Prediction of bioconcentration factors in fish and invertebrates using machine learning

Abstract

The application of machine learning has recently gained interest from ecotoxicological fields for its ability to model and predict chemical and/or biological processes, such as the prediction of bioconcentration. However, comparison of different models and the prediction of bioconcentration in invertebrates has not been previously evaluated. A comparison of 24 linear and machine learning models is presented herein for the prediction of bioconcentration in fish and important factors that influenced accumulation identified. R and root mean square error (RMSE) for the test data (n = 110 cases) ranged from 0.23-0.73 and 0.34-1.20, respectively. Model performance was critically assessed with neural networks and tree-based learners showing the best performance. An optimised 4-layer multi-layer perceptron (14 descriptors) was selected for further testing. The model was applied for cross-species prediction of bioconcentration in a freshwater invertebrate, Gammarus pulex. The model for G. pulex showed good performance with R of 0.99 and 0.93 for the verification and test data, respectively. Important molecular descriptors determined to influence bioconcentration were molecular mass (MW), octanol-water distribution coefficient (logD), topological polar surface area (TPSA) and number of nitrogen atoms (nN) among others. Modelling of hazard criteria such as PBT, showed potential to replace the need for animal testing. However, the use of machine learning models in the regulatory context has been minimal to date and is critically discussed herein. The movement away from experimental estimations of accumulation to in silico modelling would enable rapid prioritisation of contaminants that may pose a risk to environmental health and the food chain.

Biomonitoring of pesticides, pharmaceuticals and illicit drugs in a freshwater invertebrate to estimate toxic or effect pressure

“…For pesticides, available EC 50 values (48 h acute in Daphnia magna ) available from the Pesticide Properties Database (Hertfordshire, 2019). The BCFs were estimated from both EPI Suite BCFBAF v3.02 (Agency, U.S.E.P., 2019) software and our own previously developed artificial neural network (ANN) for prediction of BCFs in G. pulex (Miller et al, 2019) (Fig. 2).…”

“…Samples were collected from 15 sites covering five river catchments and used to estimate toxic/effect pressure. Internalised concentrations determined herein and a previously developed model for prediction of bioconcentration factors in G. pulex ( Miller et al, 2019) along with the well-established EPISuite (Agency, U.S.E.P., 2019) BCF predictions in fish were used to calculate internal toxic units (TU int ) and effect units (EU int ) for pesticides and pharmaceuticals, respectively.…”

Exploiting machine learning for end-to-end drug discovery and development

“…Environmental impact 90 Pharmacoeconomics / cost effectiveness analysis / policy decisions 91 Clinical trial: recruiting, design, optimization, success and failure 6…”

Machine Learning for Environmental Toxicology: A Call for Integration and Innovation

“…In a recent study, ML outperformed animal testing approaches in chemical safety assessments [6]. In our work, we used ML to predict bioconcentration in aquatic fauna as part of persistent, bioaccumulative and toxicity (PBT) assessments [7]. However, there is a critical lack of literature concerning ML development for environmental exposure and effect assessment.…”