Project facts

  • Location
  • Period
  • Challenge
    Accelerating responses to adverse pharmaceutical events, and improving the consistency of service for affected customers.
  • Solution
    A text-based AI solution that interprets the risk and severity associated with reported events, and provides automated decision support.


  • Improved consistency of customer service responses

  • Streamlined compliance with global adverse event reporting regulations

  • Accelerated adverse event reporting and response times

The challenge

Improving response consistency with AI

With 104,000 employees, our client is one of the biggest animal health providers in Germany, catering to tens of thousands of customers around the world. When you’re providing so many products, adverse events and reactions are an unfortunate fact of life. In many cases, they can’t be prevented. But, by responding quickly, and in the right way, a pharmaceutical provider can still make a positive difference.

On top of that, there are also diverse regulatory requirements that dictate how adverse events must be reported and responded to in each country – making tracking them consistently across geographies extremely challenging.

When our client came to Royal HaskoningDHV Digital, they wanted to find a new, intelligent way of improving the consistency of adverse event responses, to help meet rising global demand for its products, support further growth, and ensure the best outcomes and experiences for buyers.

The approach

Supporting rapid decision-making

Our first task was to meet with the client’s domain experts and map out their current adverse event handling processes to identify areas where automation and decision support could be applied. It turned out that these processes required many manual search actions and expert judgements to be made. A time-consuming process.

From there, we worked with the team to gather relevant data – including historical event cases and medical industry protocols – to effectively train and validate new automation and decision support models.

The majority of this data was text-based, so we created a solution that would also generate text-based outputs. The model was trained to recognise and extract key information from new events, such as described symptoms, and then intelligently interpret the level of risk involved.

By analysing the case details provided by the customer against the historical data, medical documentation and clinical protocol data used for training, the model can help response teams better understand the severity of each case – and respond appropriately.

That analysis and insight is also flowed into answer templates, which are automatically generated to be sent back to the client, and any relevant local authorities, as quickly as possible.

The impact

Faster responses and streamlined compliance

Today, with our solution in place, the client’s global customer service teams can handle adverse events as quickly and professionally as possible.

AI-based decision support empowers case handlers to make the right choices and provide the correct responses at every stage of the event handling process. Plus, the whole process can be completed in a far shorter time.

From a compliance point of view, it enables the client to ensure that all global adverse events are responded to in a consistent manner, while still meeting local reporting regulations.

Plus, in the event that the model uncovers a serious adverse event, it can quickly be routed to the right experts – ensuring that all potential negative outcomes are minimised, for both customers and the company itself.

To make our models usable for the adverse event handlers, we created a UI around the models that runs in parallel with their administrative systems.