Hosted by Alicia Johnson, MPH

Clinical algorithms assist healthcare providers with decision-making based on a small set of demographic and clinical characteristics. While they can be useful tools, algorithms are created using data from sources such as electronic health records or longitudinal studies. They reflect the features present in that data – which may include a lack of representation of various subgroups of the population, selection bias that may differ among subgroups, systematic measurement error, and more. In clinical settings, these biases can manifest as results that may be inaccurate for certain groups of patients, potentially leading to misdiagnosis, missed opportunities for treatment, and other undesirable outcomes.

This talk provides an introduction to the concept of algorithmic bias, illustrating how it may show up through examples from various areas in medicine. The goal of this talk is to support attendees in understanding algorithmic bias, the limitations of clinical prediction rules, and the impact algorithmic bias may have on patients. We will also discuss how to begin to identify and minimize the impact of algorithmic bias.

This introductory seminar is appropriate for those who are familiar with clinical algorithms and are curious to learn about their limitations. Healthcare providers, analysts, students, and others are welcome.

Register via Compass. This seminar will be recorded.