Isaac Kohane, Chair, Department of Biomedical Informatics, Harvard Medical School

Isaac Kohane chairs the Department of Biomedical Informatics at Harvard Medical School and practices pediatric endocrinology at Boston Children's Hospital. He came to medicine from a family with no doctors in it, left medical school midstream to study artificial intelligence, and has spent four decades trying to make computers useful at the bedside.

Finding Computing at Brown

He grew up in Switzerland, and no one in his family was a doctor. He came to the United States for college, majored in biology at Brown University, class of 1981, and used Brown's open curriculum to take a heavy load of computer science. His honors thesis, with population geneticist James Kidwell, became his first paper in the medical literature.

Leaving Medical School for Artificial Intelligence

Midway through his first year at Boston University School of Medicine he found that medicine was not the science he had expected, and he left to do a PhD in computer science during an early boom in AI research. His advisor at MIT, Peter Szolovits, sent him back. "Zak, you should finish your clinical training because I'm not getting a lot of respect from clinicians," Szolovits said. He earned his MD and PhD from Boston University in 1987.

Training as a Pediatric Endocrinologist

He completed pediatrics and then pediatric endocrinology at Boston Children's Hospital, where he has practiced since, and he started a computing lab there. He joined the Harvard Medical School faculty in 1992.

His lab grew into an institution. From 2005 to 2015 he directed the Countway Library and co-directed the Center for Biomedical Informatics, and in July 2015 that center became the Department of Biomedical Informatics, with Kohane as its inaugural chair and the Marion V. Nelson Professor.

Testing GPT-4 Before Its Release

In October 2022, before ChatGPT was public, Peter Lee at Microsoft gave him early access to GPT-4. He tested it on rare pediatric endocrine cases from his own training, including a newborn with ambiguous genitalia, and the model walked through the hormonal and genetic workup. He and Lee wrote The AI Revolution in Medicine: GPT-4 and Beyond (2023) with Carey Goldberg, and he is editor-in-chief of NEJM AI.

Diagnosing Children Without a Diagnosis

His current clinical focus is children whose disease has no name yet. At a 2024 lecture at Brown he walked through the case of a child who had stopped walking and talking. Specialists had missed the diagnosis, and a genome sequence plus an AI read of the child's history found it. The child was walking and talking within weeks of treatment.

Links

← All people · Featured in Voices of Healthcare: 30 People Changing Health Tech (2026)