Thanks for explaining Unlearn's approach, it is novel and robust. I have a clarifying question on calculating the control group sample size reduction enabled by digital twins.
Quote: "So the more correlated the forecast is with observed control outcomes, the fewer participants need to be assigned to the control group."
Question: If observed control outcomes are an input into the control group calculation, how can the sample size reduction (and associated cost and time savings) enabled by digital twins for a given clinical trial be calculated before the trial begins? Or is that something that is refined as the trail progresses (ex: start the trial with a target of 500 test and control populations => enroll people into both groups normally => track digital twin outcomes predicted and actual outcomes for the control group as the trial progresses => estimate bias between predicted and actual in parallel => use that info in ANCOVA to bring down the control group)?
Thanks for explaining Unlearn's approach, it is novel and robust. I have a clarifying question on calculating the control group sample size reduction enabled by digital twins.
Quote: "So the more correlated the forecast is with observed control outcomes, the fewer participants need to be assigned to the control group."
Question: If observed control outcomes are an input into the control group calculation, how can the sample size reduction (and associated cost and time savings) enabled by digital twins for a given clinical trial be calculated before the trial begins? Or is that something that is refined as the trail progresses (ex: start the trial with a target of 500 test and control populations => enroll people into both groups normally => track digital twin outcomes predicted and actual outcomes for the control group as the trial progresses => estimate bias between predicted and actual in parallel => use that info in ANCOVA to bring down the control group)?