Many healthcare purchasers are concerned about partnering with organizationsthat do not fully appreciate the need to lock down patient data.(Image: Shutterstock)

Note from health care executives to developers of artificialintelligence-driven health products: You just don't understand us.And until you do, we will adopt your products cautiously, if atall.

This message came through loud and clear from a recent survey commissioned by the global lawfirm Ropes & Gray in collaboration with a division of Crain'sNew York Business. The survey group itself wasn't large–284responded to the emailed questions. But it was a relevant sampling,with 85 percent of respondents hail from either the C-Suite, theboard of directors, or a decision-making management position withhealth care industry employers.

Related: Health care systems doubling down on digitaltools

Follow-up interviews with a smaller select group served tosupport the overall conclusions: health care purchasers worry aboutthe security of their data once it is shared withAI vendors, and they also suspect that most vendors have not takenthe time to truly understand the market they are attempting topenetrate.

Among the findings:

• 60 percent said health care's "strongly entrenched business andreimbursement models make it difficult to bring digital healthproducts to market." In other words, if the insurance company orpayor refuses to reimburse for the service, it won't get muchmarketplace traction.
• Nearly half do not believe the AI developers "fully understandthe health care market." These respondents want the productdevelopers to demonstrate health care expertise beyond the specificapplication of their product or service, and they want them toprove that they can maintain data security.
• Product/service pricing was another obstacle cited byrespondents. In an evolving field, purchasers are uncertain abouthow to evaluate these new offerings.

The report focused on factors that currently are viewed asobstacles to partnerships among players on both sides of thetransaction. Many AI product developers expect to fuel their growththrough partnerships with customers in the health care field. Butmany of those surveyed were concerned about partnering withorganizations that did not fully appreciate the need to lock downpatient data. Responses included:

• 70 percent worry that "a digital health partner would fail tosecure or encrypt data prior to it being shared."
• 34 percent fear an accidental data breach by the AIpartner
• 27 percent worry that an AI partner will fail to secure "clearpatient consent for the use of data."
• 21 percent cited a concern that the AI vendor "would sharepatient data without proper de-identification."

Still, 42 percent of respondents said it was "likely or somewhatlikely they would partner or contract with an AI company over thenext year." About a third said that would probably not happen, withanother one in five uncertain about whether such a partnership wasin the cards. What do those that are willing to partner with an AIvendor expect in order to move forward: Three primary qualities:evidence of efficacy, complete compliance with all relevantregulations, and complete compliance with all data security andprivacy standards.

"All sides now recognize the enormous potential at theintersection of these sectors," the report concluded. "Digital techis beginning to understand the importance to providers and pharmaof proving efficacy in a clinical setting, while providers are morewilling to embrace digital tech. But there are still culturaldifferences that need to be bridged, and significant roadblocksthat must be overcome."

Read more: 

• Health care data breaches highlight limits ofHIPAA's vendor oversight
• Digging deeper: Health data mining platforms surgeahead
• Walgreens launches digital health care providermarketplace