Executives are bullish on the potential of artificial intelligence to improve healthcare. But they say adoption is not happening quickly enough due to a lack of workforce training, high costs, and privacy risks, according to a survey by audit, tax, and advisory services firm KPMG.
KPMG's survey of healthcare leaders was part of a larger study of how executives across five industries view the future of AI in their sectors, and the steps they are taking to maximize its benefits and mitigate its challenges.
"The pace with which hospital systems have adopted AI and automation programs has dramatically increased since 2017. Virtually all major healthcare providers are moving ahead with pilots or programs in these areas. The medical literature is showing support of AI’s power as a tool to help clinicians," Melissa Edwards, managing director, digital enablement, KPMG, said in the report.
An overwhelming majority of healthcare respondents (89%) think AI is already creating efficiencies in their systems, and 91% believe it is increasing patient access to care.
Many of the AI-related services and solutions being advanced in healthcare today are largely in the clinical, patient-facing space.
"Basic forms of automation are proving to be the ‘gateway drug’ to advanced forms of AI—such as scanning documents to determine the urgency of a referral. Applying AI to make earlier diagnoses of critical illnesses is a key area," Edwards said.
- Nine out of 10 healthcare executives are confident that AI will improve the patient experience with the greatest impacts being found on diagnostics, electronic records management and incorporating robotics into tasks.
- More than two-thirds of healthcare stakeholders (68%) are confident AI will eventually be effective in diagnosing patient illnesses and conditions, and close to half (47%) believe that diagnostics will have a significant impact soon—within the next two years.
- Healthcare executives also anticipate gains in process automation, with 40% seeing X-rays and CT scans being handled robotically.
Recent findings indicate that function may be close to reality. Google Health reported that an AI model developed and deployed by its DeepMind subsidiary was more effective in screening patients for breast cancer than human doctors using recent X-rays only, despite having access to patients’ previous records.
But the pace of progress is too slow, according to one-third of executives, citing barriers such as a lack of workforce talent and the high cost of implementing AI tools.
To date, only 44% of healthcare insiders say their employees are prepared for AI adoption, which is substantially lower than some of the other industries surveyed. Less than half of healthcare organizations (47%) offer AI training courses to employees.
Just 67% of healthcare insiders say their employees support AI adoption, the lowest ranking of any industry, according to KPMG.
Many healthcare institutions lack a breadth of individuals who “speak” the language of AI, Edwards said.
"Comprehending the full range of AI technology, and how best to apply it in a healthcare setting, is a learned skill that grows out of pilots and tests. Building an AI-ready workforce requires a wholesale change in the approach to training and how to acquire talent. Having people who understand how AI can solve big, complex problems is critical," she said.
Health systems have already made significant capital investments to meet electronic health records (EHR) requirements. To get AI off the ground requires even more of an investment, and, as a result, some health systems are slower to allocate full funding for AI.
More than half of executives (54%) believe that AI to date has actually increased rather than decreased the overall cost of healthcare. Decision-makers are struggling to determine where to place their AI best bets.
"The question is, ‘Where do I put my AI efforts to get the greatest gain for the business?' Trying to assess what ROI will look like is a very relevant point as they embark on their AI journey," Edwards said.
Healthcare executives also are concerned that AI could threaten the security and privacy of patient data. Relatedly, 86% say their organizations are taking care to protect patient privacy as it implements AI.