AI and Machine Learning Facilitates Faster Path to Personalized Medicine


November 11, 2019
Artificial intelligence (AI) expert Jackie Hunter reveals how mining the depths of millions of hours of research identifies potential targets and solid candidates for drug discovery as it builds a future for individualized patient care.

“A lot of people talk about the future of personalized medicine, but I think with the technology we have now it’s possible to make that a reality sooner rather than later,” says Jackie Hunter, Ph.D., D.Sc., C.B.E. “Artificial intelligence and machine learning can change drug discovery…now. We can use technology to delve into the whole range of public and proprietary information from the clinic to understand patients a lot better and come up with the correct therapy for the right patient.”

Initiatives all over the world generate enormous amounts of biomedical research that hold the potential to transform biotech, “but only if AI and machine learning can unlock insights for better medicine,” Hunter continues. “We need to leverage this technology to mine biomedical information for the most salient facts to develop new hypotheses.”

Hunter explores how digging into big data is transforming drug discovery during an upcoming keynote presentation at SLAS2020, January 25-29, 2020 (San Diego, CA, USA). “I want the presentation to open people’s eyes to the huge potential of AI and machine learning technology in all areas of drug discovery,” says Hunter, who is chief executive, Clinical Programs and Strategic Partnerships at BenevolentAI, (London, UK), an AI company founded in 2013 that develops and applies advanced technologies to transform the ways that drugs are discovered and developed.

AI platforms cut the time to develop a lead molecule into a candidate from three years to just over one year, Hunter states. "Researchers are able to test their hypotheses more quickly because they have the right molecules, and also because they haven’t wasted time on 90 percent of the molecules they might have worked on otherwise,” she explains.

Applications for this technology span the breadth of life sciences from drug discovery and long-term management of chronic diseases such as diabetes to areas such as radiology and pathology. “If a company can use AI to cut the time for analyzing MRI images from two hours to about three minutes, you have to believe that this is going to have a major impact on healthcare cost and delivery,” Hunter comments.

For the SLAS2020 audience, Hunter wants to expand how AI could be used to make specific systems, applications or processes used in the lab more cost effective or valuable. “I want researchers to understand how they can combine the outcomes of multiplex assays to do everything from interrogating assay positions and reproducibility to increasing information and insight,” Hunter says. She also hopes to inspire the life sciences community with new ways to collaborate.

“We’ve spent a lot of time at BenevolentAI thinking carefully about how to get people working together,” she says, adding that much consideration has gone into how to get data scientists to work effectively with biology’s messy data.

“I want biologists and biochemists to focus on the questions they really want to explore,” she continues. “I want manufacturers engaged in making and selling new equipment or devising new processes to embrace this technology because it’s the way manufacturing capabilities will be enhanced in the future.”

She acknowledges the learning curve for adopting AI and machine learning is steep, and comments that “mastering it is critical to succeed. Those companies and individuals who are going to be able to work and think across their silos and even create new industries need to embrace this new technology.”

Advancing Science by Exploring Existing Research

From her perspective of 30-plus years in pharmaceuticals, Hunter says the industry hasn’t been able to maintain the drug discovery success it had from 1960 through 1990. “We have basically taken all the low-hanging fruit, the easy targets such as the beta agonists, and we have not been able to deliver effective treatments to patients or society in some devastating diseases,” she continues.

Getting drugs to market is hampered by the inability to identify and validate the right targets in a particular patient population, according to Hunter. “In the past, the industry had issues in predicting how a drug would behave pharmacokinetically in people, but now we can predict that much better. However, we are much less able to predict whether a molecule will have unwanted side effects or even toxicities.

“I believe that’s an area where analyzing data using AI technology will be able to make an impact,” Hunter comments. “Patient derived materials are more reflective of environment under which the target of interest is working in the patient population itself, but they are often scarce and hard to come by. Therefore if we can use AI to improve our ability to get more information from assay systems – using these  patient-derived materials – then this will help accelerate drug discovery.”

Hunter also believes researchers can use AI to access greater chemical diversity. “Machine learning is able to analyze the data and come up with synthetic routes almost instantaneously so that computers suggest new molecules to make and also how to make them,” she explains. “This couldn’t happen before because the synthetic tractability had not been incorporated into the suggestions and the new synthetic routes proposed. Now we can do that.”

Hunter describes how BenevolentAI’s technology platform builds a knowledge graph drawn from billions of data points harvested from scientific literature that provides a map to navigate what is understood about a particular mechanism or system in a disease and allows researchers to quickly home in on tractable nodes up or downstream of an intractable, but highly relevant target. This technology gives you a way to maximize the benefits of your investments while exploring questions you couldn’t before,” says Hunter.

She offers as an example BenevolentAI’s quest for treatments for amyotrophic lateral sclerosis (ALS), or Lou Gehrig's disease, an incurable condition that progresses rapidly and lacks a standard of care. Research has uncovered more than 30 genes that have implications in the disease, yet approximately 85 percent of patients do not have mutations in these genes. Many patients die within two years of diagnosis.

To pursue options for ALS patients, BenevolentAI's team used the company’s technology platform to produce a ranked list of potential treatments, together with biological evidence, and classified these predictions using strategies focused on pathways implicated in multiple ALS processes, and selected the five most promising compounds. The team from BenevolentAI then presented this data to the Sheffield Institute for Translational Neuroscience (SITraN), a world authority on ALS, for further testing. From this collaboration, an ALS lead molecule emerged from a breast cancer drug, which showed delay of symptom onset when tested in the gold standard disease model.

