Healthcare execs want to implement diagnostic AI—but providers still largely distrust it

The data: Nearly 72% of healthcare execs say they’d trust AI to support nonclinical, admin processes that take away time providers could be spending with patients, according to Optum’s fourth annual healthcare AI survey of 500 senior healthcare execs.

2 areas where AI for admin can have a big impact: High levels of trust in administrative AI could signal healthcare execs’ intention to invest—especially in the following two areas, which take away the most time from patient care.

1. Prior authorizations. Most treatments require prior authorizations to ensure they’ll be covered under a patient’s insurance—but obtaining this greenlight is a drawn-out process that can take up to a month.

That’s why AI like Olive’s that electronically streamlines the prior authorization process could be attractive to healthcare execs:

  • Out of 1,000 providers, 90% reported prior authorizations had a “significant or somewhat negative” impact on their care delivery—including patients’ hospitalization or death, according to the American Hospital Association.

2. Clinical documentation. Providers say too many bureaucratic tasks contribute to their admin burden and take away from patient care. AI voice tech like Suki and Microsoft’s Nuance could reduce the time spent on clinical notes:

  • Suki claims its health system partners saved 4,375 administrative hours per week with its AI voice tech, for example.

More data: Nearly 40% of healthcare executives are excited about AI’s potential to improve diagnosis and predict outcomes, per Optum. Moreover, 36% of leaders are excited about the tech’s potential to improve medical imaging.

The bigger picture: Companies like GE Healthcare are already teaming up with large health systems to enhance medical imaging interpretation—a trend that’ll likely continue as healthcare execs reassess their AI strategies for 2022.

For example, New York-based Hospital for Special Surgery is leveraging GE Healthcare’s deep learning tech to “de-noise” raw digital data produced during an MRI to deliver a clearer signal—which helps generate images in a shorter period of time.

Why this could backfire: Although healthcare execs are excited about AI’s potential to improve medical imaging, providers may not trust it completely to inform their patient care.

  • About 95% of clinicians believe AI for diagnostic imaging is inconsistent or doesn’t work at all, according to a recent FDA study.

"Behind the Numbers" Podcast