The evolving COVID-19 pandemic is forcing healthcare systems to find new ways to meet the needs of affected patients with greater speed, agility and efficiency than ever before. While advanced cutting-edge data analytics are available with the potential to provide the tools healthcare professionals need for quick answers to pressing medical questions, they are dependent on large-scale access to real-world data. Hospitals and healthcare systems are instrumental in contributing to the body of data needed to realize their full potential of this health data revolution.
Due to advances in machine learning, we can now analyze millions of medical cases in mere minutes. That speed was unthinkable five years ago. Hospital systems across the country, including MedStar Health where I work, are starting to responsibly share sizable repositories of de-identified patient data, which is the fuel powering this faster, more representative body of medical research. We are positioned to ensure that aggregated, de-identified patient data is used in the public interest to find quick and accurate answers to the most pressing medical questions.
Currently, 20 health systems across the country are pooling their resources to apply the information drawn from interactions with doctors, laboratories and medical devices. But we need more hospitals involved to further unlock the promise of generating better insights to improve patient care and find faster cures. The advantages of putting together large, generalizable datasets are endless. Here are three to keep in mind.
First, we now have the ability to find trends and connect information from seemingly disconnected cases by applying advanced analytics, artificial intelligence (AI) and machine learning. Connecting those crucial dots provides greater diagnostic accuracy or new insights into how a treatment or clinical practice performs in the real world. For example, a recent study showed that ultrasound images of the heart assessed by AI could predict mortality in COVID-19 patients even when the same image interpreted by a human medical expert could not.
Second, scalable data platforms offer clinicians and researchers the advantage of speed. If we have enough combined data at our fingertips, we can shorten the time it takes to solve emerging and pressing medical questions as they unfold. Consider the pause of a COVID-19 vaccine rollout earlier this year due to concerns about a rare side effect. With the right dataset in place across several healthcare systems, we could have analyzed the de-identified medical records of all those who had received the vaccination at that point in less than a day and identified possible causes in a quick, efficient manner.
Finally, large datasets allow us to include de-identified data of patients from a diverse array of communities, geographies and races. Clinical trials are notorious for not including enough participants from under-represented communities. Medical research from real-world datasets is much more likely to effectively represent our communities at a large, national scale than ever before. Research from data from across diverse community can accelerate our understanding on how social determinants may impact health.
Researchers like me are genuinely excited by the advances in health data science and the role academic health systems can have to improve the health for everyone. Machine learning has transformed the focus of our research from "how do we get the data we want?" to "what questions can we ask of this data today?" That's why MedStar recently joined with other healthcare systems as members of Truveta, a company that is helping bring this heath data revolution to life. The healthcare systems involved are overseeing the appropriate and ethical use of the data.
Data in healthcare is too fragmented, incomplete or disparate across disconnected systems due to limits of how electronic health records are shared and work together. As healthcare professionals, we understand the public is concerned with how their personal medical data is handled. However, healthcare systems have a long history and strong track record of being responsible stewards of this data, bolstered by laws that incur substantial punishments if violated. We are uniquely placed to ensure that patient data is protected and used only for legitimate academic purposes.
Together, we can bring this approach to implementing data science to not only optimize our approach to addressing our patients' current needs but also to reshape the future of medicine. Healthcare leaders should start to have conversations with their chief technology officers and data scientists and look into the possibilities of how their systems can partner with other organizations to unlock the promise of this evolving health data revolution.