In pharma R&D, the adverse event case intake process, which takes up so much of Safety/pharmacovigilance professionals’ time and is far from efficient in its support of timely interventions, remains ripe for disruption. And now that promise is finally being honoured by next generation AI technologies. Specifically, Generative AI and Large Language Models are enabling the automation of humanlike decision-making, leading to earlier and more accurate conclusions about Safety events. ArisGlobal’s Emmanuel Belabe explains the tangible difference this has begun to make.
In the modern world it is expected that any authorised medicinal product designed for human use is safe for patients to consume. Pharmacovigilance (PV) processes, which continuously monitor the effects of drugs once on the market, are intended to uphold that position over time once products have been approved for distribution. Approaches to post-market Safety monitoring have changed little in decades, however, despite soaring volumes of available information. Today these are submitted in an increasing array of formats, via a proliferating range of channels.
The approach of “booking” cases, or determining whether the mandatory elements are present, remains prevalent, with a view to quickly assigning an identifier. This approach doesn’t take into account the actual contents of a case, however, which forces PV teams to apply the same treatment to all information. The effect is that all cases are assigned the same priority in the early stages; there is no discretion to allocate teams’ bandwidth according to a potential case’s complexity/risk.
Tracking all the potential signals, assessing their validity, and responding swiftly to relevant cues, is both an absolute mandate and a very costly and labour-intensive administrative burden. Adverse event (AE) case intake, in particular, represents one of the most overlooked and broken workflows in pharma in its current form.
Modern AI: Moving Away from Rigid Process Automation Towards More Nuanced Deductions
Technology-enabled process automation has long promised to transform the speed, efficiency, and accuracy of AE case intake and triage, by capturing and assessing relevant Safety signals arising via a wide range of channels (including self- or clinician-reported AEs submitted by email, post, phone call, or web portal, as well as mentions via online forums).
Up to now, machine intelligence has not come close to mimicking human powers of data extraction, filtering, inference, or deduction. Early excitement about this potential, while valid, was premature. Early automation systems had to be highly structured and painstakingly trained to recognise every possible format and variant of how important data might show up – from the basics such as a patient’s date of birth, to richer detail such as the combination of possible contributors to the adverse event (from the individual’s stage of life and overall state of health to other drugs they may be taking). As well as presenting challenges in how systems would recognise and extract the right data, this limited the scope for step changes in Safety process efficiency.