By Dr Preeti Verma, Associate Vice President, Services Solutions, PV Operations, Qinecsa
In this blog we dive into the latest advancements in pharmacovigilance (PV) literature screening and the emerging trends shaping its future.
We share insights into overcoming industry challenges and how to leverage artificial intelligence (AI) and automation.
Marketing Authorization Holders (MAHs) need to comply with a myriad of regulations to meet global and local requirements.
This requires a proactive risk management approach and overall compliance with safety standards.
To succeed, MAHs must understand how to build a quality-driven compliant literature management process, overcome common challenges and seize the opportunities offered by new technologies.
How to build a quality-driven compliant literature management process
The key responsibilities of MAHs begin with identifying the countries where products are authorised to understand the applicable global and local regulations which will be governing their literature management process.
Regular literature screening can include validated global literature databases, local country specific databases such as LILACS, local non-indexed journals and meeting abstracts and draft manuscripts.
Sponsors can then identify valid Individual Case Safety Reports (ICSRs) and new safety information and signals. Search strings should be executed at the right frequencies to ensure the most conservative regulation is followed.
Preeti Verma
Once the strings have been executed and resulted obtained, they should be distributed for screening and quality review throughout the time duration before submitting to regulators in a timely fashion.
Finally, the entire process should be documented and records archived so the system is always audit and inspection ready.
Challenges
The first set of challenges MAHs face when designing and managing a compliant literature management process is around data.
The sheer volume and diversity of data sources can lead to significant overlap in the data under review and make timely review a mammoth task.
Additional complexities are added by the limited digital availability of local journals and the need for specialist translation and interpretation.
Interpretation of local literature becomes especially critical in assessing and evaluating data in the context of local conditions and cultures.
Unless a reviewer has cultural insights, there is always a risk of either missing the safety information altogether or misinterpreting information.
The next set of challenges is technical. Keeping track of, and staying compliant with, ever-evolving global and local regulations is a huge challenge.
Search strategies must continuously adjust to remain dynamic and fit for the future.
If legacy systems are being used, they may not have the robust search algorithms needed to identify new safety information.
The third set of challenges is operational.
The biggest operational challenge is resource constraint – both in terms of expertise and experience – and the number of resources required to deliver a task with quality and timeliness.
Any delay in searches or review can lead to non-compliance or delay in risk mitigation. The quality of review is equally important.
The net must be wide enough to capture what is relevant while simultaneously eliminating the noise.
The final set of challenges is financial constraints. These include operational costs, regulatory compliance costs, manpower expenses and scalability.
Solutions
There are specific solutions to overcome these challenges. We need to develop a harmonised framework and unified processes to address global regulatory requirements and stay agile to cater to diverse and unique local needs.
It is imperative to centralise data and processes across all PV domains which utilise literature outputs so there is always a single source of truth.
We need to collaborate with localized specialists to access and review local literature and invest in robust training.
The critical solution is the automation of literature management processes and innovations including artificial intelligence (AI).
Automation and AI technologies including and natural language processing (NLP) enhance the efficiency, accuracy and scalability of managing end-to-end scientific and medical literature and will continue to do as the technologies become more robust and more advanced.
Next-generation innovations in literature screening
Automation helps streamline literature management workflows, reduce manual workloads and ensure faster turnaround times for identifying adverse events.
AI technologies, especially NLP for extraction of data, can automatically scan vast amounts of scientific literature very quickly and recognize specific keywords, significantly reducing manual efforts and improving quality.
AI can also support automated journal monitoring while web scraping (keeping in mind global and local legalities) and RSS feeds track journal updates in real time.
It allows continuous monitoring of scientific and medical publications worldwide, resulting in quick and real-time identification of new relevant safety information.
Advanced AI translation algorithms allow for the processing of literature in multiple languages, broadening the scope of safety signal detection and ensuring comprehensive coverage across diverse linguistic resources.
Machine learning (ML) models can be deployed to detect patterns and correlations and identify anomalies within larger data sets.
These help with the early identification of safety signals that might have been missed through traditional methods.
AI can also assist in the prioritisation of critical safety signals by assessing the severity and potential impact of findings.
An extension to ML is predictive analytics. These models can predict potential areas by analysing historical data and current real-world evidence.
The value of all these innovations lies in being able to proactively detect safety signals early on, intervene promptly and enhance patient safety and public health.
The literature platforms available today are not just review platforms. They help streamline, manage performance and improve status and compliance reporting to relevant stakeholders.
Future trends
The advancements already mentioned underscore the pivotal role of AI in changing literature management. Now the focus has shifted to proactivity and more precise drug safety surveillance.
For example, large language models (LLMs) are being adapted to generate SQL queries from natural language inputs. This allows non-technical users to access and analyze PV data more effectively.
Autonomous AI agents employ LLMs for prompt generation and iterative optimisation, automating data collection and analysis.
Explainable AI models can assess how specific drugs contribute to the likelihood of serious events, providing insights from real-world data to inform and modulate PV monitoring.
A new trend is blockchain for PV transparency.
This offers significant opportunities to reinvent the way in which pharmaceutical companies access, collect, distribute, share, leverage, monitor and audit clinical trial data on medical and patient records.
There is also a growing emphasis on patient centricity in PV activities.
This includes empowering patients to report adverse events through user-friendly digital platforms, enhancing the detection of previously unknown drug reactions.
The final trend is increasing regulatory harmonisation and international collaboration in PV. Efforts are being made to standardise adverse event reporting requirements and streamline safety monitoring processes globally.
These efforts will eventually lead to more efficient PV practices and help us share safety data and best practices, ultimately improving safety outcomes worldwide.
About the author
Preeti is a qualified MD with over 15 years’ experience in clinical practice and the pharmacovigilance industry.
She is a highly experienced client partner and consensus builder, enabling her to develop strategic solutions that address client needs.
