How AI Will Help the World’s Top Hospital CEOs Transform Health Care

Leaders at some of the world’s best hospitals find themselves at a unique moment. Just a few years after the COVID-19 pandemic pushed global health care systems to a breaking point, many of these top hospital executives are now on the leading edge of what promises to be one of the biggest leaps in modern medicine.

Artificial intelligence is quickly showing the potential to revolutionize many aspects of care giving, from cancer diagnoses to clerical work. But AI also presents new risks, uncertainties and vexing ethical questions.

With the release of Newsweek‘s ranking of the “World’s Best Hospitals,” executives at leading hospitals around the world told us how they are using the power of AI.

“Health care needs to embrace artificial intelligence,” said Mayo Clinic President and CEO Dr. Gianrico Farrugia, who also serves on a National Academy of Medicine panel tasked with developing a code of conduct for appropriate health care use of AI.

Although the work on the guidelines will continue into the coming year, Farrugia said health leaders should not hesitate to adopt AI tools in the meantime.

“If health care were perfect, we could afford to wait,” Farrugia said. “Health care is not perfect, there’s too much pain and suffering.”

The COVID-19 pandemic laid bare many underlying weaknesses, including staff shortages, equipment and supply problems and systemic inefficiencies that both impede care and add cost. The CEOs told Newsweek that AI can help them better address many of those challenges.

“The hospital of the future is a hospital driven by data,” said Paulo Nigro, CEO of Hospital Sírio-Libanês in Sao Paulo, Brazil. “And I’ll tell you, the [COVID] crisis has accelerated this process of using data to make decisions.”

But hospital leaders also spoke of a tension between the promise AI holds and its challenges.

“We want to be doing the latest and greatest, but we want to be doing it in the safest and the most thoughtful way,” UCLA Hospital System CEO Johnese Spisso said.

Ultimately, the financial and demographic pressures hospitals face will likely require them to find the most useful and ethical ways to adopt the technology.

“Every health care system in the world has the same challenges, which are rising costs, an aging population and a decreasing workforce,” said Dr. Aymeric Lim, the CEO of National University Hospital in Singapore.

“The only way that we can get around this is actually using AI, so, we have no choice.”

AI Potential for Early Detection

AI is already widely used for record keeping, to improve note taking and documentation needed during patient consultations, and to enhance the security and integrity of patient data.

“There’s a lot that we’ve been using for several years before we even really called it AI,” Spisso said.

Those applications can cut costs and reduce the clerical burden on staff, freeing them to devote more time to patient care rather than paperwork. But it is AI’s ability to augment medicine, especially diagnostics, that these hospital leaders find most exciting and challenging.

It is already helping doctors to better evaluate CT scans and X-rays and identify early signs of many diseases, including breast cancer, pancreatic cancer and osteoporosis. The Mayo Clinic has developed ways to use AI to identify early signs of heart failure, and at the Cleveland Clinic, it is helping to address one of the most common causes of death among emergency hospital patients: the bloodstream infection sepsis, when the body does not respond in the right way to fight an infection.

“Successful treatment of sepsis lies in early detection,” Cleveland Clinic CEO Dr. Tom Mihaljevic said, but the telltale collection of symptoms indicating sepsis can be very subtle.

“The use of artificial intelligence helps us to summarize all the relevant data and helps us identify patients with the potential for sepsis much, much earlier,” he said.

Mihaljevic said the hospital has seen a 40 percent improvement in sepsis treatment using AI-powered algorithms.

“That is a phenomenal success for an otherwise deadly condition,” he said.

In Brazil, Nigro said that during the pandemic his hospital saw an opportunity to apply AI more broadly to population-level analyses of health trends to support preventative care.

“This is where I believe artificial intelligence will make a huge impact,” Nigro said.
Sírio-Libanês now has part of its business that uses algorithms to look for indications of health risks in the demographics of workforces at large companies.

“For instance, if it’s a mining company, probably employees will be more exposed to certain sorts of diseases,” Nigro said.

The hospital then recommends interventions or health initiatives targeted to the specific traits and trends the AI identifies among the employees.

“They can anticipate or even eradicate certain diseases,” he said. Nigro added the program already includes about 200,000 Brazilians and his “moonshot” goal is a million.

