• The challenge for Microsoft and Novo Nordisk will be to integrate AI and pharma into a process that will bring cheaper medicines faster with less risk.
• Besides Novo Nordisk, Microsoft entered into a similar partnership with UCB in 2021 as part of the COVID Moon Shot project.
Developing a new drug can take over 12 years and cost $2.6 billion or more. The process is as risky as a trip to Las Vegas as nine out of ten potential new therapies fail to become market-ready products. The cost of a failure is paid for by the company, but ramifications go beyond the bottom line and include loss of life and increased costs to healthcare systems. Even with several high-profile failures, the promise of AI represents a substantial opportunity to transform drug development.
The September 2022 announcement by Microsoft and Novo Nordisk has several components. Microsoft will provide computational services, including cloud and AI technology. Microsoft engineers will work directly with Novo Nordisk data scientists and domain experts to accelerate early research and development projects. The collaboration will first use automated summarization to analyze scientific literature, patents, reports, and discussion forums. The second project will develop models that can predict the risk of a person developing atherosclerosis, including establishing biomarkers. Biomarkers are measurable substances whose presence is indicative of some phenomenon, such as disease, infection, or environmental exposure. Novo Nordisk believes that building machine-learning models can accelerate the creation of new molecules that have specific properties to rapidly do in minutes which might take humans months to accomplish.
The path toward AI-developed drugs is accelerating; however, most pharma scientists believe that it will take a decade until the delivery of AI-influenced drugs follows a predictable cadence. Still, there is a growing list of AI-developed drugs that are entering preclinical trials, in addition to announcements from Microsoft and Novo Nordisk. This suggests that a wave of AI pharma will reach the market. Assuming that AI-pharma continues to develop what should companies do to make sure they are not caught flat-footed?
As with all technology, it is important to develop a vision for how your company will build out its AI-pharma implementation. Adding benchmarks to measure progress will be critical. Before investing large amounts of capital in platforms, build a proof concept to test out simple but essential algorithms. Finally, ignore the temptation to believe that AI will solve many problems of current new drug production. With the capability to build in-silico models, AI will create a healthier future. However, AI should always be seen as a guide rather than the sole arbiter when making new drugs and compounds.