Every medicine we rely on today, from painkillers to cancer drugs, has gone through a long and complex development process. Before a drug ever reaches patients, it must be carefully tested to check if it is safe and effective. Today, much of this testing is still done in animals. While studies using animals can provide important information, they are also ethically sensitive and scientifically challenging. A striking number highlights this challenge: 5% make it to human patients. The remaining 19 drop out somewhere along the way.

Why drug development is so complex

The path from a scientific idea to an approved medicine is long. On average it takes 5 years to reach a first human study, 7 years to reach a clinical trial and 10 years to reach approval. Many factors influence whether a drug continues to human trials, including how studies are designed and carried out. Small choices, such as how many animals are included, which measurements are taken, or how the models are handled, can affect how reliable or comparable the results are. Some of these factors are clear, while others are still unknown. And because the process is complicated, no single study can reveal what makes a drug successful.

The STRIDE Lab’s approach: learning from millions of studies

This is where the STRIDE Lab at the University of Bern comes in. Led by Benjamin Ineichen, the team uses systematic reviews (a structured way of collecting and analysing all the scientific studies on a particular question, so we can see the full picture instead of relying on just a few papers), statistics and data science, such as large language models, to analyse many scientific papers, sometimes thousands or hundreds of thousands at a time.

Instead of manually reading a handful of studies, they build large datasets that bring all available evidence together. With this overview, they might identify patterns that are invisible in individual studies, such as:

• which experimental designs lead to more reliable results
• whether testing a drug in multiple species increases the chance it will work in humans
• which research practices are linked to success or failure

Their work has already shown that several research areas suffer from low rigour. For example, inadequate dosing of animal studies not representing what would be given to human patients.

How this work supports the 3Rs

Replacement

Systematic reviews help highlight where non-animal approaches, such as organoids, computational models, or human-based methods, are already providing strong evidence. This opens space for replacing some animal studies altogether.

Reduction

Systematic review allows researchers to learn from existing data rather than starting from scratch. This can reduce the number of new animal experiments required and help avoid unnecessary or less well designed studies.

Refinement

By identifying study designs and animal models that are most informative, researchers can avoid experiments unlikely to provide useful insights. This can reduce the severity of procedures and improve animal welfare.

A more responsible and efficient drug development pipeline

When we learn which research practices truly lead to better treatments, we can focus resources where they matter most, reducing avoidable animal use while strengthening the science that ultimately helps patients. Better evidence means better decisions, for humans and for animals.