Facebook
  1. Home
  2. Insights
  3. Blog
  4. How AI could transform literature surveillance for pharmacovigilance

How AI could transform literature surveillance for pharmacovigilance

The use of Artificial Intelligence (AI) across several industries has increased in recent months. Healthcare research is no exception. In pharmacovigilance (PV) the focus is on how AI can be used to detect adverse events sooner. This use could potentially lead to faster risk identification. But as with any new technology, there are important regulatory considerations associated with AI’s use, accuracy and precision.

Managing information from literature surveillance

In a commercial setting, literature surveillance draws references from multiple sources. These could include global and local databases, journals and agency websites. Even when this information is regularly accessed bottlenecks can quickly form. This may be especially so if using processes with manual steps and interventions. Despite the benefits of a semi-automated system, team members will still spend a lot of time identifying relevant references.

Potential benefits of AI applications

A system with machine learning capabilities could help in repeatable scenarios like this, by reviewing the initial batch of references and discerning the more relevant ones. What is being searched for often does not change for periods of time, making it appropriate for machine learning technology. In practical terms, the initial screening of references would take a person an average of three minutes per reference. By contrast, a machine learning solution could process each reference within a few seconds. This approach can increase efficiency and optimise resource allocation. Team members can spend their time where they will contribute most value: on scientific analysis of the final results. This could enable earlier detection of safety information and identification of case reports.

Balancing innovation, pragmatism and objectives

Adopting AI is becoming synonymous with a company being perceived as innovative. However, to avoid embracing the wrong type of AI a pragmatic approach is called for. This should align the objective with the direction of the organisation. Current solutions include Generative AI (GenAI), Large Language Models (LLM) or Natural Language Processing (NLP). It is essential to appropriately validate the methodology, checking the AI-produced results and re-running the results to ensure repeatability. If the AI is a machine learning solution, spot checking results at regular intervals should be conducted to ensure that the program is still producing the expected results. These measures will help to protect the methodology and ensure it holds-up under scrutiny.

The future of AI in pharmacovigilance

Our industry is faced with ever increasing amounts of data and expectations from regulators. Processing this data quickly, accurately and efficiently remains critical to identify potential risks to patient safety. Investment in systems that support these increased demands, and are scalable, is essential to ensure we continue to deliver for patients. AI is a tool with the potential to address these needs, if adopted appropriately. The implementation of AI solutions is the next stage of our technology journey. Being ready to introduce the right solutions at the right time will be key to optimise success.

Learn more about how AI is solving challenges in healthcare research.

Author:

Andrew Purchase

Director, Pharmacovigilance Specialised Services & UK QPPV; Site Head, Swansea (UK); Pharmacovigilance & Safety Reporting Services

In this section

Connect with us