Artificial intelligence (AI) and AI-enabled devices are already making their way into the market and should be able to help make meaning of data in order to improve care delivery, said speakers at the 15th Annual World Health Care Congress.
Already, artificial intelligence (AI) is starting to spread into healthcare with AI-enabled devices. During a session at the 15th Annual World Health Care Congress, Mark H. Michalski, MD, executive director of MGH & BWH Center for Clinical Data Science, a collaboration by Massachusetts General Hospital and Brigham and Women's Hospital, and Tom McGuinness, chief strategy and commercial officer at GE Healthcare, discussed how AI can be used to advance care.
Diagnostics, radiology, and pathology were the first parts of healthcare to really embrace AI technology, according to Michalski. The technology allows the healthcare industry to do the work faster and add value to the final report. He is also excited for what comes next that AI can assist with, such as finding novel biomarkers or building patient cohorts throughout the entire medical system.
McGuinness added that AI has the potential to make sense of the deluge of data in the healthcare system, which is slowing down decision making and making it a challenge to create a holistic 360-degree view of a patient.
“Big data and AI really go hand in glove,” he said. AI thrives in data-rich environments, and the more data you throw at an AI algorithm, the smarter it becomes.
“Why artificial intelligence is so useful is we’ve got this big boatload of data we don’t know what to do with,” but once you apply AI to it, you can start to suss out meaning from it all, Michalski agreed.
Eventually, AI will be one of the many tools at a physician’s disposal to create value. It will help to find disconnects and create efficiencies. He provided the example of looking at the number of errors taking place in radiography. Big data can allow someone to look at the errors in a retrospective analysis and identify the drivers afterward, but AI allows you to bring the information to the front at the time of decision making so clinicians can understand the errors, predict them, and prevent them.
The challenge with such novel technology is regulating it, because systems using AI can adapt and change over time. As the systems adapt and change, how will FDA apply standards that were meant for systems that were built to never change?
“How do those paradigms for regulation evolve?” asked Michalski.
McGuinness agreed. When things are regulated, they get tested and locked, he explained, but that’s not the nature of AI. “The inherent nature of artificial intelligence is that it moves forward.”
Both speakers also agreed that AI will never replace the clinical assessment of a provider after the moderator questioned whether the use of AI will lead to the loss of unspoken indications, such as when a patient comes in with bruises and says he or she fell, but the expression signs seen by the clinician indicate that there is abuse.
McGuinness stressed that AI will complement and augment the work of clinicians. Michalski added that once there is an algorithm to help make meaning of data, providers can determine if they’re losing something and figure out a different way to capture that additional information.
Finally, they spoke about how they think AI will impact healthcare in the short term. Michalski hopes that AI will change the way providers interact with data so that the most important data is highlighted for them to act on.
McGuinness added that some AI is already out in the market and impacting healthcare, but that it’s just in the beginning stages.
“As we demonstrate the value of these, there will be more proliferation of [AI] as a tool,” he said.