Mitigating Data Bias for Life Science Models
You downloaded the dataset and the bias came with it. Unless you catch it, it flows straight into your model, your paper, and the decision at the end.
This 60-minute session is for life-sciences researchers and managers who use public and private datasets and know the history of those datasets. We work through a practical arc you can run from detecting the bias you inherited, mitigating what is fixable in the data you already have, and designing collection so the next dataset does not repeat it.
We ground it in real, recent cases like a polygenic score or a skewed biobank. You will leave with a repeatable four-question scan for any dataset, a working sense of which fixes rescue biased data and which do not, and a short list of design habits.
No coding required. Bring a dataset you actually work with.
Who this is for: Post-docs, research scientists, PIs, and anyone in life sciences who collects data.
Hosted by:
