated that ROI-based training improves external F1-score and reduces background bias
Found evaluation design affects reported performance more than model architecture
Ayaan Rustagi, IEEE published and 3rd place science fair