Beyond Cost, Patient Safety and Physician Burnout Should Factor Into AI Decisions in Health Care

Wide application of artificial intelligence (AI) in health care could greatly improve patient safety, reduce physician burnout, and improve the overall efficiency of health care from scheduling and billing to making surgeries more safe. But the health care industry lags in AI adoption due to inherent technical challenges and because the financial benefits of AI adoption are often difficult to measure, according to a new review published in The New England Journal of Medicine.

The study authors detailed the AI challenges confronting health care compared with other industries.

“Early AI took root in business sectors in which large amounts of structured, quantitative data were available and the computer algorithms, which are the heart of AI, could be trained on discrete outcomes—for example, a customer looked at a product and bought it or did not buy it,” they wrote. “Qualitative information, such as clinical notes and patients’ reports, are generally harder to interpret, and multifactorial outcomes associated with clinical decision-making make algorithm training more difficult.”

Another major factor slowing utilization of AI are the financial incentives, or lack thereof, that influence decisions at different health care organizations.

“In our experience, the environment in which some health care organizations operate often leads these organizations to focus on near-term financial results at the cost of investment in longer-term, innovative forms of technology such as AI,” they wrote. “Health care organizations that prioritize innovation link investment decisions to ‘total mission value,’ which includes both financial and nonfinancial factors such as quality improvement, patient safety, patient experience, clinician satisfaction, and increased access to care.”

Although AI has many potential uses, the authors wrote that utilization is uneven across health care, as some potential uses are harder in practice than others.

Nikhil R. Sahni, MBA, MPA-ID

“Specialties such as radiology, pathology, dermatology, and cardiology are already using AI in the process of image analysis,” they wrote. “In radiologic screening, for example, up to 30% of radiology practices that responded to a survey indicated that they had adopted AI by 2020, and another 20% of radiology practices indicated that they planned to begin using AI in the near future.”

Reimbursements

The authors cited processing of reimbursements as the area of health care where AI is most used.

“Uses of AI in this domain are both common and among the most advanced uses, with a higher-than-average total mission value,” they noted.

More than 10% of claims are denied or delayed because of eligibility issues, the authors wrote, but up to 85% of claim denials are avoidable. They cited as an example an unnamed, large health care system that ran claims data through a regression model to sift out correlations between submitted claims and denials.

The health care system now has a pilot program that uses the top 10 root causes to flag claims and address them before submission.

“This could further improve the denied-claims record of the health system and potentially reduce the administrative spending needed for claims processing and reprocessing, all while improving the patient’s experience by reducing the number of frustrating denials,” they wrote.

Clinical operations

Applications of AI to improve clinical operations is “an area of intense research,” but despite encouraging early results, widespread application is uncertain. One specific area the authors examined is the use of AI to optimize operating room capacity, which could reduce surgical backlogs and clinician shortages, they wrote.

Brandon Carrus, MSc

“Organizations such as the Mayo Clinic and Lucile Packard Children’s Hospital at Stanford have estimated that utilization would potentially be improved by 15% to 20% if AI were implemented,” the authors wrote. “However, this step remains largely in the pilot phase, and whether the improvements will be realized is not known.”

Quality and safety

Adverse drug events, decompensation, and diagnostic errors have been identified as problems with the greatest potential for improvement from AI, the authors wrote, but “a substantial portion of value comes from nonfinancial factors.” Consequently, they noted, “the current level of adoption of AI in this domain is limited.”

Nonetheless, AI applications for the early detection of sepsis are already believed to have saved the lives of thousands of patients, according to the authors. They noted an AI algorithm that uses electronic health record data in combination with blood pressure and heart rate measures to predict whether a given patient in the intensive care unit (ICU) might have sepsis.

“Over a period of 5 years beginning in 2014, this monitoring had reportedly saved approximately 8000 lives across the network of the health system,” they wrote.

Why is adoption of AI lagging in health care?

AI has immense potential to improve health care for patients, clinicians, and administrators, but the practical challenges are just as immense, the authors observed. Using AI to make a movie suggestion to a customer browsing Netflix is child’s play compared with the challenges of applying AI to improve health care.

There is also the question of incentives for developing the algorithms that would result in better care. The authors specifically cited the fee-for-service payment model that is the norm in US health care as an impediment to investing in AI applications.

“Another major reason (for slow AI adoption) is the fee-for-service model of payment as compared with a value-based payment model,” they wrote. “The latter payment structure would fund measures that improve care or make it safer, which is where the benefit of AI in health care delivery could be of substantial importance. Under a fee-for-service model, these incentives are substantially less prominent or absent altogether.”

To speed the adoption of AI in health care will require those who make investment decisions to give as much weight to improving patient safety and reducing clinician burnout as they do to dollars and cents.

“Historically, the decision to invest in AI has been based on financial return,” the authors wrote. “This calculation should be expanded to include nonfinancial factors as well. Otherwise, AI adoption could continue to lag in certain domains in which a large portion of its effect is nonfinancial, such as quality and safety.”

Reference

Sahni NR, Carrus B. Artificial intelligence in US health care delivery. N Engl J Med. 2023;389:348-358. doi:10.1056/NEJMra2204673