AI is going to transform the pharmaceutical industry. Here's how.

From lowering risk during drug trials to better study design and faster results, Big Data, the cloud, and AI are transforming how medicine gets made.
Written by Greg Nichols, Contributing Writer

The pharmaceutical industry could be on the cusp of a major transformation as artificial intelligence, Big Data, and the cloud begin transforming how drugs are made and tested.

A recent Intelligence Unit report from The Economist and research sponsor PAREXEL, one of the biopharmaceutical services companies driving the AI push in clinical trials, found that Big Data innovations reduced time for recruitment in clinical trials by 37 percent and that drugs developed with new AI tools were 16 percent more likely to reach market launch.

To get some context to those findings, I reached out to Dr. Isabelle de Zegher, VP of Engineering at PAREXEL Informatics.

GN: I was surprised to learn that the cloud and Big Data analytics were not really being used in clinical drug trials until very recently. Why is that?

Dr. Z: There are probably two reasons. For one, the need for the cloud and big data have historically been limited. While the pharma industry is processing a lot of documents and data as a support to drug development and regulatory submission, this has been in limited volumes compared to other industries such as aeronautics.

In addition, pharma is a highly regulated industry and as such companies need to be compliant with regulatory guidelines. We are no different than other regulated industries, such as finance. However, the pharma industry has the additional constraint of ensuring data privacy on patient identification. This has created a lot of reluctance to be more "open" to the cloud.

GN: How will AI impact the way clinical trials are designed, which I understand has an outsized affect on whether or not the trials will produce quality data?

Dr. Z: AI can be applied to optimize protocols in different ways by simulating different clinical trial designs and identifying the most appropriate one considering patient population, evidence of diseases, and lessons learned from previous similar trials. Study optimization involves comparing cost and speed of various potential scenarios for trial execution - across countries, site, patient population - to identify the most streamlined protocol.

GN: Who's using AI in drug development and trial design right now? How pervasive has it become?

Dr. Z: There is still limited use of AI in drug development and more particularly in clinical development. One of the main hurdles is the quality - or lack thereof - of data researchers have historically had access to. We start to see new solutions emerging across drug development from protocol optimization to outcome measurement and reimbursement, though these are still in the early stages. As processes and technologies like the cloud continue to improve data collection, it will enable researchers to incorporate AI into more areas of the healthcare continuum.

GN: What will the drug development process look like a decade from now? Will it simply be more streamlined, or are there more foundational changes to come with enhanced AI capabilities?

Dr. Z: In ten years we will see fundamental changes in the traditional Phase 1 to 4, site-centric models of drug development. This will not be only because of AI, but rather a combination of Big Data - supported by sensors and electronic medical record (EMR) integration - improved telecommunication/ telemedicine, and AI.

There will be much less patient life trials, and instead they will be replaced by "in silico" trials which can mimic the patient. This will be especially prevalent for the Phase 1 studies that require healthy volunteers.

There will be more patient centric and remote trials that are more accessible for all patients. Direct access to EMR, linked with de-identification techniques, will allow researchers to find patients far more effectively than today. There will be limited need for patients to visit a clinical site. A combination of sensors, study apps, bots, and local home care nursing supported by referral physicians, will allow patients to participate in a trial while remaining at home. This use of AI will be most common for confirmatory Phase 3b and 4 trials that have large patient populations.

Editorial standards