BioInvestGPT simulated clinical trials predict Novartis drug failure

Novartis Trial Failure Triggers Multi-Billion Market Loss As AI Simulators Gain Traction

Artificial intelligence platforms are challenging traditional pharmaceutical development after accurately predicting the late-stage failure of Novartis’ experimental muscular dystrophy drug, del-desiran, an outcome that erased $30 billion in market value when shares fell 11% last month, Reuters reported. The drug had previously drawn peak annual sales predictions of $5 billion from the Swiss company’s CEO. Copenhagen-based AI startup BioinvestGPT ran a simulated clinical trial in July using DNA sequencing to match virtual patient eligibility criteria, correctly predicting an insignificant clinical benefit before trial results were released.

Global Biopharmaceutical Spending And The High Failure Rate Of Clinical Trials

The global biopharmaceutical industry spends approximately $140 billion annually on human clinical testing, yet regulators approve only about 12% of tested drug candidates, a success rate that has remained largely stagnant for decades. Traditional clinical pathways require small Phase 1 safety studies, mid-stage Phase 2 trials, and large Phase 3 trials that take years to complete. In contrast, AI simulation platforms process virtual patient cohorts in a month or less to help drugmakers decide which programs justify capital investment and which acquisition targets possess real clinical value.

Investment in AI drug discovery more than doubled to $8.4 billion in 2025 compared to 2023, according to a McKinsey report. While current spending primarily targets molecule design rather than lengthy trials, pharmaceutical companies are increasingly using modeling tools to evaluate candidates early. US health regulators recently announced initiatives aimed at accelerating drug trials, which a federal health official notes could eventually establish a formal pathway for predictive AI in clinical development.

BioinvestGPT Track Record Across High-Profile Pharmaceutical Trials

BioinvestGPT shared detailed outcome analyses with Reuters ahead of multiple major clinical readouts, achieving a correct prediction record in five out of six tracked trials. Beyond forecasting the negative outcome for Novartis’ del-desiran, the AI platform correctly anticipated success for the Moderna and Merck melanoma vaccine, weak clinical benefit for AstraZeneca and Ionis’ heart drug Wainua, a failed trial for Novo Nordisk’s heart drug ziltivekimab, and success for Vaxcyte’s pneumococcal vaccine.

The platform’s methodology involves building virtual bodies through DNA sequencing to match specific trial parameters and testing models of the investigational therapies. However, the technology is not infallible. BioinvestGPT predicted positive results for Novartis’ pelacarsen, which lowers lipoprotein(a) cholesterol levels. In September, Novartis reported that the late-stage trial failed to cut the risk of major heart attacks or strokes for patients with a genetic risk factor. Bragi Lovetrue, who co-founded BioinvestGPT in 2024 with his wife Idonae Lovetrue, explained that a subsequent investigation revealed the model got the mechanism wrong by failing to account for genetically set variations in lipoprotein size.

BioinvestGPT Predicts Failure for Biogen and Takeda Trials

For trial results expected before the end of the year, BioinvestGPT forecasts failure for two Phase 3 trials of Biogen’s litifilimab in the most common type of lupus, along with Phase 2 studies of Takeda’s zasocitinib in Crohn’s disease and ulcerative colitis. The AI model indicates both drugs are suboptimal for those specific patient populations.

Industry executives express caution regarding artificial intelligence predictions. Takeda research chief Andy Plump stated he has immense confidence in the mechanism and machine learning used to polish molecule structures, noting he does not believe the industry is near a point where AI can make definitive predictions. Diana Gallagher, Biogen’s head of clinical development for multiple sclerosis, immunology, and Alzheimer’s, noted that the company uses every available tool including AI, but emphasized that algorithms relying on historical data might lean toward predicting negative results for conditions with few approved biologic treatments like lupus. Meanwhile, QuantHealth chief medical officer Francisco Beca argues that as predictive accuracy improves, ethical questions will emerge regarding whether it remains acceptable to expose human patients to clinical trials that AI models indicate will likely fail.