A Year in Review: Generative AI

Generative AI has quickly become a buzzword of 2023 because of its exponential growth and fast adoption. So far, generative AI leverages data-driven algorithms to create new content, such as text, images, audio, and synthetic data that can uncover hidden insights. These models learn from the input of information and data in which they are trained on, to then be able to produce new outputs.

Because of its widespread uses and capabilities, gen AI has been adopted across a variety of industries. 79% of all respondents say they’ve had at least some exposure to gen AI, either for work or outside of work, and 22% say they’re regularly using it in their own work” (McKinsey, 2023). But in healthcare, pharma, and medical device industries, only 6% of people say they regularly use gen AI for work (McKinsey, 2023). This is expected to rapidly change, as almost three-quarters (70%) of CEOs report that they’re accelerating gen AI investments to maintain competitive advantage (EY, 2023). 

Leading organizations, such as Sanofi have already seen success in implementing AI into their operations at a large scale. Additionally, Insilico successfully had their gen AI-discovered drug enter phase 2, and new therapies are being discovered with AI for women’s cancers (Drug Dev, 2023). 

Adoption of generative AI, in part, can be hindered because AI is only as trustworthy as the data it’s trained on. This is especially critical in the healthcare and life sciences industry where sensitive personal and private information is at stake. There have been instances that prove the technology, while helpful, is not always flawless—inaccuracy being one of the most commonly cited risks (McKinsey, 2023). 

Generative AI at Medidata 

Medidata has developed sophisticated technology leveraging gen AI to gather and unlock sensitive data out of regulated environments. With over 20 years of experience running global trials across a wide range of diseases and treatment types, Medidata provides unparalleled expertise that can be coupled with AI to offer unique solutions for clinical developers. Medidata has worked across nearly all disease areas, and has created unique solutions for clinical developers working in areas including cardio-metabolic, oncology, and treatments like CAR-T. These solutions can help with speeding up trial times, determining appropriate endpoints, evaluating treatment effects, and allowing for companies to derive insights that may have otherwise been impossible without historical clinical trial data and advanced analytics.

Building on our success in AI and advanced analytics using historical clinical trial data, this year, Medidata AI launched Simulants. This suite of innovative, proprietary algorithms uses the most advanced techniques in generative AI to create synthetic clinical trial data based on actual historical clinical trial data. Synthetic clinical trial data is designed to mimic the patterns and characteristics of the actual, underlying historical trial data while ensuring that it remains anonymized and compliant with privacy regulations. Simulants is a first-of-its kind solution that allows for expanded access to the novel insights generated from past clinical trials, while safeguarding participant and sponsor privacy. 

The Medidata AI Simulants team secured a patent on the innovative process for synthetic clinical trial data generation in 2023. This model was showcased at numerous well-known and peer-reviewed conferences in the industry, including ICML, where the team also won the highly competitive best paper award. 

Because of Medidata's unique gen AI and synthetic trial data, organizations are interested in applying it to their clinical development programs in a variety of ways, including, but not limited to; predicting treatment responses and patient outcomes to select optimal endpoints; conducting risk assessments to understand the likelihood of serious adverse events within sub-populations; augmenting datasets to upsample underrepresented groups. The impact that gen AI will have on the clinical landscape is only just beginning.