- MIT’s Andrew Lo is known for studying Wall Street, but he’s turned his attention to biotech.
- He wants to predict which drugs will fail clinical trials so money and patients’ time can be saved.
- But experts say his tech needs more work and that getting it wrong may cause harm.
Between October 2019 and March 2020, about 300 data scientists from Novartis embarked on a special project.
The $194 billion drug company teamed up with a 15-person consulting firm in Cambridge, Massachusetts, to test whether a computer program could predict whether a drug would make it through the gauntlet of clinical trials and regulatory hearings necessary for it to end up on pharmacy shelves.
Such a tool is the drug industry’s holy grail, Iya Khalil, the global head of Novartis’ AI Innovation Center, told Insider. It takes nearly $1 billion to develop a new drug, and more often than not, it’s a lost cause. A study from three companies focused on the industry, Bio, Biomedtracker, and Amplion, found that just 9.6% of drugs made it through in-human tests.