It seems like a simple question. “How do we know if a drug works?” But in the future this question will be answered differently. Bayesian statistics are surpassing frequentist statistics in the age of the internet and machine learning. Historically, the scientific gold standard has always been group-based evidence. A drug might work on some patients, but if it doesn’t work well enough across the group it’s considered a failed trial—and it’s not even made available to those who were benefiting. But biosensors and continuous monitoring can change the math. Genetic biomarkers have prepared us for this conceptual leap. “Efficacy” can be personalized. We can prove a drug’s benefits patient by patient, and stop the drug’s use whenever the benefits disappear.