The Science, Art of Entrepreneurship, Artificial Intelligence

Friday, Nov. 30, 2018

What motivates venture capitalists and health care CEOs to invest in artificial intelligence (AI) startup companies is no longer the hype of AI; it's the potential that they see in the wide ranging implications for AI across all of medicine, according to a Thursday session, "Entrepreneurship and Artificial Intelligence."



"AI is both a science and a clinical art," said Ajay Kohli, MD, radiology resident at Drexel University College of Medicine, Philadelphia. "While AI will definitely turbo-charge the science and research of medicine in the coming years, the primary operational target for most companies looking to work or invest in AI is geared toward the clinical art side, which is the type of AI technology that can aid radiologists on a daily basis."

Companies working on bringing AI into medical imaging should note that the barriers are three-fold: solve the problem, obtain medical data and implement clinical AI algorithms within health systems.

"Identifying and trying to solve the real problems facing the field of radiology, such as slow workflow and long wait times for patients, is where AI companies should be focusing their efforts," Dr. Kohli said.

Concerns regarding the security of sensitive patient data prevent many institutions from sharing their data with developers, and that lack of data is slowing the process of building effective algorithms. However, Dr. Kohli believes that AI startups need to focus on finding novel data sets as well as developing strategic partnerships with clinical institutions to help interpret this data.

Since health care is complex at many different levels and there is a lot of red tape affecting the implementation of new technology, Dr. Kohli also recommended that radiologic AI startups focus on improving diagnostic accuracy and workflow.

"Building AI products is not the problem. Implementing the algorithms is," Dr. Kohli said. "AI technology in imaging analysis should not affect the infrastructure of hospitals or physician practices. If anything it should improve their workflow structures and potentially help them increase their revenue, but this is only if these algorithms are implemented judiciously."

Given the complexity of AI technology and the gravity of implementing this type of technology in the health care setting, Dr. Kohli recommended that AI companies work with a team of specialists — computer software engineers, data scientists and clinicians — to launch a successful startup.

Radiologists with experience partnering with imaging centers, hospitals or insurance companies, can be especially valuable to startup companies. Including the input of department or institutional leaders is also an important way to ensure a modicum of success.

"There is a need for radiologists to pursue a more 'intrapreneurial' role or working on special ideas within a larger company or health care system," Dr. Kohli said. "Physician intrapreneurs can help health care systems navigate the growing climate of health tech vendors to ensure that physicians are applying products in the most efficient workflow for the hospital."