Drug development faces a forecasting problem, not a measurement problem. In the same way that temperature, pressure, and humidity readings have been collected for centuries, it wasn’t until we developed models and computational power that accurate weather forecasting became possible.

Similarly, in drug development, we’ve long had access to extensive pre-clinical data and measurement tools. However, without predictive models to interpret this data, we will continue to struggle to forecast a drug’s safety and effectiveness in humans.

Interesting coincidence that for the last several decades, the rate of success of a drug advancing from a phase 1 clinical trial to approval is about 10%, which was just about how good we were at forecasting tomorrow’s weather between 1900-1950.