Several prominent health insurance companies have begun to leverage artificial intelligence in the process of approving or rejecting claims, raising crucial questions about the need for human oversight and the potential for the technology to perpetuate existing inequalities within the healthcare system.
Jude Odu, a seasoned healthcare technology expert with 25 years of experience, emphasizes the shift towards automation that AI has brought. "In contemporary settings, decisions are increasingly driven by AI," he explains. "Previously, a human had to review a claim thoroughly before a denial, but now that process has increasingly been delegated to AI systems."
Odu's background includes significant time at one of the largest health insurance companies in the U.S. He is also the founder of Health Cost IQ, where he employs AI to pinpoint inefficiencies in health plans sponsored by government and employers. Furthermore, he recently published a book discussing the impact of AI on health plans.
In an interview with WUSF’s Gabriella Paul, Odu highlighted the dangers of allowing AI to make medical coverage decisions without human input.
Reflecting on his early career at United Healthcare, Odu recalls, "I worked in the appeals and denials department, and while I wasn't directly on the committee that made those decisions, I observed that nurses and medical directors would quickly assess cases before making a denial. This experience gave me insight into the inner workings of our healthcare system."
Because Odu has experience in international healthcare systems, he was able to provide a broader context for the U.S. approach. "Healthcare operates differently in many developed countries," he notes. "For instance, individuals don't face bankruptcy due to medical debt, which starkly contrasts with the situation in the U.S. where healthcare serves as a major profit-driven enterprise."
He illustrates this point by explaining how denial practices can be beneficial for large shareholder companies—less payout in claims translates to higher profits. Introducing AI into this equation complicates matters, particularly in light of existing health disparities. "AI's effectiveness is dependent on the quality of the data it’s trained on and the design of its algorithms," he cautions. "In cases of health disparities, AI can inadvertently amplify discrimination because it learns from existing biases."
Odu presents some compelling examples, including United Health Group's acquisition of NaVi Health for $2.5 billion, where an AI algorithm is under scrutiny for producing a high reverse rate during patient appeals. He also notes that another AI system tasked with scheduling appointments resulted in longer wait times for Black patients, exacerbating disparities based on socioeconomic factors used in its algorithms.
While he acknowledges the pitfalls of AI, Odu believes there's potential for its beneficial application in health insurance. "Implementing AI systems wisely is key. It's vital to actively seek out biases within these systems," he states. For instance, healthcare organizations could employ AI to audit all claims, identifying discriminatory trends related to specific demographics or ZIP codes.
He argues that AI can be a proactive tool in closing care gaps for at-risk populations when trained with inclusive datasets meant to improve overall health outcomes. "The focus on AI in healthcare should not just be efficiency, but enhancing patient health," he emphasizes, noting that efficiency gains will naturally occur, but what's efficient for the insurer might not be for the patient.
Odu also touches on political discussions around AI in healthcare, hinting at government-backed plans potentially utilizing AI for decision-making. "AI is already being scaled by the Centers for Medicare and Medicaid Services, and it's likely to penetrate Medicare and Medicaid services soon," he predicts. He stresses the importance of establishing safeguards around AI, as its capabilities could lead to unintended consequences if not managed wisely.
"While AI has tremendous potential," he concludes, "it’s crucial to ensure it channels its power ethically and effectively."
