Artificial Intelligence in Health Care

Portfolio Overview

The field of artificial intelligence is rapidly growing and evolving. Artificial intelligence (AI) refers to the ability of a machine to perform a task that is normally done by humans, including problem-solving and learning. AI-enabled software can be divided into two categories, with the classification depending on how the software is developed. Rules-based algorithms use expert-derived rules to turn the inputs into an output, through a defined and logical process. Data-based algorithms are given sets of labeled input data (called “training data”) and use programmed processes to derive relationships between the inputs and the labels. The relationships can then be used to predict how new input data would likely be labeled. While forms of clinical decision support (CDS) software has been available for many decades, recent advances in data-based AI may have the potential to significantly improve software performance, opening to the door to an explosion of new products, some of which are already in the market. With this explosion will come a host of regulatory, implementation, and adoption challenges in the near- and long-term. 

Current Work: 

• Trust, But Verify: Informational Challenges Surrounding AI-Enabled Clinical Decision Software
  • January 2020 | Public Event
  • September 2020 | White Paper Press Release
• Accountability, secrecy, and innovation in AI-enabled clinical decision software (November 2020 | Journal of Law and the Biosciences)
• AI-Enabled Clinical Decision Support Software: A “Trust and Value Checklist” for Clinicians (November-December 2020 | NEJM Catalyst)
• Accessing Real-World Data to Evaluate Postmarket Performance of AI-Enabled Clinical Decision and Diagnostic Support Software
• Understanding the Business Case for AI-Enabled Diagnostic Support Software

Past Work:

• Current State and Near-Term Priorities for AI-Enabled Diagnostic Support Software in Health Care