Musings
Essays on real-world evidence, healthcare AI, research methods, and leadership.
Treatment Failure Is an Endpoint, Not a Code
In real-world evidence, treatment failure and non-response are endpoint contracts. The analyst has to define the adequate trial, the failure components, the source of each signal, and the intercurrent events before the data can mean anything.
Read the full noteEarlier writing
1 entry
Target Trial Emulation, Explained for Operators
Most observational studies fail in the design, not the analysis. Target trial emulation is the protocol habit that prevents it: specify the randomized trial you would run if you could, then emulate every element - eligibility, time zero, strategies, and contrast - explicitly in real-world data.