Why Health Care CIOs Can’t Delay on Embracing AI

Artificial Intelligence in Health Care
It is crucial for CIOs to lead organization-wide AI integration to improve efficiencies and remain competitive.

Artificial Intelligence (AI) has the potential to transform the health care landscape over the next few years, helping clinicians achieve better outcomes in patient care and enabling organizations to be more efficient. But many Chief Information Officers (CIOs) and other IT leaders today are still poised at the brink of change, having not yet adopted the latest AI capacities in a comprehensive way, according to Laura Craft, vice president and analyst at Gartner, a Connecticut-based global research and advisory company.

Craft is also on the faculty of the Harvard T.H. Chan School of Public Health’s Executive and Continuing Professional Education program, Leadership Strategies for Information Technology in Health Care.

Harvard’s program is geared to help health care CIOs fully understand the potential of AI and other emerging technologies, and to see how these tools fit in as part of their organization’s broader strategy. It also strives to empower participants to effectively partner with clinical and administrative staff to make the most of the latest AI capabilities in order to lead their organizations to success in this area, explains Mary Finlay, MBA, who serves as the Program Director. Finlay is also a lecturer in the Chan School’s Division of Policy Translation and Leadership Development at the Department of Health Policy and Management.

Both Finlay and Craft agree that over the next few years, CIOs must be up to speed on AI technologies or risk being left behind by the competition.

Exploring AI Uses in Health Care

Craft points out that under the broader health care umbrella‚ providers, insurers, and life sciences will find that AI holds similar opportunities—but how it applies can vary, depending on the needs and the setting.

“For health care providers specifically [including health care organizations, hospitals, and medical practices], AI can reduce medical errors, improve patient outcomes, and improve organizational efficiencies,” Craft says. In fact, when applied to their full potential, the latest AI technologies, such as machine learning, language processing, and predictive analytics, are taking medical care to new levels of accuracy.

CIOs must be up to speed on AI technologies or risk being left behind by the competition.

For instance, organizations are using predictive algorithms to improve cancer diagnosis rates; reviewing patterns to predict patients at risk for sepsis (a life-threatening infection); performing remote monitoring of intensive care patients’ vital signs; identifying patients at risk for serious diseases, complications, and medicine non-compliance; and automating time-intensive administrative tasks.

In order for AI to be effective, Finlay and Craft both point out that there are some important criteria that organizations must meet, such as using the right data points, and training staff to input results properly and then interpret them correctly. Further, AI needs to be positioned as part of the organization’s broader strategy.

“With all technology, you need to think about how it aligns with your overall agency strategy, such as with improving the patient experience and becoming more effective at managing costs. That alignment requires working with your clinical partners to bring it to fruition,” Finlay says.

Looking to the Future of AI in Health Care

While some organizations are ahead of the curve in using AI—particularly at larger academic medical centers—Finlay points out that many others are still struggling to find their footing in this area.

“CIOs need to learn as much as they possibly can about the opportunity and challenges of new technology, with the understanding that they may need to seek expertise outside of their departments and even outside of their organizations,” Finlay says. In fact, she points out that one of the big differences that sets AI apart from other technologies, such as Electronic Health Records (EHR), beyond the analysis component, is that the management of AI does not reside strictly under the IT department. Rather, it requires a strategic partnership with the clinical side of care. “CIOs have to be open to being more of a learner, rather than an expert, in this space,” she stresses. That’s where partnerships with vendors, and case studies from other organizations, can be helpful to guide CIOs efforts.

AI can reduce medical errors, improve patient outcomes, and improve organizational efficiencies.

Assessing Health Care CIOs’ Adoption of AI

“At Gartner, we recently did a survey of 50 CIOs at health care organizations,” Craft says. “I asked them where they currently are with AI, or when they plan to implement it.”  The findings reveal that only 11 percent of CIO respondents have currently implemented some type of AI capability (such as algorithmic medicine or AI for diagnostic imaging), but 50 percent of them have plans to implement AI over the next 24 months. This illustrates how rapidly the field will be evolving in the near future—and indicates the need for CIOs to get up to speed in this area.

Gaining Trust for AI

One stumbling block that needs to be addressed in using AI to improve outcomes and processes is getting people to trust the results.

“We are not saying AI is going to replace clinical thinking and diagnoses,” Craft says. Rather, it’s a tool to enhance care, helping organizations achieve value-added care. “We need to introduce this as a way to help doctors make a better diagnosis. It’s not a replacement for doctors, but a tool to help them do their job better,” she points out.

For example, a project using deep learning for breast cancer diagnoses conducted by a team from Harvard Medical School’s Beth Israel Deaconess Medical Center (BIDMC) found that the clinical pathology error is 3.9 percent, she says. With AI, the algorithm had a 2.5 percent error rate. But when AI and clinical diagnosis were combined, the error rate decreased to 0.5 percent. This makes a compelling case for how AI, when used in conjunction with traditional diagnostic criteria, can help clinicians get the best possible outcomes for the patient, she adds.

One of the big differences that sets AI apart from other technologies is that the management of AI does not reside strictly under the IT department. It requires a strategic partnership with the clinical side of care.

The Need for Oversight

 Despite the clear promise that exists in using machine learning and other AI capabilities to transform the health field, there is also a need for governance to ensure that the data is used and applied correctly.

“Gartner predicted that by 2020, the first medical malpractice case regarding AI will be heard,” Craft says. “Either an algorithm will not be used when it should be, or it will be used incorrectly.”

With so much at stake, she stresses that health care organizations need to develop governance on how to apply AI in the safest and most responsible way.

“There’s a real imperative in the industry now to adopt AI—but we still need strategies to make it safe and pervasive,” she explains, adding that the technology is moving faster than the regulatory and legal requirements.

Navigating the AI Landscape

 CIOs have a tall order to meet in order to help their organizations navigate the complex landscape in front of them.

“This is an emerging technology, and there is still a lot for CIOs to learn about what this means to their enterprises,” Finlay says. “The more we learn from others who are deploying this technology, so we can apply their lessons and take them to scale, the better off our organizations will be.”