To Meet Future Needs, Health Care Leaders Must Look at the Data (Science)

Clinician Looks At Digital Technology Data of Patient
Medical technology concept. Electronic medical record.

As a health care leader or clinician, do you play an active role in your organization’s data science strategy? If not, you may miss out on important opportunities to help improve value, workflow, and patient outcomes.

Today, health care organizations have access to more information than ever before—from emergency department records to radiology reports to cases of surgical errors and complications. Yet the data alone isn’t enough to transform your operations. The data needs to be formatted, stored, and organized in a way that is meaningful and actionable for the organization. Only then can the tools of Artificial Intelligence (AI) and machine learning be used to make sense of all the information that exists and truly inform your efforts.

One of the best ways to do this is by forming a data science team (or joining one that exists), explains Heather Mattie, PhD, a lecturer on biostatistics and co-director of the Health Data Science master’s program at the Harvard T. H. Chan School of Public Health’s AI for Health Care: Concepts and Applications, which is designed to help people on the clinical and administration side of care get up to speed on AI and its potential at their institution.

Understanding the Potential of AI

In layman’s terms, AI is a form of technology that analyzes data and uses algorithms (or a series of steps) to identify patterns. Most people have experienced AI while browsing online. Think of when Google shows you ads targeted to your interests, or when Facebook identifies a person by their photo. In health care, these and many other advanced capabilities can go a step further to help clinicians and administrators make educated decisions on everything from increasing diagnostic and treatment capabilities to using supplies and manpower more efficiently.

However, according to Mattie, the logistics of how AI informs medicine can be difficult to grasp for people who haven’t been exposed to or trained in data science. Further, data scientists are limited in what they can accomplish without having input from medical experts.

“Data scientists are often lacking the deeper health care knowledge required to provide the context needed to capture and analyze the data they collect in a meaningful way,” says Trishan Panch, MD, MPH, a primary care physician who co-founded the digital health company Wellframe and who also joins Mattie as co-director of the Applied Artificial Intelligence for Health Care program.

The Need for Diverse Data Science Teams

Mattie and Panch both agree that the best way to realize the full potential of data science in the health care sector is by having leaders, clinicians, and data scientists work collaboratively in multi-disciplinary data science teams so they can pool their knowledge, experiences, and perspectives.

“For optimal outcomes, you need to have a data science team that includes as much expertise as possible,” Mattie stresses. “For example, the team needs to involve a leader who is good at understanding both the clinical process and the organizational context of health care organizations, and data scientists who are familiar with the data and what can (and cannot) be done,” she says.

Having diverse racial and cultural representation on your data science team is also essential to prevent perpetuating stereotypes by not looking at an inclusive picture when it comes to collecting and analyzing data.

While the value of having a data science team is clear, getting busy health care professionals and administrators to participate can be challenging in some organizations. But Gopal Kotecha, MD, MS, who also shares the title of co-director of Harvard’s Applied Artificial Intelligence for Health Care course, points out that the benefits people can expect in return for getting involved may provide strong motivation. “If you build data science solutions that directly improve the workflows of key stakeholders and educate them along the way, people will be invested to participate. By focusing on implementation and buy-in early, data science teams can save time and money, so everyone wins,” Kotecha stresses.

Finding the Best Health Care Data Science Solution for Your Needs

Kotecha also points out that how your data science team operates typically will depend on the size of your organization and the resources at hand, as well as the extent of your goals. For instance, you may find your organization has challenges with scheduling. Can you improve appointment times using AI? Or your radiology team may be unhappy with missed diagnoses. Can you find software to help radiologists hone their accuracy? Or the marketing team needs to reach more dermatology patients. How can you let more people know your services exist? Where does AI fit in? These are the kinds of questions that leaders need to be able to answer, Mattie chimes in.

Panch adds that in many organizations, it will come down to one of the following possible scenarios on how to operationalize your efforts:

  • Hospitals can develop their own AI solutions and involve clinicians to advise the creation and make sure they will achieve their goals.
  • They can partner with established vendors or big tech companies to use pre-existing solutions.
  • They can look to other organizations that have an effective data science strategy and use the products and lessons learned to guide their own efforts.

He says that for smaller hospitals and organizations, partnering with vendors or with other health systems is usually the most feasible approach rather than trying to start from scratch. But each organization needs to determine what will work best for their situation.

The Need for Human Expertise to Guide AI Efforts

Finally, Panch offers reassurance for people who are concerned that AI will take away jobs from people working in health care. “Health care is full of complex problems. Data can help to solve them, but the information contained in the data is usually hidden. Therefore, we will always need experts working in a team to find and apply the insights. You can’t build an automated solution to improve diagnoses unless you know what steps are required and what data you need; and this requires everyone to work together,” he stresses.


Harvard T.H. Chan School of Public Health offers AI for Health Care: Concepts and Applications, Innovation with AI in Health Care, and Implementing Health Care AI into Clinical Practice, online programs that helps medical and technology professionals improve health outcomes using AI.