Health care data play an important role in shaping how doctors make decisions, how researchers develop new medical knowledge, and how health care systems improve patient care. But when that data contains biases, it can affect future research and clinical decisions, potentially leading to poorer outcomes for patients.
A recent publication by James J. Cimino, M.D., Distinguished Professor of Medicine and chair of the Department of Biomedical Informatics and Data Science, examines how biomedical informatics can help address these challenges and improve the way health care systems use data.
Titled “The Roles of Biomedical Informatics for an Equitable Learning Health System: Lessons from a Year of Scientific Informatics Literature,” the publication examines findings from 240 scientific papers reviewed as part of the American Medical Informatics Association's (AMIA) year-in-review presentation. Cimino and a team of trainee volunteers identified 19 major research topics, many of which highlighted connections between data bias, health care delivery, and patient outcomes.
Understanding how health care data affects patient care
A learning health system uses data from everyday health care to generate knowledge that can be applied to improve clinical practice. However, when that data reflects differences in how patients receive care, it can introduce bias into research and influence future clinical decisions.
“Health care differences produce biased data, which interfere with knowledge generation,” Cimino explained. “When we apply that knowledge back to care, we can worsen the differences as well as the resulting outcomes and their data, and the cycle continues all over again.”
The researchers found that many of the papers examined systematic bias in data and differences in health care delivery. Their findings highlight how biomedical informatics can help identify and correct these issues, leading to more accurate research and better patient care.
Bringing together informatics, artificial intelligence, and health care
Building an effective learning health system requires collaboration across several disciplines. Biomedical informatics, artificial intelligence (AI), data science, and clinical research each contribute to different parts of the process.
For example, clinical informatics and AI can support the development of clinical decision support tools that help providers make informed decisions. Clinical research informatics and machine learning can also help researchers analyze health care data to generate new medical knowledge.
“When we are ready to address care differences from the top down, we look to informatics to help with every part of the problem,” Cimino said.
Ultimately, the research highlights the importance of data quality in improving patient care. By identifying and correcting bias in health care data, researchers can generate better knowledge, support informed clinical decisions, and improve patient outcomes.
About the publication
“The Roles of Biomedical Informatics for an Equitable Learning Health System: Lessons from a Year of Scientific Informatics Literature” was co-authored by Humayera Islam, Pushkala Jayarman, Nick F.H. Reid, Pratishtha Guckhool, Malin Britt-Lalich, Parand Shams, Zihan Xu, and James J. Cimino.