Questions linger around AI's potential in healthcare

The expectation is that AI will revolutionize healthcare for patients and providers. It’s also left many experts hopeful that the technology will positive impact the human race in the near future. But before AI’s potential can translate into action, several key questions must first be addressed, a recently published viewpoint argued in JAMA.

In order for AI to be successful in healthcare, the technology needs to be applied to the right tasks. If it’s not, the industry may not ever realize its potential. AI is a tool that’s better deployed for some tasks than others, and it works at it’s best when primarily used to identify clinically useful patterns in large, high-dimensional data sets,
viewpoint author Thomas M. Maddox, MD, MSc., of the Washington University School of Medicine in St Louis, Missouri, et al. wrote in the viewpoint. 

Other tasks—like clinical risk prediction, diagnostics and therapeutics—are more challenging for AI technology.

AI also needs the right data. Though AI will most likely succeed when used with high-quality data sources, current clinical data from electronic health records (EHRs) is prone to bias and can be ill-defined and insufficient for effective exploitation by AI techniques.

Additionally, the writers noted the need to find the right evidence standard to demonstrate outcomes and consequences from AI, and how to effectively integrate the technology into clinical care.

“AI is a promising tool for healthcare, and efforts should continue to bring innovations such as AI to clinical care delivery. However, inconsistent data quality, limited evidence supporting the clinical efficacy of AI, and lack of clarity about the effective integration of AI into clinical workflow are significant issues that threaten its application,” Maddox et al. wrote.

“Whether AI will ultimately improve quality of care at reasonable cost remains an unanswered, but critical, question. Without the difficult work needed to address these issues, the medical community risks falling prey to the hype of AI and missing the realization of its potential.”