If you are experiencing a medical emergency, please call 911 or seek care at an emergency room.
Artificial intelligence in healthcare is no longer science fiction. Today, it is science fact—with the power to significantly improve efficiency. However, concerns have been raised about whether or not AI is safe in the complex healthcare environment and whether it will introduce inherent biases. Research, which has been traditionally used to develop and test AI algorithms, should be part of the solution to address AI safety and equity.
Healthcare is embracing AI and machine learning in a big way: 86% of health companies use some type of AI. Research is part of the reason why—it plays a critical role in developing, testing, and implementing new technologies. Most healthcare AI applications are currently in non-clinical roles and as AI moves to the bedside, the accompanying concerns become more evident.
The excitement and anxiety surrounding healthcare AI reminds me of the early days of GPS navigation, which now comes standard in most cars.
The first time I got turn-by-turn GPS directions, I didn’t trust them. I’d been driving for decades; I knew the way to go! I turned right instead of left because I “knew better” than the technology and ran right into a traffic jam. The next time the GPS said to turn left, I complied, and now I, like most people these days, use apps like Waze which uses smart GPS to adjust to changing traffic and find the quickest route to my destination.
That’s where AI is headed in healthcare. MedStar Health Research Institute is at the forefront, blending real-world clinical experience with expertise in data science and research. We’re applying AI tools to make healthcare more accessible, provide real-time information to the clinical care team, and improve patient compliance and outcomes. We are also using AI to collect and analyze data safely to identify ways to make healthcare safer with improved results. And we’re exploring how these AI products can impact patient care while advancing health equity.
AI can improve patient health and address bias.
Creating groundbreaking AI algorithms requires the best data. If these software systems are built using inaccurate data, their analyses will be inaccurate. If they are programmed with diverse data that is not representative of the general population, the result will be biased.
Academic health systems like ours that develop AI must uphold a deep commitment to population health to ensure healthcare algorithms are accurate, are built to reflect our diverse patient community, and developed with safeguards to prevent biases.
Our novel AI tools:
- Use representative data from our diverse community. MedStar Health patients are urban and rural, academic and non-academic, and they represent a broad swath of racial and ethnic backgrounds. Our setting is ideal for generating data and training algorithms for application across the nation.
- Bring deep expertise to the scientific process. With dozens of data scientists in the MedStar Health system, we have the knowledge and experience to develop and test new algorithms in collaboration with clinical leaders.
- Keep equity in focus. The MedStar Center for Health Equity Research leverages partnerships with community organizations and the expertise of a diverse team of researchers and clinicians to help ensure new AI programs advance equity in care.
This unique combination of capabilities allows MedStar Health to deploy an impactful approach to AI research, development, and testing. Our researchers collaborate with clinicians and fellow scientists to create innovative approaches in AI that push the boundaries of our knowledge, driven by scientific rigor and guided by ethical safeguards.
For instance, we employ an AI co-pilot to assist with analyzing chest X-rays. Our algorithm identifies potential areas of concern in a patient’s imaging. This can help reduce the likelihood of subjective human interference like regret bias, which can lead a radiologist to overdiagnosis because of potential negative outcomes. The AI also serves as a double-check to further enhance the radiologist’s expertise.
We also use AI to improve echocardiogram visualization. An algorithm directs the clinician where to move the probe to capture the most useful images of the heart, getting novice caregivers up to speed quickly and improving efficiency across the board. The AI-assisted echo also reduces variability between human operators, helping eliminate biases.
- Analyze risk factors, early detection, and personalized interventions to reduce disparities in infant and maternal health
- Develop a personalized voice assistant for patients with heart failure
- Detect, diagnose, and classify breast and lung cancers
- Find solutions to nursing workforce challenges
- Identify trends in patient safety event report data
- Interpret coronary angiograms
On AI, MedStar Health is swinging for the fences.
Developing and testing algorithms in healthcare is a bit like baseball. Each time we research advances in AI we come up to bat.
Sometimes we strike out—the algorithm doesn’t work as intended, and we try again. Sometimes we bunt, advancing the field incrementally. But sometimes, we hit a home run. One game-changing algorithm can be incredibly impactful for millions of people, improving their health in ways we can’t yet imagine. At MedStar Health, we’re swinging for the fences.
The MedStar Center for Biostatistics, Informatics, and Data Science (CBIDS) was established to provide expertise in high-quality statistical methods, clinical informatics, data science, and health information technology development. We built CBIDS so our experts can come together, working to make sure the data we produce fits the needs of researchers and clinicians.
We’ve also created new leadership positions to help support our work in this area. Our Chief Research Information Officer, Aaron Zachary Hettinger, MD, is a board-certified in clinical informatics, and a federally funded investigator with expertise in research and data science. Our Chief, Research Data Science, Nawar M. Shara, PhD, works to coordinate our collaborative efforts between data, research, and clinical expertise.
With these experts and dozens more, we are on a journey to develop the novel approaches to AI in healthcare that could become game-changing innovations, like the GPS in your car. Together we’re leveraging the latest AI and machine learning technologies to help us move toward our shared destination: advancing community health and achieving health equity for everyone.