The rise of artificial intelligence (A.I.) in the pharmaceutical industry signifies a transformative period, placing unprecedented power in the hands of tech-driven drug discovery and igniting debates about innovation, control, and patient safety. Subject to regulatory scrutiny, the FDA has expressed a commitment to balancing the promotion of innovation with the protection of patient safety. Companies like Immunai are spearheading this movement through partnerships, such as their $18 million deal with AstraZeneca to leverage Immunai’s A.I.-powered immune system model for clinical trial optimization. Startups like Insilico Medicine and Recursion Pharmaceuticals also champion A.I. as their key to discovering new drugs, though some critics argue these claims could be exaggerated, a phenomenon known as A.I. washing.
The Role of A.I. in Drug Discovery
Raviv Pryluk, co-founder and CEO of PhaseV, acknowledges the spectrum of opinions on A.I.’s efficacy in drug development, positioning himself between optimistic and skeptical viewpoints. Drawing from his experience as SVP of operations and analytics at Immunai, Pryluk now focuses on using A.I. and machine learning to streamline clinical trials with PhaseV. He emphasizes that A.I.’s value in this domain is contingent upon careful and deliberate implementation.
Since 1995, the FDA has received over 300 submissions for drugs and biological products containing A.I. components—a figure that illustrates the growing interest and investment in A.I. technologies. Dave Latshaw II, former head of A.I. drug development at Johnson & Johnson, notes the exponential growth in the complexity and promise of A.I., driven primarily by the scale and quality of data now available. While A.I. has already influenced clinical trials by better matching patients with appropriate studies and targeting specific populations, these advancements come with their own set of challenges. Narrowing patient types can yield more accurate results but simultaneously restrict the commercial appeal of a drug by limiting the diversity of the population it serves.
Optimizing Clinical Trials with A.I.
Latshaw, who co-founded BioPhy after leaving Johnson & Johnson in 2020, seeks to bridge the gap between technological advancement and commercial priorities. By refining the entire drug discovery pipeline, including clinical trial optimization, BioPhy aims to overcome inefficiencies that larger, more bureaucratic organizations struggle with.
A.I. also has the potential to transform mergers and acquisitions in the pharmaceutical industry. By sifting through vast amounts of pre-clinical and clinical data, A.I. can help identify the most promising drugs for acquisition, thereby optimizing portfolio management for pharmaceutical giants. According to Pryluk, machine learning and A.I. can analyze data and literature to guide better decision-making in acquisitions.
Another significant advantage of A.I. is its potential to enhance the diversity of clinical studies, which has been a persistent challenge despite long-standing mandates, such as the NIH’s 1993 requirement for the inclusion of women in research. Advanced A.I. methods, like causal machine learning, could tailor studies to ensure findings are applicable across demographics, potentially increasing participation from women and minority groups in clinical trials. Pryluk envisions this as a step toward precision medicine—treating each patient as a unique case—which he believes will be the industry’s next major shift.
Addressing Skepticism and Challenges
However, there is no shortage of skepticism regarding A.I.’s role in pharma. Every A.I. application must pass rigorous risk-benefit analysis, addressing concerns about data sharing, cybersecurity, algorithmic transparency, and biases. Genetic data, fundamental to many A.I. models, raises significant privacy issues, as evidenced by ongoing debates around companies like 23andMe.
Steven Aviv, CTO of Pentavere, underscores the importance of data trust. Pentavere’s DARWEN A.I. system examines unstructured healthcare data to identify patients for specific treatments. To ensure data integrity and compliance, Aviv and his team utilize Databricks’ Data Intelligence Platform. Latshaw asserts that transparency in A.I. solutions is crucial to avoiding misinterpretation of outputs. He warns against prioritizing marginally innovative methods over asking the right research questions. For example, predicting protein structure does not necessarily equate to accurately forecasting its function and disease impact.
The Future of A.I. in Pharma
Latshaw, after departing from Johnson & Johnson in 2020, co-founded BioPhy with the mission of bridging the gap between technological innovation and business goals. BioPhy works to refine the entire drug discovery pipeline, including optimizing clinical trials, to address inefficiencies often seen in larger, more bureaucratic companies.
A.I. also holds the promise of revolutionizing mergers and acquisitions in the pharmaceutical sector. It can analyze vast amounts of pre-clinical and clinical data to identify the most promising drugs for acquisition, thereby optimizing portfolio management. Pryluk notes that machine learning and A.I. can scan data and literature to inform better acquisition strategies.
Another notable benefit of A.I. is its ability to improve the diversity of clinical trials, a challenge despite mandates like the NIH’s 1993 requirement to include women in research. Advanced methods, such as causal machine learning, could customize studies to ensure they are relevant across various demographics, increasing participation from women and minority groups. Pryluk sees this as a move toward precision medicine—treating each patient as a unique case—which he believes will be the industry’s next major shift.
