Keeping It Real: Challenges And Benefits Of Integrating AI And Machine Learning Into Pharma R&D
Contributed Commentary by Tom O’Leary
January 3, 2020 | Productivity in the pharmaceutical industry is rapidly declining. In fact, the mean projected return on new drug research and development (R&D) investments by a dozen large biopharma firms fell from 10.1% in 2010 to 1.9% in 2018, according to a 2018 report by the Deloitte Centre for Health Solutions. This crisis has created a need to re-think the way clinical trials are conceived, designed and conducted. As such, artificial intelligence (AI)—which is defined as the simulation of human intelligence processes by machines or computers—is poised to transform the industry.
With the potential to automate processes, increase efficiency and enable more data-driven decisions in pharma R&D, interest in AI-driven solutions is growing steadily among industry leaders. In truth, the market volume for AI-based medical imaging, diagnostics, personal AI assistants, drug discovery, and genomics is projected to reach $10B by 2024. According to an ICON survey of more than 300 executives, managers, and professionals in biopharma and medical device development companies, nearly 80% of respondents said their firms plan to use, or are using, AI or Big Data approaches to improve R&D performance. Moreover, the umbrella category of AI and advanced analytics was seen as the digital technology with the most potential to improve R&D productivity, according to ICON’s 2019 report on digital disruption in biopharma.
While there are various applications of AI, one of the most disruptive is machine learning, which essentially allows the computer to automatically learn and improve its performance based on experience, rather than being continually programmed. Here we describe the various uses of machine learning in pharmaceutical R&D, as well as the challenges to adopting these technologies.
Uses of Machine Learning
One area of drug development where AI has substantial implications is in the development of biomarkers. For example, Berg is a biopharmaceutical company that is applying AI-driven modeling to develop diagnostics and biomarkers in the fields of oncology, endocrinology and neurology. In 2017, Sanofi Pasteur announced a partnership with Berg’s Interrogative Biology platform and bAIcis’ artificial intelligence tool. The partnership will allow these two companies to combine their expertise to identify molecular signatures and potential biomarkers for assessing the influenza vaccine immunological response.
Radiology / radiotherapy planning
Radiology and radiotherapy planning is also being impacted by AI, and this industry should anticipate further growth as imaging technology advances. For example, Google’s DeepMind Health is developing machine learning algorithms to detect the differences between healthy and cancerous tissues with the goal to improve the accuracy of radiotherapy while minimising damage to healthy organs at risk.
Clinical trial research
Machine learning also has many potential applications for improving clinical trials. When applied to data sets, such as social media data, genetic information or electronic health records, machine learning can be used to identify clinical trial participants more efficiently. Moreover, it can be used to facilitate remote monitoring and real-time data access to increase safety and patient adherence, and to determine the optimal sample size of a trial, adjust protocols for different trial sites and reduce data errors such as duplicate entries.
Finally, AI and machine learning have many potential uses in drug discovery. For example, earlier this year, AstraZeneca announced a deal with BenevolentAI, a UK-based company focused on combining computational medicine and advanced AI. The collaboration will combine AstraZeneca’s disease area expertise and large, diverse datasets with BenevolentAI’s AI machine learning capabilities to discover and develop new drugs for chronic kidney disease and idiopathic pulmonary fibrosis.
Barriers to Adoption
Despite the promise of AI and machine learning to transform the industry, putting these technologies into practice comes with a range of challenges. This is mainly due to the lack of relevant expertise and understanding. Some of the challenges that will need to be addressed include:
- Data governance challenges
- The need for transparent algorithms to meet drug development regulations
- Recruiting data science professionals
- Breaking down data silos
- Streamlining electronic records
To address these challenges, many pharma and biotechnology companies will continue to shift towards R&D outsourcing as interest in AI-driven technologies continue to rise.
AI and machine learning have the potential to address declining ROI in pharmaceutical R&D by improving biomarker development, radiotherapy planning, clinical research, drug discovery and much more. Despite its potential, the complex nature of AI and machine learning, in addition to the need for sophisticated infrastructure, are driving a trend towards outsourcing these capabilities.
Tom O’Leary is CIO at ICON. He can be reached at Thomas.OLeary@iconplc.com.