Using Data Science To Reduce Animal Testing

I've continued a productive collaboration with Bayer's Thomas Steger-Hartmann studying at ways to reduce animal testing. Our recent letter in Toxicological Pathology as a follow up to our webinar for the International Academy for Toxicological Pathology.

• Can Historical Control Group Data Be Used to Replace Concurrent Controls in Animal Studies?

The idea of virtual control groups will eventually be the standard practice. The need will drive adoption of FAIR data principles (Findable, Accessible, Interoperable, Reuseable) as the recording of the details of studies becomes more standardized.

I'd like to point out a 2023 PhD dissertation from Peter Wright which has has an excellent analysis of this topic.

Analysis of Virtual Control Groups and Inter-species Concordance using Preclinical Toxicity Data

Dr. Wright also dicusses one of my favorite topics - understanding how well animal studies predict human response. It includes inter-species results comparison which is highly relevant for understanding the impact and predictive ability of animal studies. That is another route to reduce animal usage - refining studies to do only those demonstrated to have predictive value, and discontinuing those that can be shown not to have human predictive value.

The concept is also gaining traction for human studies driven by the desire for better ways to evaluate real-world effects (RWE) and understand how to best predict human response to drugs.

• Virtual controls as an alternative to randomized controlled trials for assessing efficacy of interventions
• Virtual control groups in pharmaceutical development – What are they and can they be trusted?In addition to using historic control data, there is also work for creating statistical models for clinical trials - and the methods and evaluation can guide animal studies.
• The use of a predictive statistical model to make a virtual control arm for a clinical trial