FB2023 Webbanner 1920X1080 INEB
Data enablement and digital business models

Let AI help you manage your drug safety workload

With Ingelise Bærentzen Melchiorsen, Head of Drug Safety, NNIT

Reports of adverse events are increasing rapidly, placing a heavy burden on pharmacovigilance teams in the life sciences industry. Ingelise Bærentzen Melchiorsen, Head of Drug Safety at NNIT, explains how advanced digital tools can help you process the growing influx of safety data to provide clarity and extract more value.

Efficient and reliable capture of pharmacovigilance (PV) data is the core of drug safety, but many life sciences organizations face increasing challenges keeping up with a rapidly growing volume of adverse-event reports. The rapid development of COVID-19 vaccines, for instance, ignited greater public interest in drug safety and the risk/reward relationship between medicinal benefits and possible adverse effects.

In addition, complex products, stricter regulation and more avenues for reporting adverse effects, such as social media, are also contributing to the increased workload for PV teams.

– We’re seeing an exponential increase in drug safety cases across the market. This puts a significant resource strain on pharma companies, forcing them to find new ways to cope with the increased workload, says Ingelise Bærentzen Melchiorsen, Head of Drug Safety at NNIT.

Some of the options available to pharma companies include outsourcing parts of their case processing and introducing more automation to their PV processes. In either case, digitalization can help with PV by reducing cycle times, improving quality and accuracy, handling case volume growth and diverse data formats, minimizing manual steps, and enhancing compliance.

Manage your pharmacovigilance inbox with AI

Efficient management of the pharmacovigilance inbox, where the PV team receives adverse-event reports from health professionals, patients, regulatory authorities and other sources, is central to keeping the drug safety workload under control.

Since most of the data submitted to the inbox must be reviewed and reported to the health authorities within a specific timeframe, it is essential to be able to sort and categorize the submissions by type and relevance. Here, artificial intelligence (AI) can do a lot of the heavy lifting.

– Using AI, we’re able to filter much of the content as soon as it arrives in the PV inbox. This makes it easier to ensure that each case is forwarded to the relevant team for further processing. You can also produce automated forecasts about how many cases each team can expect, allowing for more efficient scheduling to ensure that there are sufficient resources to handle the cases within the assigned deadline, says Ingelise Bærentzen Melchiorsen.

When designing AI algorithms for this type of process, it is important to know how to validate the system and ensure that GxP compliance is maintained at all times. For example, while AI may be used to analyze videos and images submitted by doctors and apply a layer of meta-data to enable structuring and searchability, it is important to preserve the original data unaltered. In addition, review procedures can be designed to ensure that a human always signs off on the work done by the AI algorithms.

Signal detection on new platforms

The PV inbox is not the only source of safety data. Today, pharma companies can pick up safety-related information through a variety of different channels and platforms, including social media, meetings with health professionals, e-mails, market research and patient support programs.

– Reports of adverse events and other relevant safety data can come from all sorts of interactions with patients and health professionals. For example, a company might operate a hotline for smokers who want to quit. If someone calls in and mentions that the nicotine patches they use don’t have any effect, that is a relevant piece of data that needs to be recorded, says Ingelise Bærentzen Melchiorsen and continues:

– This means that you must have sufficient structure and governance in place to filter out the noise and ensure that relevant safety information is recorded, formatted correctly and submitted to the right place. Ideally, you want to automate as much as possible and implement measures that turn unstructured data into structured data, such as checklists and pre-defined fields in the customer support system.

Build the foundation for safety data enablement

AI, data lakes and better structure do not automatically create additional value, but such measures provide a solid foundation for data enablement by making safety data easier to manage, more accessible and more transparent.

– The future is bright when it comes to improving the value of safety data. We have many new tools to help us structure, process and share data. We can see more of the iceberg, which allows us to spot problems and opportunities we can act on. Does our customer support staff need better training? Should we train more doctors in the use of our products? Is a particular site dosing their batches incorrectly? There’s a lot of potential, says Ingelise Bærentzen Melchiorsen.

Want to know more about how your organization can work with data enablement, digital business models and processes, and new agendas?

Read the other articles where NNIT experts share their concrete experience and advice for different aspects of the life sciences industry.

Do you want us to help you get more out of your safety data?

If you have been inspired to enhance your safety data, NNIT has the industry insights and technical expertise to advise and assist you. Let us help you identify new opportunities and implement the digital solutions and organizational change management needed to realize the full potential of your data.