Nigro took an unconventional path to his position. His educational background is not in medicine but mechanical engineering, and he led companies in food services and pharmaceuticals before joining the hospital.

“My passion was always innovation,” he said.

Nigro said data-driven innovation can promote medicine that keeps people healthy, rather than just treating disease, and could realize enormous savings across health care systems. An ounce of data could be worth a pound of cure.

Expanding With Remote Care

AI is also playing a role in the expansion of remote care that has become more widespread since COVID-19.

Spisso recalled the difficult pandemic period as Los Angeles County suffered through four surges of widespread infections. COVID-19 patients filled beds and people in need of other care stayed away for fear of infection.

“During the COVID-19 pandemic we went from like 1,000 telemedicine visits a week to tens of thousands a week,” she said.

The growth in telemedicine was an emergency measure but it is proving its value well after the pandemic has passed. Many hospitals have kept and built upon the expanded telemedicine and remote care options that were borne out of necessity during lockdowns.

“That’s really here to stay,” Spisso said. “It’s allowed us to continue to expand without adding more bricks and mortar.”

Farrugia said ideas for remote care or telehealth were “accelerated by the pandemic and now they’re becoming part of the fabric of Mayo Clinic.”

One example is the Mayo Clinic’s 2023 launch of “Cancer Care Beyond Walls,” in which intravenous chemotherapy is delivered in the home rather than requiring a hospital visit.

Telemedicine and in-home treatment can improve the patient’s experience and reduce costs, these leaders said, and several hospitals find that AI assistance makes remote care better.

At the Cleveland Clinic, Mihaljevic said, AI is helping physicians answer routine patient questions that come via email.

Nigro said that Hospital Sírio-Libanês has been able to resolve 90 percent of cases via telemedicine in the first consultation with patients due in part to assistance from AI. That’s especially important for the more rural eastern parts of Brazil the hospital serves.

The digital helper allows physicians to draw on the experience of many previous consultations and more thoroughly and quickly check symptoms that might indicate the probability of, say, Zika virus or dengue fever.

“This is the future,” Nigro said.

The Ethics of AI

Not everyone is convinced of the speed with which AI will change medicine.

“I’m myself sometimes skeptical about the AI,” Dr. Yasuhiro Kodera, director at Nagoya University Hospital in Japan, said. Kodera is a surgeon and professor who has been in leadership positions at the hospital for eight years, five of them as director, and is retiring from the position this spring. In his long career he’s seen many new medical technologies arrive and, with them, new challenges in how best to apply those technologies.

Kodera said he sees the potential benefits of AI. His hospital is participating in some AI projects, including large-scale research of an AI system to use chest X-ray images to predict risk of osteoporosis. But his experience with some other AI applications leaves him leery.

“We have some discussions about how much we should rely on the AI,” he said. Kodera is concerned about whether the information that feeds AI programs will be complete or robust enough to produce reliable results. He pointed to the problem of AI “hallucinations” in some popular large language models such as ChatGPT as a cautionary tale.

“They sometimes lie,” he said of the popular AI tools. “They’re not always telling the truth, so there is some skepticism.”

Spisso said some complicated questions are already arising from use of rather simple AI tools intended to relieve some of the clerical burden on caregivers. For example, an ambient listening system uses AI to assist physicians with gathering and organizing the notes from their conversations with patients.

“Well, who else is listening to that information, right? Where is that data?” Spisso asked, giving voice to the potential concerns that patients might have. Spisso started her health career in nursing, and she said her approach to hospital management is rooted in the importance of the patient relationship.

“We don’t ever want to do anything that would invade the trust of our patients and the confidentiality of their information,” Spisso said.

AI also carries the risk of magnifying underlying biases in medical information if tools are poorly designed or rely on flawed data. A 2019 study found that a frequently used algorithm that guides decisions about which patients need extra care had used basic data that reflected racial bias. As a result, the algorithm reduced the number of Black patients identified for extra care by more than half.

Kenneth Goodman directs the Institute for Bioethics and Health Policy at the University of Miami and has written extensively on the ethics of AI in health care. He also contributed to recent World Health Organization guidelines on AI health applications.

Goodman said most ethical quandaries in medicine over the past 50 years have been triggered by new technology such as mechanical support for end-of-life care and gene therapy. AI follows that pattern but, he said, there is one major difference about this technology.

