How we predicted 50,000 patient reactions to a new drug

The challenge

A pharmaceutical company needed to understand how different patient profiles would react to a new drug before launching expensive clinical trials. Traditional methods took months and cost millions.

What we did

We built 50,000 digital patients with unique medical profiles, histories and preexisting conditions. Each one reacted to the drug based on its simulated biology. We observed adverse effects, adherence and dropout patterns.

Results

• —94% match with phase III trials six months later
• —3 risk subgroups identified that had not been considered
• —6 months saved in the pre-clinical research timeline
• —Estimated $4.2M saved in failed trial costs