Facebook Twitter LinkedIn Instagram Email Printer Google Plus
Drive Innovation December 13, 2018

MIT Healthcare Experts Talk Artificial Intelligence’s Untapped Potential

Digital Health Physician, Icahn School of Medicine at Mount Sinai
Key Takeaway
Learning models are outperforming human pathologists in metastatic cancer diagnosis.

Seeking to more closely examine where healthcare is regarding the latest and greatest artificial intelligence and machine learning advancements, many groundbreaking healthcare executives, professors, and leading researchers presented their industry perspectives at this year’s MIT Sloan Healthcare and BioInnovations Conference, held earlier this year in Boston.

The conference captured the essence of artificial intelligence-based disease diagnosis in various fields of medicine as shown by a couple of different startups depicted below, which have been met with variable success. Besides, the hot artificial intelligence robotics has now been extended to healthcare, proving its impact in patient engagement rates and outcomes delivery. The below startups drive home the need for solving U.S. healthcare’s $3 trillion conundrum with artificial intelligence-based technologies, a case in point for creating radical new solutions.

Here’s more from two of these individuals – both leaders of machine learning-focused companies working to transform the industry – on what they had to say about artificial intelligence’s untapped opportunities for healthcare.

Aditya Khosla, Co-Founder, PathAI

Aditya, whose company provides artificial intelligence-powered pathology diagnosis and works to decrease diagnosis error rates, said Deep Learning models are outperforming human pathologists in metastatic cancer diagnosis. Error rates of Deep Learning models (0.65 percent) versus human pathologists in clinical practice (13-26 percent) versus human pathologists on small tumors/micro-metastasis (23-42 percent), can significantly affect diagnosis outcomes, he said. Inter- and intra-observer discordance in commonly seen in some studies, he noted.

Aditya also quoted a few examples of other known healthcare artificial intelligence startups in disease diagnosis, such as Arterys for Cardiac MRI, Stanford’s skin cancer detection tool, ZebraMed for Liver CT, Crosoft InnerEye for Brain MRI, and Babylon Health’s chatbot diagnosis.

Cory Kidd, PhD, Founder and Chief Executive Officer, Catalia Health

Cory talked about the huge psychological impact of an actual robot looking into a person’s eyes, thus impacting engagement rates. His startup uses a wide-eyed, interactive health companion, named Mabu, to drive artificial intelligence-based conversations for chronic disease management and delivery of improved outcomes to providers and pharma.

(For a more detailed overview of Cory’s machine learning perspective, listen to our interview with him on the Oliver Wyman Health Podcast.)

Cory offered MIT attendees looking to partner with an artificial intelligence company the following pieces of advice:

  • Have good rationale for partnering, and remember why you choose to do so
  • Partner compatibly – have the same vision, culture, and personalities
  • Design the partnership for success, but also allow for escape if things don’t work out
  • Generously reward value creating contribution
  • Leverage the partnership by being prepared to delegate


Insights in your inbox