#AI22 is a series of articles highlighting what we believe to be 10 developments that will be impacting AI this year.
This series is co-written by Dr. Johannes Otterbach, Dr. Rasmus Rothe and Henry Schröder.
While AI in the pharmaceutical and biological industry has stepped out of its infancy, the next steps will be of exponentially increasing stride. From Novartis using AI to predict the drug sufficiency of untested cell components in order to accelerate the drug discovery process to BioNTech applying an AI tool in detecting the risk of Covid-19 variants to Tencent using AI in determining the severity of Parkinson symptoms, there are already major developments in this field that have been applied. Nonetheless, the full capacities of these technologies are still to be adapted.
Notably, two extremely capital-intensive elements of R&D in the pharmaceutical industry will be revolutionized through AI: drug discovery and clinical trials. A paradigm shift in drug discovery, enabled by AI, was first materialized by Google’s AlphaFold in 2020 and significantly advanced by AlphaFold2 in 2021. Its ability to predict the protein structure of cells, a technique known as protein folding, by using AI, allows scientists to work at an exponentially higher pace. As proteins are molecules essential to life in all organisms, this ability to predict their shape and thereby also gain insights on its inner workings in a much shorter period gives life science researchers the ability to research a much more extensive variety of proteins - and therefore also the possibility of discovering new therapeutic compounds. While by the mid-2000s, the available protein structure data contained information from 100 million different compounds, by the end of the 2010s there were 200 billion compounds discovered to have potential therapeutic value. AI-powered drug discovery companies will further extrapolate the benefits to speed, precision and thereby heavily decreased R&D costs, through AI and become leading in the field.
In the case of clinical trials, AI will aid in problems like study design, site identification, patient recruitment, or processing of datasets. Clinical trials are expensive to perform and challenging to design. 85% of all clinical trials fail to meet their timeline. With the help of AI, companies try to prevent these expensive delays, for example by applying AI to identify eligible patients based on their cancer diagnosis and stage, to enroll in clinical treatments. Similar applications will take off in this space: from bio-simulations that predict the clinical value of drugs before starting expensive trials, or federated learning-based programs that connect clinical trial data from around the globe to identify patient populations of interest or discover certain treatment reactions.
The use of AI in pharma is very specialized and specific for certain diseases and processes and thereby prone to specialized start-ups. Therefore big pharma players in the already M&A intensive industry will increase their AI acquisition and partnership activity in the coming years.