Seminar Series: Implementation and Evaluation of AI in Real-World Clinical Settings

November's AI seminar will feature Finale Doshi-Velez, PhD, and her presentation on "Interpretability and Human+AI Interaction for AI Decision Support in Health" .

In many health contexts, we think of the core question as what treatment or intervention to recommend to each patient (or perhaps, relatedly, determining the diagnosis). However, there's a "micro" problem hidden inside this "macro" problem: how do we maximize the efficacy of the human+AI team? The machine learning algorithms need to expose appropriate information about themselves such that the user can confirm alignment and understand the strengths and limitations of the AI. The interaction needs to be designed such that the user does not fall into their own cognitive biases such as over-relying on the machine advice. I'll discuss how we can design models that are both performant and inspectable, as well as steps toward personalizing the interaction to the user's specific cognitive needs. I'll also speak more broadly to the rich design space and opportunities when it comes to creating human+AI teams to help people achieve their health and wellness goals. 

About Finale Doshi-Velez, PhD, MSc

Finale Doshi-Velez is a Herchel Smith Professor in Computer Science at the Harvard Paulson School of Engineering and Applied Sciences.  She completed her MSc from the University of Cambridge as a Marshall Scholar, her PhD from MIT, and her postdoc at Harvard Medical School.  Her interests lie at the intersection of machine learning, healthcare, and interpretability.

Please email Stephanie Chuc at [email protected] for invite details.