Measuring ROI For Healthcare AI May Require A New Approach

While the last two years have seen significant progress and uptick in AI implementation, many organizations are contemplating the transition from small, limited pilots to large-scale implementations and AI enabled transformation programs. A big part of this transition is answering a key question: has value been captured from these tools sufficiently to justify increased investment, and if so, how should value from AI in healthcare settings continue to be measured?

This question of investment and value capture varies markedly across the industry. In July, UnitedHealth Group, one of the largest payor organizations in the world, announced that it would be investing nearly $1.5 billion in AI across its different business units, ranging from insurance and back-office automation functions to care delivery and technology operations. Competitors such as Humana and Centene are investing similarly, understanding that these initiatives will likely lead to long term productivity gains and operational success.

With regard to the return on investment (ROI), evidence is slowly emerging that there is significant value to be captured if these tools are incorporated correctly into existing workflows. In fact, a 2026 Productive/Edge report indicates that for every $1 dollar invested in AI tools, there is roughly $3.20 captured in return over 14 months; additionally, per the study, organizations were found to achieve nearly 147% ROI within 3 years, with 45% of organizations achieving measurable and positive ROI within just 12 months.

Not all studies have been so encouraging, however. These numbers are in stark contrast to many earlier reports which highlighted that AI investments frequently fail in returning value on investment. A key study released by MIT last year caused significant uproar across the industry after finding that nearly 95% of AI based pilots were failing to deliver measurable returns. Another landmark McKinsey study indicated that while nearly 8 in ten companies have invested in some type of Gen AI functions, the same amount reported no significant improvement or impact to their bottom line.

Therefore, the ROI capture has not been so clear.

However, this highlights a fundamental problem with novel innovation and technology adoption: ROI cannot be measured using conventional means. While the two opposing camps regarding the return on investment for AI may seemingly propose opposing views on the value of the technology, the reality is that they are just describing different measures of capturing value. The word “value” therefore, has to be incredibly well defined.

Take for example ambient scribing. Studies are increasingly showing that there is indeed significant upside to using this technology. A recent JAMA study found that “AI scribe adoption was associated with 13.4 fewer minutes of EHR time, 16.0 fewer minutes of documentation time, and 0.49 additional weekly visits delivered.” While the minutes saved may seem minimal, over the course of multiple physicians across a clinic, and multiple clinics across a health system, the numbers add up. Another study found that “in radiology, AI-based interpretation reduced the interpretation time for abnormal contrast-enhanced brain CT lesions by 11.23%, the interpretation time for lung lesions by 52.82%, and the analysis time for peripheral blood smears by 61%.”

While these numbers are not always jaw-dropping and may not add up to significant, tangible cost savings or increased visits/encounters completed, they are still important, and indicate that return on investment needs to be measured with specific context in mind. One such novel measurement must be a focus on physician satisfaction. Across 17-22 patients per day, a few minutes saved per patient encounter can mean a huge quality of life uplift for a physician. Although not long enough to “squeeze another patient in,” there are still significant benefits to this, given that healthcare attrition and burnout are at all time highs. Additionally, these tools have increasingly becoming table-stakes for physicians. As medical education programs and hospitals are increasingly adopting AI tools in their daily workflows, there will soon be a critical mass reached where these tools will become a mainstay of “modern medical practice.” Systems that thereby do not adopt these tools may fall behind the curve, leading to poor talent attraction and challenges in maintaining a sustainable workforce.

It has been well documented that hospital and healthcare margins are razor-thin: “The median operating margin across U.S. health systems ran -0.1% in 2023, recovered to 1.6% in 2024, slid to 1.0% in 2025, and has fallen to a 0.4% year-to-date median as of March 2026.” Therefore, organizations cannot invest in multibillion dollar AI adoption programs without determining tangible ROI. However, as witnessed many times in decades past, when industry upending technologies arrive, traditional value models often do not conform with the realities of modernization. AI is undoubtedly one of the largest technological shifts in the history of humanity and in modern medicine. Therefore, organizations have to reform traditional and outdated financial and investment methods, and instead, must be incredibly strategic and thoughtful about what “value” truly means in this day and age.