For COVID-19 vaccine updates, please review our information guide. For patient eligibility and scheduling availability, please visit

NIH Grant Funds Research to Identify High-Risk Suicide Patients Using AI

Weill Cornell Medicine received a four-year research grant (R01MH119177) for $2.84 million from the National Institutes of Health to identify people at high risk of self-harm, suicide attempt and suicidal death using large-scale data from electronic health records and insurance claims.

Photo of Jyotishman Pathak

Dr. Jyotishman Pathak

Suicide is among the leading causes of death in the United States. While self-harm and suicide attempts are considered consistent risk factors for suicide, predicting and monitoring for them can be challenging.

The project is led by principal investigator Dr. Jyotishman Pathak, Frances & John L. Loeb Professor of Medical Informatics and Psychiatry, chief of the Division of Health Informatics, and vice-chair of the Department of Healthcare Policy & Research at Weill Cornell Medicine. Using de-identified electronic health records and insurance claims data from over ten million patients in New York, Dr. Pathak and his team will develop novel natural language processing and machine learning models to identify high-risk patients.

Risk tiering and risk predicting algorithms will be the focus of the first half of the project. “We are not just looking at clinical risk factors," explained Dr. Pathak. "You can have a history of anxiety and depression, but there are social and interpersonal aspects that might trigger someone to take their life." Unlike clinical information, however, social determinants of health data, such as housing status, family situation, and income, are not as well captured in electronic health record systems for every patient. Such data is highly temporal and subject to change over time. "Part of the project will be developing methods that will take these issues into consideration," said Dr. Pathak.  "We are going to develop more advanced AI-based algorithms, including deep learning methods, that will extract social determinants of health data from patient electronic health records in a more robust way."

Photo of

Dr. Gregory Simon

The second half of the project is a validation study, in partnership with Kaiser Permanente Washington Health Research Institute (KPWHRI), to see how the algorithms initially developed and tested using data from New York will perform in a different data set. Compared to New York, Washington is more rural and has a more homogeneous population in terms of race and ethnicity. Led by Dr. Gregory Simon, a staff psychiatrist and mental health services researcher, KPWHRI is also the lead site for the NIH-funded Mental Health Research Network. “We know that simple data from health records can identify many people at risk for self-harm.  This new research will use more detailed records and information about real-life risk factors to identify people we have been missing,” said Dr. Simon.

Researchers hope that by improving the process for identifying at-risk individuals, more appropriate interventions could be designed in the future and make a difference in addressing suicidality in the U.S.

Population Health Sciences 402 E. 67th St. New York, NY 10065 Phone: (646) 962-8078