With the potential to disrupt care as we know it, AI continues to surge in interest among different parties in the healthcare industry. But in all the excitement some stakeholders are concerned that investment in the technology are hurt by disorganized and inconsistent approaches and methods.

“Everyone’s investing but there’s a little chaos in the salad going on in terms of AI products,” said Ingrid Vasiliu-Feltes, chief quality and innovation officer for MEDNAX, in a panel discussion this week at the Ai4 Healthcare conference in New York City. “Everybody wants to do a use case, but there’s no synergy and clear understanding of what you’re aiming to achieve and how it’s going to impact the population health, or precision medicine or whatever product you want to build," she continued. "I’m a little frustrated sometimes we’re not doing it [investing] in the most meaningful way.”

More than 85 speakers from across the U.S. and abroad gathered for the two-day event to discuss the impact of AI on their respective healthcare segments, which ranged from payors and pharma developers to medical device and biotech firms.

Sitting in on technical and non-technical lectures, participants listened as panel after panel delved into current and potential applications for AI, such as insight extraction for medical imaging, pharmaceutical decision-making, digital pathology multiplex image analysis, and fighting physician burnout.

Like Vasiliu-Feltes, many speakers had much to say about the issues plaguing the development and effective deployment of AI, particularly how providers and other entities are choosing which AI and machine learning solutions to use in their practices.

“For any organization, it’s important for us to make sure we are looking at the benefits of how the AI is connecting to our patients and the decisions that we’re making are related to connecting to AI partners,” said Tanya Scott, clinical services director for Kaiser Permanente, who was a member of the same panel discussion as Vasiliu-Feltes, How AI is Revolutionizing the Hospital. “Is the decision one that is going to provide us with the affordability that we need and at the same time give us the patient outcomes we’re looking for? When our outcomes are such, that we’re able to drive the cost of healthcare down and affordability is realized as a result of that, then it makes sense to make the investment in AI.”

This and other conversations looked into other necessary elements for leveraging the full potential of AI, including the need for clinicians' input during development stages for technology; the process in which data is delivered to patients; the elimination of mundane tasks by AI for radiologists; and the future direction of where AI is expected to go.

One in particular titled, Data Management, Privacy and Security, delved into data extraction and management, and the reaches of cybersecurity and interoperability for ensuring safe and effective communication of data in a timely manner. The panelists in this discussion agreed that it is not so much challenges within the technology itself but regulations governing its actions that prevent users from leveraging the full potential it has to offer.

“With that increased level of regulation, it is more challenging to make sure you have consensus and that you are giving data subjects the right to view their data, to have their data deleted or to have updates,” said Jon Wilkinson, technology privacy officer and U.S. privacy officer and director at PAREXEL. “That obviously can run into conflict with AI where you want to put data in the data lake and you want to have those data repositories where you can use them for all different kinds of applications in analytics. The question is, do you have the right to do that with respect to those regulations? How are you managing those consents to ensure you are in the scope of the law?”

Many view AI as a positive addition to their work as providers and manufacturers in healthcare. That said, they are in agreement that such technology requires a uniform regulatory framework regarding how it is deployed, and how to measure its effectiveness so different entities can achieve the outcomes they want for their patients.