BioinvestGPT called Novartis's $30B trial flop weeks ahead
What's the deal? Copenhagen AI startup BioinvestGPTDealroom has a profile for this one. Try Dealroom → predicted the failure of Novartis's late-stage muscular dystrophy trial weeks before the result wiped $30 billion off the Swiss drugmaker's market value. In July, the company ran a simulated trial using virtual patients and forecast that del-desiran would offer insignificant clinical benefit.
What happened: The late-stage trial missed its primary objective last month, and Novartis shares fell 11%. The outcome diverged sharply from the chief executive officer's earlier projection of peak annual sales reaching $5 billion.
The track record: BioinvestGPT has correctly called five of six outcomes it shared with Reuters in July. Those included del-desiran's failure, a ModernaDealroom has a profile for this one. Try Dealroom →-Merck melanoma vaccine's success, weak benefit for AstraZeneca and Ionis's heart drug Wainua, the initial failure of Novo Nordisk's ziltivekimab, and success for Vaxcyte's pneumococcal vaccine.
How it works: The platform uses DNA sequencing to model a human body matching a trial's eligibility criteria, then runs a virtual trial with a model of the test drug. "We can pinpoint the reason why a drug is effective and safe, and in many cases, why not," said co-founder Bragi Lovetrue, who launched the firm with his wife Idonae Lovetrue in 2024.
Why now? The global biopharmaceutical industry spends roughly $140 billion a year on human trials, yet only about 12% of candidate drugs win approval — a rate stuck for decades. AI firms say simulations, which can take a month or less versus years, help identify which programs to advance and which acquisition targets to pursue.
What could go wrong? The simulations are not infallible. The platform wrongly forecast success for Novartis's pelacarsen, which failed a late-stage heart trial in September. A review found "we got the mechanism wrong," by failing to account for genetic variations in lipoprotein size, Lovetrue said.
The signal: BioinvestGPT already has predictions on the table, including failure for two Phase 3 trials of Biogen's lupus drug litifilimab and Takeda's zasocitinib in Crohn's disease and ulcerative colitis. If more hold, virtual trials could reshape how drugmakers decide where to spend their billions.
Read more: independent.co.uk (Reuters)
Image credit: Centers for Disease Control and Prevention
