Scroll Top

How artificial intelligence and machine learning are changing drug discovery

Imagine an industry where a failure rate of greater than 90% is considered acceptable or even pretty good, and in some especially competitive segments, could be even lower4. Where the product development timeframe can easily run more than a decade and the costs to develop that product could exceed $2 billion1. Now consider that even when the products reach the market, there’s still a chance they could be pulled later, even 30 or more years after launch2,3.  It should come as no surprise to anyone in healthcare and the life sciences that this scenario describes the pharmaceutical industry today.

How to change the pharma paradigm has been the topic of many analysts4,5. But for all the ideas about homing in on the right biomarkers, leveraging new clinical trials study designs, integrating real-world evidence, and tightening up the intended patient population, there has been little impact on streamlining the drug development process or reducing the capital costs to bring a new drug to market. Simply stated, the current system is too expensive, takes too long, and has too many pitfalls for patients, providers, researchers, and investors.

In MoneyBall Medicine: Thriving in the New Data-Driven Healthcare Market, I described how one company is changing the drug development paradigm by taking advantage of the data deluge in healthcare, emerging technology, and a back-to-basics biological approach. Boston-area’s Berg, founded in 2006 by Carl E. Berg, Mitch Gray, and Niven R. Narain, is using artificial intelligence for a proprietary Interrogative Biology platform and bAIcis analytics software6. In a 2015 WIRED UK talk, Narain describes how their software builds a “map” of the patient that includes everything from all of their clinical data to the biological information of their condition (e.g., genomic profiling of a tissue sample) to more completely describe the patient in terms of their disorder, allowing them to gain a better understanding of disease process along the way1,7.

Using artificial intelligence to analyze the massive amounts of data available for each patient gives Berg the opportunity to dive deeper into what makes someone respond to treatment or not, identifying the key parameter more quickly and accurately than the human-driven methods of today. Unlike their human counterparts, the AI program can be hypothesis-agnostic, so finding a new correlation between a drug and a cancer target may take weeks or months rather than years of trial and error through pre-clinical and clinical trials. Use of data and technology in this way is just one application of MoneyBall Medicine ideas—and one that has garnered Berg partnerships with pharmaceutical companies such as AstraZeneca, SanofiPasteur, and BD, and others in the traditional academic healthcare and government sectors.

In this episode of the (Episode 2) MoneyBall Medicine podcast, I sit down with Berg co-founder, President and CEO Niven Narain. We discuss where Narain thinks the pharmaceutical industry is headed by using AI for drug development and the business and human asset implications. He describes the success Berg has had with one of their drug targets for pancreatic cancer, a disease with one of the highest mortality rates and why their AI-backed platform allows them to go after ultra-rare diseases where other companies might shy away. Join us for this episode of the MoneyBall Medicine podcast!

  1. DiMasi, J. A., Grabowski, H. G. and Hansen, R. W. 2016. Innovation in the pharmaceutical industry: New estimates of R&D costs. J Health Econ 47 20-33.
  2. Flapan, D. 2004. Vioxx Pulled From Global Market. Medscape. https://www.medscape.com/viewarticle/490355
  3. org. 2014 35 FDA-Approved Prescription Drugs Later Pulled from the Market.https://prescriptiondrugs.procon.org/view.resource.php?resourceID=005528
  4. Huss, R. 2016. The High Price of Failed Clinical Trials: Time to Rethink the Model. Clinical Leader. https://www.clinicalleader.com/doc/the-high-price-of-failed-clinical-trials-time-to-rethink-the-model-0001
  5. Lo, C. 2017. Counting the cost of failure in drug development. Pharmaceutical Technology. https://www.pharmaceutical-technology.com/features/featurecounting-the-cost-of-failure-in-drug-development-5813046/
  6. 2018 Company Info.http://www.berghealth.com/media/company/
  7. Temperton, J. 2015. Artificial intelligence is making better drugs. Wired. https://www.wired.co.uk/article/niven-r-narain-ai-drugs-wired2015