Focus: AI companies run virtual drug trials, aim to improve success of human studies

• BioinvestGPT correctly predicted Novartis' del-desiran failure in July, ahead of the actual trial results in September, along with five other accurate predictions including Moderna and Merck drug successes
• Investment in AI drug discovery more than doubled to $8.4 billion in 2025 from 2023, though spending focuses on molecule design rather than clinical trial bottlenecks
• The AI platform predicts failures for upcoming Biogen litifilimab Phase 3 lupus trials and Takeda zasocitinib Phase 2 inflammatory disease studies, though company executives express skepticism about AI's definitive predictive capabilities

Summary: AI Companies Deploy Virtual Drug Trials to Reduce Pharma Failure Rates

AI startups are increasingly partnering with pharmaceutical companies to run virtual clinical trials, aiming to improve the historically low 12% success rate for drug candidates seeking regulatory approval. The biopharmaceutical industry spends approximately $140 billion annually on human clinical testing.

Key Companies and Technologies:

BioinvestGPT, a Copenhagen-based AI startup founded in 2024, accurately predicted five out of six recent trial outcomes in July, including Novartis' failed del-desiran trial for muscular dystrophy (previously projected for $5 billion peak sales). The company also correctly forecasted results for trials from Moderna, Merck, AstraZeneca, and Novo Nordisk.

Tel Aviv-based QuantHealth is another platform using real-world data to simulate patient responses. BioinvestGPT predicts upcoming failures for Biogen's linabrothon in lupus Phase 3 trials and Takeda's zasocitinib in Crohn's disease and ulcerative colitis Phase 2 studies.

Market Impact:

Investment in AI drug discovery more than doubled to $8.4 billion in 2025 from 2023, according to McKinsey. The technology promises to reduce expensive trial failures, with AI simulations taking a month or less versus years for traditional human trials.

Industry Response:

Pharmaceutical executives remain cautious. Takeda's research chief Andy Plump stated AI isn't yet capable of "definitive predictions," while Biogen's Diana Gallagher noted AI algorithms may struggle with diseases having limited historical data, like lupus with only two approved biologics.

US health regulators recently announced initiatives to accelerate drug trials, potentially creating pathways for integrating predictive AI into clinical development. However, experts acknowledge human trials will remain necessary despite AI advancements.