Ai Tool Exceeds Physician Accuracy on Medical Licensing Exam in Buffalo, Ny

A clinical AI tool, Semantic Clinical Artificial Intelligence (SCAI), developed in Buffalo, New York, has demonstrated high accuracy on the United States Medical Licensing Exam (USMLE). Developed by University at Buffalo researchers, SCAI outperformed most physicians and other AI tools. The tool achieved a 95.2% score on Step 3 of the USMLE. SCAI contains 13 million medical facts and their interactions. It uses semantic triples to create semantic networks, enabling logical inferences. The AI tool can reason and converse, augmenting physician decision-making, according to Peter L. Elkin, MD, from the Jacobs School of Medicine and Biomedical Sciences at UB. SCAI's capabilities include improving patient safety and democratizing specialty care. While SCAI can access vast amounts of data, it is designed to augment, not replace, physicians. Elkin emphasizes that AI-assisted doctors may replace those who do not use AI.