Artificial intelligence is already helping to analyse scans and support diagnostic decisions. But when does it actually improve patient care? Eric Topol, one of the world’s most prominent physician-scientists, argues that medical AI should be judged by clinical outcomes — and by whether it gives doctors more time for their patients.
Stories about medical technology often centre on a single inventor credited with a breakthrough and millions of lives saved. The reality is more complicated. Discoveries must pass through research, trials, implementation and evaluation in ordinary clinical settings. At every stage, the promise may prove greater than the demonstrated benefit.
That is why Eric Topol belongs in People & Leadership in 2026. He is not presenting a universal algorithm that has supposedly solved healthcare’s problems. As a practising cardiologist, researcher and founder and director of the Scripps Research Translational Institute, he asks a harder question: which technologies can make medicine more accurate and more humane — and how do we prove it? According to his Scripps Research profile, Topol has published more than 1,300 peer-reviewed papers and is among the most cited researchers in medicine.
Cardiology is a branch of medicine that deals with the disorders of the heart as well as some parts of the circulatory system.
From cardiology to digital medicine
Topol did not arrive in healthcare from a technology company. He previously led cardiovascular medicine at Cleveland Clinic, helped establish its medical school and remains clinically active as a cardiologist. That experience informs his view of AI: a digital system must help clinicians make decisions in real practice, where a doctor is responsible for an individual patient, has limited time and often works with incomplete information.
His influence extends beyond the United States. The UK government commissioned Topol to lead an independent review of how to prepare NHS staff to use artificial intelligence, genomics, robotics and other digital technologies. The resulting Topol Review was published in 2019. Its central message remains relevant: healthcare professionals need training and the conditions in which to develop new skills. Buying technology alone will not improve care.
Genomics is an interdisciplinary field of biology focusing on the structure, function, evolution, mapping, and editing of genomes.
His work has also received public recognition. Topol appeared in the inaugural TIME100 Health list in 2024, and Scripps Research reports that he was included among Best Leaders in the United States for 2025. These distinctions demonstrate his profile and influence; they do not establish an objective ranking of the world’s ‘most popular’ medical scientists.
Why write about him now?
In his 2026 essay The Future of AI-Facilitated Medicine, Topol examines three possible areas of progress: more accurate diagnosis, a stronger doctor–patient relationship and earlier prevention of disease. He also describes the obstacles: model errors, biased data, protection of medical information and insufficient evidence to support the widespread use of some systems.
The distinction matters. A technology that performs well on prepared sets of scans or clinical test questions does not necessarily improve outcomes in an everyday clinic. It must be evaluated with different patients, in different institutions and within the workflow doctors actually use. Topol points directly to the gap between medical AI research and its implementation in practice.
Some results already merit serious attention. In his essay, Topol discusses a large randomised trial of mammography screening. According to the results he cites, radiologists supported by AI detected approximately 30% more breast cancers and 25% more invasive tumours than radiologists in the comparison group. Those findings relate to a particular study and screening process. They cannot automatically be applied to every hospital or AI system.
Mammography is the process of using low-energy X-rays to examine the human breast for diagnosis and screening.
Research published in 2026 also shows that how clinicians and algorithms work together matters. In one randomised study involving 70 clinicians, diagnostic accuracy was 85% when a purpose-built AI system offered its opinion first and 82% when it acted as a second opinion, compared with 75% using conventional resources. Topol was not an author of this study. It is an example of developments in the field he examines, not evidence that all medical chatbots are effective.
Can AI give doctors more time with patients?
One of Topol’s most compelling ideas concerns human attention rather than diagnosis. If a system helps prepare clinical notes, identifies relevant details in a patient’s history and organises information, a doctor could, in principle, spend less of a consultation looking at a screen and more of it speaking with the patient. Topol developed this argument in his book Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again.
But the time saved will not necessarily go to patients. A clinic’s management could instead use higher productivity to schedule more appointments per shift. Topol explicitly raises this risk in his 2026 essay. Leadership in medical AI therefore involves more than selecting software. It requires a decision about how its benefits are measured and who receives the time it saves.
Predicting disease is not the same as preventing it
Another area of Topol’s work concerns identifying the risk of age-related diseases earlier by bringing together test results, medical histories and other information. His book Super Agers: An Evidence-Based Approach to Longevity appeared in 2025. In his 2026 essay, he considers how such data might help medicine move from detecting disease late to taking preventive action earlier.
Precision of language is essential here. A risk prediction is not a diagnosis. Detecting a disease early is not the same as preventing it from developing. An algorithm may indicate who needs closer attention, but its practical value depends on whether there is a proven action that can improve that person’s health. Topol explores the potential of large predictive models while acknowledging that some of the infrastructure he envisages does not yet exist.
Fairness matters as much as predictive accuracy. If training data mainly represents patients in well-resourced health systems, an algorithm may perform less well for other groups. If only major hospitals can access it, the technology could widen existing inequalities. Topol warns that, without deliberate efforts to ensure fairness, medical AI may increase disparities in care.
What makes Eric Topol a leader?
Topol combines three roles rarely found in one person: practising doctor, extensively cited researcher and public commentator on biomedical science. Through books, academic papers and his Ground Truths publication, he helps bring new findings into discussion beyond the laboratory. His influence does not need an invented ‘doctor of the year’ award, nor should research carried out by other teams be attributed to him. It lies in the questions he insists we ask before adopting a technology.
Does AI help a doctor notice something they might otherwise miss? Do patients benefit, beyond an improved score in a test? Does the system work for people of different ages and backgrounds? Is their information protected? And, once the tool is introduced, does the patient receive more of the doctor’s attention?
That is why Topol’s work deserves attention now. Medical AI is no longer solely a promise for the future, but it has not earned unconditional trust. Eric Topol makes the case for welcoming new capabilities while demanding evidence — with human health remaining the measure that matters.
We think you might also like these articles:Dr Olga Azarova: “Children Should Not Learn to Suffer — They Should Learn to Solve Problems”
Revolution in Education: How WESIO Is Transforming Schools
Which Countries Are Benefiting from Global Instability Today?