Table 3.

– Suggested ways forward to increase quantitative information on AOPs.

Activity	Description	Notes, examples
Development of a general qAOP modelling framework	A harmonized approach for regulators and scientists that would facilitate the qAOPs modelling. A framework should ideally allow for natural integration with (physiologically-based) pharmacokinetic models.	A framework for qAOP development was proposed and three case studies conducted (Paini et al., 2022).
Prioritization of current qualitative AOPs for further qAOP development	Pragmatic prioritization considering (1) the foreseen regulatory application domain (e.g., potency ranking vs quantitative hazard characterization for risk assessment), (2) the existence of established methods for the MIE/KEs, and (3) the expected time lapse between exposure and health effect.	The design of qAOPs may be complicated for endpoints where the adverse outcome only occurs after years of chronic exposure.
Concerted action through crowdsourcing and promoting the contribution to smaller units (e.g. quantitative KERs)	Stimulation of concerted activities on smaller parts of quantitative AOPs (quantitative KERs) to facilitate larger interactions and a more rapid generation of quantitative information.	A larger interlaboratory variation could be a limitation (Svingen et al., 2021).
Develop in silico extrapolation methods between assays using toxicokinetic models	Account for the toxicokinetic differences between species, assays and level of biological organization.
Establish standardized approach for omics data	Harmonise and standardize the approaches for interpreting and quantitatively connecting omics data (e.g., gene expression and signalling pathways) to a phenotypic outcome.	An example is the Signalling Pathways Project for discovering consensomes, i.e. downstream genomic targets of signalling pathway nodes (receptors, enzymes, transcription factors and co-nodes) and cognate bioactive small molecules (Ochsner et al., 2019).
Flexible approach in qAOP development with respect to the available data and modelling tools	Consider other models if dose-response models are not applicable (do not take into account the dynamics of a system).	Hybrid approaches to properly quantify the KERs, i.e., combining different types of equations may be of value.