Where AI can have the biggest impact in healthcare—and where it can’t

The U.K. is taking on a big pilot program with 500,000 people being remotely monitored at home using AI to analyze all the incoming data. The program by the National Health Service underscores where AI is likely to have the biggest impact in healthcare––non-consumption, or areas where there isn’t an affordable or convenient solution for consumers.

Authors Moni Miyashita and Michael Brady, both of consulting firm Innosight, laid out three lessons in using AI to address healthcare challenges that can actually be solved with these new technologies in the Harvard Business Review. Innosight was founded by Harvard Business School Professor Clay Christensen. The authors also addressed why non-consumption is the best application of AI in “the new world of patient-centric healthcare.”

In one particular area, medical imaging, AI isn’t likely to make that much of an impact. That’s because the area of diagnosis from medical images is complex, completed by highly specialized experts in the field who are already receiving help at the margins.

But when it comes to smaller––especially the personal––aspects of healthcare, AI’s impact can matter leaps and bounds. Taking prescriptions properly and on time, following an exercise and diet regimen and reducing stress can all be aided by AI technologies in the home and vastly improve health as well as potentially prevent adverse health events.

While monitoring patients at home isn’t exactly new, AI is set up perfectly to manage large populations and monitor large volumes of data in these programs in health systems, the authors wrote.

AI may in fact be the answer to scaling up these types of programs, and providers can draw on three lessons for success:

Target critical metrics. Taking aim at impacting one critical metric, such as hospital readmissions, can improve both patient outcomes and financial sustainability, the authors wrote. For example, an AI tool in an Atlanta-based hospital reduced readmission rates 31% through system alerts identifying high-risk patients, saving $4 million over a two-year period.

Find partners. “Don’t try to do everything alone,” the authors wrote, encouraging alliances with partners with similar goals. Recent blockchain initiatives across healthcare stakeholders are good examples of different partners coming together to solve a shared problem.

Collaborate. Instead of seeing AI as competition with highly-trained professionals, it should be viewed as a collaborative technology. AI can help deepen knowledge and reasoning and quell patient fears of being treated solely by algorithms when working in conjunction with professionals.

AI can have a big impact on non-consumption leveraging these lessons and bring together healthcare providers and patients, the authors concluded.