Artificial Intelligence in Health Care Is Just What the Doctor Ordered

IN HEALTHCARE, THE battle cry never changes: Improve quality. Reduce cost. Make the doctor-patient experience better.

Artificial intelligence may finally help solve those problems and so many others. One study by Accenture estimated $150 billion in annual savings from key clinical health AI applications by 2026. With the increase in data collected from health records, to the increased use of wearable fitness trackers, to environmental factors, AI can help shift the balance from reactive health care to continuous health and wellness.

AI is already being applied behind the scenes to detect drug interactions and predict which patients are most at risk for readmissions or experiencing a hospital-acquired condition. But the promise of AI is much greater – think robot-assisted surgeries and virtual nursing assistants – and that promise could be realized if more healthcare organizations are willing to adopt AI.

In a recent survey of leaders in healthcare organizations conducted by Intel, we learned that only 37 percent of them use any form of AI today. Surprisingly, clinical-use cases for AI such as predicting patient risk and analyzing medical images are the most common; 77 percent of organizations that use AI do so for clinical-use cases, compared to 41 percent for operational-use cases and 26 percent for financial-use cases.

In our survey, trust was presented as the biggest obstacle to implementing AI in healthcare. More than one-third of those surveyed say patients won’t trust AI to play an active role in their health care, and 30 percent assumed that clinicians won’t trust it either.

During a closed-door workshop at HLTH in May, top experts from across the industry weighed in on ways to do overcome the trust issue. Among their suggestions: Address the “black box” perception, lean in to areas where clinicians are ready for change, highlight the benefits to practitioners and consumers alike, and create a more nimble and flexible framework that would allow for easier adoption of AI.

Fatal error was named as the No. 1 reason for fear of AI in health care. Providers, regardless of level of education, have invested significant education and training on understanding the processes and procedures to treat patients, and are accountable for describing exactly what steps were taken when something goes wrong. The technology industry needs to focus on building accountability, transparency and interpretability of how AI applications are developed and deployed to help clinicians feel more comfortable with the new technology.

But the patient-provider experience must come first. Ensuring providers can understand how the technology works so they can provide their point of view to patients should be paramount, and then organizations can lean into areas where clinicians are ready for change and where fatal error is less likely.

A recent American Medical Association survey of physicians concurs. The AMA states that “physicians are most receptive to digital health tools they believe can be integrated smoothly into their current practice, will improve care, and will enhance patient-physician relationships.” Instead of clinical moonshots, organizations should focus on incremental advances in supporting administrative work that can help with clinician burnout and even provide predictive analytics for intervention while slowly building overall trust in the technology.

The benefits to using AI are numerous and should be emphasized among organizations and consumer. When implemented correctly, automation can increase operational efficiency, improve quality, reduce cost and will help build solid ROIs that can be invested in more complex AI use cases. By using AI to reduce administrative burden and take away mundane tasks, providers can spend more time with the patient, focus on more complex cases and operate at the top of their licensure. AI will augment, not replace, providers, and will provide actionable data that can help create more meaningful patient-provider interactions.

It’s worth noting that there is an opportunity cost of not embracing AI: Eighty-four percent of those surveyed by Intel believe that companies that do not invest in AI will fall behind competitors. In the current market, it stands to be a differentiator.

To enact these recommendations, a more flexible and nimble regulatory framework will be required, and the Trump administrationseems to agree. AI isn’t one size fits all, and the regulation can’t be either. The AMA in June announced that they would promote the development of thoughtfully designed, high-quality, clinically validated AI in healthcare, which will be extremely helpful but the technology industry also needs to continue to take an active role in these conversations, showing regulators and provider groups the benefits of AI technology and working to address feedback and concerns.

At the end of the day, we are all consumers of healthcare, and we should feel confident that advances in technology can ensure we receive high-quality, affordable care. Together, we can ensure patients and providers realize the benefits of AI in healthcare today, building trust and understanding that it will help us unlock incredible advances in the future.