S3: Predictive toxicology and big data

Predictive toxicology and big data - about the challenge of knowing which dots to connect

OpenTox Asia 2018, Predictive toxicology and big data

Michael Riediker


Swiss Centre for Occupational and Environmental Health (SCOEH), Switzerland



1. Swiss Centre for Occupational and Environmental Health (SCOEH), Switzerland
2. School of Materials Science and Engineering,
Nanyang Technological University (NTU), Singapore

Big data is a term that describes the collection, management, and use of voluminous data sets of large variety that are often rapidly changing. In recent time, big data made headline news because of its use by companies and political entities that aim to predict the behavior of their target groups. However, big data is also frequently encountered in natural sciences from quantum physics over environmental exposure research to the “omics” in biosciences.

Predictive toxicology is one of the fields where a lot of data is being generated. It aims to identify how substances modulate or perturb biological pathways at a molecular level and how this translates via cellular and organ level into biological outcomes. Could big data help towards these goals? The challenge, here, is not just to get the data and to manage it. Different than in marketing, it is not sufficient to predict some fleeting trends in consumer preference. Instead one needs to provide solid, repeatable and reliable facts that can be examined by regulators and that inform about actual biological pathways. Thus, in-depth expertise needs to accompany the novel computational sciences approaches, and quantity of data should not be a replacement for quality of understanding.

Another challenge is to provide predictions that are of real-world value. Here, predictive toxicology needs to meet epidemiology and exposure sciences. Today, many epidemiological studies collect detailed metabolic and genetic data, which will inform about the variability of the target populations, while exposure sciences generate large datasets of real world exposures of consumers, workers and the general population. This data needs to be smartly combined with predictive toxicology. On one hand Big data can guide toxicological experiments to ask the right questions, on the other hand it can be used to predict the risks to humans, and thus help identify the best strategies to ensure the safety of workers, consumers and the general public.