Over the last 10 years, use of AI has proliferated across the industry, but there is the sense that the envisioned potential is yet to be realized. There are limited examples of where we’ve seen success in drug development, for both speed and candidate progression, but also widespread implementation across R&D.
To ensure the industry utilizes full capabilities of these tools, Pharma needs a deep understanding of what is on offer, and how this can be integrated in internal ecosystems.
What does AI for target discovery offer?
Aiming to reduce time and cost, AI provides additional data-driven insights to R&D teams, including:
- Identification of genome-wide targets using -omics data
- Mapping disease pathways
- Protein binding site identification
- Understanding protein interactions, functions, and “drugability”
- Exploration of candidate molecules
- Understanding molecule-protein interactions
Applying pre-existing datasets to predictive models can help increase R&D teams’ likelihoods of success, reducing wasted efforts. Utilizing protein structure data with platforms such as Deepmind can help interpret protein interaction and drug-binding impact.
With this understanding, what is Pharma doing to implement these tools within R&D?
What progress has Pharma made?
Big Pharma and Biotech are no strangers, having previously embraced AI somehow. Primarily through partnerships, sharing capabilities has led to significant advancements, more so recently.
Some of the notable examples of “early success” in the discovery process include AstraZeneca and BenevolentAI’s continuing alliance to discover new druggable targets for chronic kidney disease, IPD, systemic lupus erythematosus, and heart failure. Similarly, Sanofi and Exscentia’s partnership is exploring novel candidates for personalized medicine across oncology and immunology.
While these are in pre-clinical development, Insilico Medicine recently announced that its application for a Ph1 trial of INS018_055 for IPD was approved by the Center for Drug Evaluation in China, making it the first AI-powered drug for this disease worldwide, and joining ~15 assets in clinical trials for other conditions.
In the short-term, due to significant investment required to develop in-house capabilities, AI-related endeavors will be driven through partnerships. However, the recent spike of companies providing an AI-based technology has made this challenging, offering little differentiation. Pharma companies will need to filter out “dead weight” finding valuable partners. Simultaneously, Pharma also needs to offer AI-based companies opportunities to grow/establish presence, as options remain for them in a crowded, largely untapped market.
It’s easy to be consumed the buzz of AI, however, previous successes need to be validated, in order to replicate achievement.
Despite seeing progress with candidates now in trials, all R&D needs to be transformed to explore all of what AI can provide. Hesitant of AI-use, companies so far have implemented tools in silo, in areas of reduced risk. However, adopting the mentality of “AI-First” across R&D should see the impact on metrics such as time saved, cost reduction, and overall success.
In an area with still untapped possibilities, greater ambition and risk appetite need to be seen across industry to explore the value of AI, with Pharma best placed to facilitate this.