Data regulations impeding AI use in drug research

Artificial Intelligence has been hailed as a way of accelerating faster timelines in drug discovery and development, but a new survey reveals data silos and outdated regulations represent some of the biggest barriers to its adoption.

The survey has been carried out by non-profit group The Pistoia Alliance, and found 52% of life sciences professionals across the US and Europe cited this reason as the biggest challenge to AI and machine learning adoption, a figure that has swollen from 28% in 2017.

Recent weeks have seen more alliances forged in the field, with Celgene working with Exscientia and AstraZeneca partnering with BenevolentAI.

The use of AI in the life sciences industry has increased in the past two years, with 70% of respondents stating they are using AI, including machine learning and deep learning in some capacity, up from 44% in 2017.

Steve Arlington

“There is now also an abundance of data streams – such as Real World Evidence, clinical trials data, and genomic data, which could have real value in drug discovery and development, as long as we’re able to analyse it,” said Dr Steve Arlington, President of The Pistoia Alliance.

“The industry must work closely with academic organisations and educators to highlight these opportunities, and attract the next generation of data scientists.”

Lack of skills was also another major concern, with 44% of survey respondents revealing that this was preventing AI uptake.

“We need highly-trained, specialist data experts to meet this challenge in order to make the technology work for this industry,” said Dr Arlington.

The survey also highlighted that lack of quality data was also presenting another hurdle, with 66% of respondents saying data quality was the biggest barrier to using AI in drug design in 2018.

“Data quality is a problem the life science industry can take immediate steps to address” said The Pistoia Alliance, which advised life sciences professionals to ensure their data complies with the FAIR principles, also known as Findable, Accessible, Interoperable and Re-useable.

Despite these challenges, less than half (46%) of respondents have a team of data scientists at their company dedicated to improving data quality, and only 15% said their company is planning to build such a team.

“Our research has shown the life science industry is very interested in AI adoption, but that the same issues are still hampering it’s use,” commented Dr Nick Lynch, Strategy Lead for The Pistoia Alliance’s AI Center of Excellence.

“This is why The Pistoia Alliance created our Center of Excellence in AI for Life Sciences. We wanted to provide an opportunity for the industry to work collaboratively on implementing AI successfully, from sharing best practices to collaborating on improving access to quality data – including working to implement standardised data formats that will accelerate adoption.

“But this won’t happen until we have the life science industry, technology specialists, vendors, and regulators, all in the same ‘room’ and working together to solve the same problems.”

The organisation has called upon the life sciences sector to offer guidance on the topics it should provide further education and training around, offering those in the industry to complete a survey.

“Given AI adoption in the life sciences looks set to flourish, in order to safeguard patients and reassure regulators, those in the industry must work together to ratify and implement data standards and protocols,” it said.