“BenevolentAI discovered some early indications from in vitro and in vivo work that the hypotheses we have generated are having some effect on ALS,” Hunter says. “One of the great things about stem cell biology is that we can take actual stem cells from patients and, in the case of ALS, use those to create astrocytes and motor neurons to show that a mechanism is important in a particular subset of patients or indeed across all subsets of patients. We believe that allows us to more accurately replicate the characteristics of disease in the lab.”

Hunter comments that BenevolentAI made significant progress in identifying the right patients and making an impact at stages all along the drug discovery and development process. “It’s challenging, but it’s stimulating and rewarding when you get something positive out of it,” she says.

The company's most recent workflows for hypothesis generation have resulted in another lead molecule, BEN-XX1, an optimized compound showing improved central nervous system exposure and profound rescue effect in ALS patient cells. BenevolentAI is currently in late-stage lead optimization of BEN-XX1 aiming to nominate a clinical candidate as soon as possible.

Gaining Perspective and Giving Back to the Life Sciences Community

Hunter’s early experiences as a patient inspire her push for effective treatments. “When I was little, I had asthma,” she explains. “I was very lucky that shortly after my asthma diagnosis, the pharmaceutical industry released beta agonists, which transformed the treatment of acute asthma attacks and had a huge impact on my life. Since that time, my educational and career choices have been motivated by understanding the causes of illness and using that knowledge to develop new and more effective treatments.”

Hunter’s journey has yielded professional achievements that span academia and industry. She started her career doing basic research on the behavior of monkeys at the Zoological Society of London, followed by a post-doctoral fellowship on reproductive behavior at St. George’s Hospital (London, UK) and then worked 19 years in the pharmaceutical industry ultimately becoming the head of one of GlaxoSmithKline’s (GSK) Centers of Excellence for Drug Discovery where she led neurological and gastrointestinal drug discovery and early clinical development.

“After that I set up my own company, OI Pharma Partners, so I’ve experienced running a small start-up as well. Then I went back to academia to run one of the research councils,” she comments, adding that these experiences led her to BenevolentAI in 2016 where now she seeks to accelerate drug discovery and development.

“I love the work that I do. It opens my eyes to just how impactful AI and machine learning are in all aspects of healthcare,” says Hunter. “I feel like I’ve been moving in this direction for my whole career.”

Her current goals at BenevolentAI include tackling other high medical need diseases – such as Parkinson’s, ulcerative colitis and sarcopenia  – where Hunter says that technology potentially could identify new hypotheses and essential biomarkers to use in the clinic to stratify patients earlier in drug development. With many new avenues opening, the busy researcher longs to be back in the lab. “I wish I was decades younger because it’s such an incredibly exciting time to be a student in any discipline of life sciences,” says Hunter.

In the face of all this opportunity and in spite of her busy schedule, Hunter still takes time to stimulate her thought process by enjoying a gentle jog or spending time in the garden. “I’m sure every scientist will relate to this, but quite often when you’re at work, you get overwhelmed. You don’t really have time to think deeply on a problem,” Hunter explains. “When you engage in certain leisure tasks, your mind has time to wander. Quite often it drifts back to work, and that is when you come up with some interesting thoughts. You can sort things out in your mind away from the workplace.”

She compares capturing a hypothesis to pulling up a dandelion in the garden. “There’s nothing like that satisfaction of really nailing it!” she says, with a laugh. “I like that. When I’m doing things like pulling dandelions, I don’t often switch off from work problems.”

When she does flip the work switch into the off position, Hunter re-establishes balance in her life by spending time with her family and dogs. “I like to cook and entertain. It’s important to set aside time for you, your friends and family.” Likewise, she says it’s important for life sciences professionals to purposely carve out time in their busy schedules for their community, no matter how many years they have invested in the profession.

She has advice for all walks of life sciences. For mid-term career people, she recommends blazing a trail with previously unexplored research endeavors: “Ask for forgiveness instead of permission when you want to try something new.” She prods students toward finding mentors and urges the seasoned pros to connect with peers.

“Many years ago, a fellow in the profession suggested that I apply to the strategy board at the Biotechnology and Biological Sciences Research Council (BBSRC). I never would have thought to do that, but I actually did get on the strategy board and that led ultimately to me being the chief executive at the BBSRC. I wouldn’t have done that without the fellow’s suggestion that I was a good fit for that opportunity,” Hunter says.

She encourages those at any experience level to take a moment to nurture the next generation of researchers. “Mentoring is one of the most rewarding things I have experienced. I always learn so much from it – it’s a two-way street,” she explains, advising senior professionals to spend more time with up-and-coming scientists. “They’ll learn a lot about themselves and give so much back to the community,” she says, adding that students need to be bold in asking for assistance. In fact, she advises students to seek out new opportunities and set aside concerns if they don’t have a grounding in a particular area.

“It’s important that students look at the impact of new technology and open their minds to collaboration between academia and industry,” Hunter says. “People always feel comfortable within their own discipline or in working with other people like themselves, but it’s important to find new avenues to learn and work.”


Learn More about AI in SLAS Technology: Browse the Special Issue, The Internet of Things in the Life Sciences Laboratory

From SLAS Discovery: Evaluation of Machine Learning Classifiers to Predict Compound Mechanism of Action When Transferred across Distinct Cell Lines

Also in SLAS Discovery: Design and Development of a Technology Platform for DNA-Encoded Library Production and Affinity Selection

Learn More About AI and Drug Discovery

Read About Jackie Hunter's Contribution to Women in AI