“AI is helping us think better, and so it’s a different organ that these machines are now helping us with,” he said. “And that is going to be transformational.”

Goodman said there are a few key questions hospital leaders must consider. Are they convinced of the machine’s reliability and accuracy? Do patients understand how and why the technology is used? And—perhaps most crucially—what is the appropriate role of the human health worker in overseeing the process?

“There must always be a physician in the loop,” regarding decisions based on AI tools, Goodman said. But exactly where that point of human control should come could well be a moving target with different and quickly evolving AI applications.

Goodman said health care officials must continually vet automated systems for reliability and be on guard against biased outcomes. Along the way they must also work to maintain patient trust by being open with the public.

“Turn your cards face up,” Goodman advised.

Keeping the Human Touch

Mihaljevic said the AI debate reminds him of his work as a heart surgeon, where his specialty was robotic surgery.

“One of the most common questions that patients would ask me was, ‘Who’s actually doing the operation, is it you or is it a robot?'” Mihaljevic said. “So that overlay between automation as well as the human touch is always there.”

He and other hospital leaders Newsweek spoke with shared some of the many ways they are working to address the AI challenges that Goodman outlined.

“With respect to health care, AI is a quantum leap,” Lim said. His hospital has established a new office of innovation to help the entire staff become AI literate.

“Not just a whole bunch of people using chatbots all the time but being innovative and creative in the delivery of health care,” he said.

Some of the AI challenges relate to how hospitals had to respond to the COVID-19 pandemic, when misinformation and conspiracy theories threatened to undermine public trust in health advice and the safety of vaccines.

During the early stages of the pandemic, Mihaljevic said, Cleveland personnel were open about what they did and did not yet know about the illness.

“The same principles apply to the implementation of new technologies as well, the principles of truthfulness,” he said.

Spisso said she leaned heavily on her hospital’s partners across the city to build channels of trusted communication during the pandemic and the COVID vaccine rollout. That sometimes included some unorthodox health communication partners, such as L.A.’s storied sports franchises.

“We had a panel of L.A. Laker legends, Kareem Abdul-Jabbar, a lot of the famous legends that are trusted in the community to help us share information on vaccine safety,” she said.

With AI, Spisso is once again looking to a team effort to guide decisions and maintain public confidence in new technology. UCLA Health has set up an advisory committee on AI and Spisso said she will soon hire an executive-level position on AI health.

Farrugia drew a similar parallel between the challenge of misinformation during the pandemic and the need to maintain public trust as hospitals adopt AI.

“It was a challenge then, it’s a challenge now,” Farrugia said. “Trust is at the center of what patients expect.”

The Mayo Clinic is a founding partner of the Coalition for Health AI, CHAI, which started in 2022 with assistance and input from corporate and academic partners including Microsoft, Duke and Johns Hopkins Universities, and federal government agencies. The CHAI goal is to develop standards for AI use, evaluate new technologies and increase trust among AI users and patients.

“Who better than health care to make sure AI is used properly in health care?” Farrugia asked. “So, we’ve got to step up.”

Adopting Rules for AI

Voluntary efforts like CHAI could help establish guidelines and principles for AI use but several hospital CEOs Newsweek spoke with agreed that AI will also require regulation.

“Regulation is something that is way behind,” Nigro said, particularly on the questions around the extent to which physicians can base decisions on AI. He said regulation would help firmly establish the principle of human control of AI in the health setting.

“The final decision will always be the decision of our physician,” Nigro said. “This is absolutely paramount for us.”

Lim said the massive amount of patient data involved in AI will require strong protection.

“You can just imagine insurance companies or a eugenics-type government using that at a massive scale, so there are obviously risks there,” Lim said.

Farrugia said he not only foresees regulation, but he is also advocating for it. A clear regulatory framework can do more than just deter harmful application of new technology, he said. Properly done, regulation can also help set a path to more innovation.

“Regulation is not simply to prevent bad actors from doing bad things but to give confidence to good people to do the right thing,” he said.

Farrugia encouraged his colleagues in health care to be “skeptical about AI, but don’t be cynical.”

Yes, AI carries risks. But, in Farrugia’s view, the biggest risk is inaction.

“Risk number one is that health care doesn’t embrace AI, and doesn’t deliver what we need to deliver, which is better health care for all,” he said.

Alex Backus and Lauren Giella contributed reporting for this story.