Reproduce “minoHealth.ai: A Clinical Evaluation Of Deep Learning Systems For the Diagnosis of Pleural Effusion and Cardiomegaly In Ghana, Vietnam and the United States of America” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Darlington Akogo, Benjamin Dabo Sarkodie, Issah Abubakari Samori, Bashiru Babatunde Jimah, Dorothea Akosua Anim, Yaw Boateng Mensah
Deep Learning systems developed by minoHealth.ai were clinically evaluated for diagnosing pleural effusion and cardiomegaly from chest X-rays, demonstrating diagnostic accuracy comparable to or superior to experienced human radiologists across datasets from Ghana, Vietnam, and the USA. The highest-performing AI model exceeded the top radiologist's AUC-ROC score by approximately 10% for both conditions, and the models maintained high performance on a larger, regional Ghanaian dataset.
Key insights
- The minoHealth.ai Deep Learning systems achieved diagnostic performance for both conditions that was comparable to or superior to individual experienced human radiologists and an aggregated radiologist benchmark.
- AI models demonstrated robust performance across diverse geographical datasets (Ghana, USA, Vietnam) and different gender demographics, suggesting strong generalizability.
- The observed variability in human radiologist diagnoses highlights the potential for AI to reduce inter-rater discrepancies and introduce greater objectivity in medical image interpretation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.