The Speed of Success

With the impact of AI and ML on the global pharmaceutical industry, a future in which most drug candidates will be successful, cost savings will decrease prices, and innovation is accessible to everyone, could be on the horizon. 

The capital and time requirements compared to the success rates of R&D programs will need to improve (it still takes between 10 and 15 years to develop and gain regulatory approval for a new drug, as well as an average cost of more than $3 billion). So how can AI and ML help organizations bring treatments to patients faster while working more efficiently and cost-effectively? How do we rewrite long-established rules? We speak with Fabrice Chouraqui, CEO-Partner of Flagship Pioneering and Chief Executive Officer of Cellarity, about what it takes to push the boundaries of science.

What is the current failure rate for new drug candidates and how can AI/ML change this for the better?
 

It takes about 10 to 15 years and an average cost of more than $3 billion to develop and gain regulatory approval for a new drug, yet 90 percent of drugs that enter clinical development fail. This equates to a return on capital which is extremely low. To sustain the level of investment needed to continue to bring new therapies to patients, we need to improve. 

The pharmaceutical and biotech industries have contributed great innovations to society – effectively eradicating diseases such as tuberculosis and others that used to be the leading causes of death, developing medicines that effectively cure diseases such as HIV and AIDS, and saving humanity from the COVID pandemic. 

This is where AI and ML come in. These technologies don’t just help organizations work more efficiently and cost-effectively. If used in the right ways, AI and ML can enable us to rewrite the rules of drug discovery by seeing things in human biology that were previously unseen. 

Human biology is highly complex. If you think about human biology as a computer, it’s always running countless programs at once – and we can’t see many of those programs because we’re humans and they weren’t invented by a human. An AI/ML-enabled platform allows us to see and better understand these programs – and predict which compounds can cause a desired change in cell behavior to stop certain disease-causing programs from running, and prioritize compounds based on those predictions. Such a platform essentially allows us to reprogram human biology with the insights it generates. We can compare healthy and diseased cell types, and then predict what we need to do to the diseased cell types to get them from a state of disease to health. 

This whole-cell approach to drug discovery will create a much higher probability of success in the clinic. It also has enormous repeatability in terms of addressing almost any disease to create drugs that are out of reach with current methods of drug discovery. 

What part can human collaboration play in developing the tech that can unlock that potential?
 

AI-enabled drug discovery requires the right people behind the AI. It requires people with multiple disciplines who can speak the various “languages” of other disciplines to collaborate and uncover novel insights effectively. 

Through a whole-cell approach, the focus of drug discovery can shift from a traditional single molecular target to the underlying cellular dysfunction. We do this by linking biology and chemistry with high-dimensional omics data to uncover drugs that act against the cellular signature of the disease itself. We call this a non-reductionist approach, and our people – experts across biology, chemistry, computational research, data science, etc. – fuel this approach. This enables them to collaborate in a way that will lead to the development of novel treatments that I believe will change the way we treat diseases and provide immense value to patients and to society. 

How soon do you anticipate the end of the single molecular target?

The single molecular target will remain a possible starting point from relatively simple diseases that can be reduced to a single protein to target. A lot of these diseases have already been researched thoroughly, but many others – especially multisystemic diseases that are some of today’s leading causes of death – cannot be reduced to just one target and so require more complex approaches. To address such complicated diseases, we need to generate novel insights – and now we have AI and ML to do that. 

How do you expect the clinical trials process to be augmented as a result?
 

With the impact that AI and ML are having on our industry, I can see a future in which most drug candidates will be successful and where the significant cost savings generated as a result will significantly decrease the price of medicines, and make innovation accessible to everyone across the world. Imagine what that could do for a person living with a debilitating disease – and the impact that could have on that person’s loved ones.