Two natural language processing (NLP) systems developed by Yizhao Ni, PhD recently finished as top performers in a national challenge hosted by Harvard Medical School and the Volgenau School of Engineering at George Mason University. The National NLP Clinical Challenges (n2c2) took place in November in San Francisco before the annual symposium of the American Medical Informatics Association (AMIA).

Image of Yizhao Ni, developer of natural language processing systems
Natural language processing systems developed by Yizhao Ni, PhD finished as top performers in a national challenge.

Ni, an assistant professor in the Division of Biomedical Informatics at Cincinnati Children’s Hospital Medical Center, is an expert in natural language processing (NLP), one of several types of artificial intelligence (AI) that are reshaping the ways we approach health care data. The technology is especially strong in training AI to identify relevant clinical information from unstructured data and free-text notes stored within electronic health records.

The first n2c2 task challenged developers to create systems to answer the question: “Can NLP systems use narrative medical records to identify which patients meet selection criteria for clinical trials?” It required NLP systems to automatically assess if a patient is eligible for a study by comparing each patient to a list of selection criteria. Such a system could both reduce the time it takes to recruit patients, and help remove bias from clinical trials.

Ni’s Automated Clinical Trial Eligibility Screener® system placed in a statistical tie with the top 5 out of 101 systems, submitted by 45 teams around the world.

The second task aimed to answer the question: “Can NLP systems automatically discover drug-to-adverse event relations in clinical narratives?” Systems were challenged to analyze approximately 500 discharge summaries drawn from a clinical care database in order to identify drug names, attributes (e.g., dosages, frequencies, durations), and their relationships with adverse drug events or reactions.

Ni’s end-to-end medication reconciliation system ranked 8th among 50 submissions.

Both systems won international recognition in earlier years.

Ni’s research interests lies in the development of machine learning, NLP and information retrieval techniques to assist clinical decision making. Working with Cincinnati Children’s Drew Barzman, MD, he has developed an NLP-based tool to help predict school violence. Other recent projects he’s worked on include safety algorithms to reduce medication errorsimproving clinical trial recruitment, and reducing cost by predicting and preventing surgery cancellations.

For more information on the systems, contact


Yizhao Ni’s Natural Language Processing Systems Among Top Performers at National Challenge