John Cogan, COO, Qinecsa outlines the opportunity presented by AI-driven automation to transform pharmacovigilance, both to address soaring workloads and costs, and to inject new strategic purpose into the function’s role.
dee karen Shutterstock
Pharmacovigilance (PV), and most notably its core task of tracking and processing adverse events (AEs), is both essential for regulatory compliance and a huge and costly drain on resources. It monopolises scientists’ time while adding very little value beyond fulfilling authority expectations and managing risk. In the context of long-established drugs, this laborious work feels much like a tick-box exercise. It makes sense then to streamline or outsource processes wherever possible, and let technology take the strain at every opportunity.
But what if PV could go further, do more? Certainly, new modalities and treatments with a higher-risk safety profile require additional vigilance and detection sensitivity, and as manufacturers become bolder in their ambitions here, it follows that their PV operations need to adapt too.
There is always a chance that unforeseen side-effects could emerge once used outside of a controlled environment, and optimised AE reporting is one of the best proactive protections against this. (The World Health Organisation estimates that adverse drug reactions account for 5% of hospital admissions globally.)
The latest new therapies and drug applications to enter the market include GLP-1 receptor agonist/weight-loss drugs to treat obesity in adults, and the use of messenger ribonucleic acid (mRNA) technology in approaches to cancer. Meanwhile, pioneering COVID-19 vaccines, given to a significant proportion of the global population, still need to be closely monitored for potential AEs that could present over time.
In this context, beyond the core remit of meeting regulatory expectations, optimised PV is essential to staying ahead of emerging issues and proactively optimising patient safety. Highlighting this, in England reports of pancreatic issues apparently linked to the use of weight-loss injections have triggered a new study into the treatments’ side effects.
Could AI be the answer?
AI solutions specifically designed for PV are entering the market in earnest now, and top-tier pharma organisations are testing them with encouraging results, hoping that the technology might hold the key both to improved productivity/cost efficiency, and enhanced proactivity.
Current solutions have been shown to reliably handle large volumes of data, extract key information from various sources, and even detect subtle patterns that might be missed by human reviewers. (Implementing AI in PV has improved the detection of potential drug risks by over 25% since its introduction, according to the US FDA.)
Yet there is no one-size-fits-all approach to AI adoption and associated PV process transformation.
Large pharma organisations, processing many hundreds of thousands of spontaneous AE cases each year across extensive and diverse portfolios, stand to gain most from intelligent automation and letting technology take the strain – whether they do this themselves, or via a service partner. But even here it isn’t just the scale of the operation that will determine the optimum path to AI use. The prospect of turning off legacy safety databases can be daunting, in which case embracing AI may need to involve sophisticated workarounds and wrap-around software. This is likely to involve giving users a single, simple, intuitive interface or portal and providing a layer where the ‘magic’ happens while preserving the core safety database’s integrity and investment.
Mid-size pharma organisations - with more modest portfolios yet concerns about resources, PV scalability and responsiveness - may want to progress to a digital-first capability, yet be unsure of the most cost-effective way to go about this. To prepare the ground, their best bet may be to start with the inbound AE case capture process, ensuring that this is digitised across all supported channels.
The more that cases can be captured digitally at source, the greater the potential impact of AI in their assessment and processing. As well as creating a substantial efficiency gain in its own right, by doing away with manual data entry, it also sets up the PV operation for all of the downstream processing to be digital.
Setting goals
Doing nothing is not an option, certainly. Potential cost efficiencies will help cement the business case, but drug manufacturers shouldn’t overlook the strategic benefits on offer.
The upshot is that PV’s role and profile needs to evolve. Already today the function is becoming steadily more strategically important, where enablers are in place. It is poised to become a key partner in drug development and product lifecycle management, for instance (a role that would involve integrating risk management and leveraging technology to improve processes and decision-making).
Capitalising on advanced technologies to hone everyday practices will help companies to realise that scenario.