Can Human Language Technologies Help Reduce Health Disparities?

The US Healthcare system is fraught with disparities and inequities caused by the differences in who gets access to healthcare and the kinds of treatments and quality of care they receive. While machine learning and human language technologies hold tremendous potential to optimize healthcare processes and practices, models trained on biased data increase the risk of perpetuating and amplifying existing health disparities. These issues may be exacerbated by Large Language Models (LLMs) which are known to encode harmful social biases and stereotypes with implications for algorithmic fairness.

This seminar, presented by Silvio Amir, Ph.D., Assistant Professor of Computer Science at Northeastern University, discusses the opportunities and challenges of using LLMs in healthcare contexts with a particular emphasis on the impacts on health disparities.