OPRA is the ONTOX AI-supported Probabilistic Risk Assessment approach currently being developed to support next-generation, non-animal chemical safety assessment. It is designed to integrate exposure assessment, biokinetics, hazard and potency information, ontologies, read-across, NAM-derived evidence, and AI-supported tools into a structured, uncertainty-aware risk assessment workflow.

As an emerging ONTOX concept, OPRA aims to support transparent decision-making by helping users understand not only whether a chemical may pose a concern, but also the level of uncertainty and probability associated with that assessment. The approach is being refined through case studies, stakeholder input, and alignment with broader European efforts to advance NAM-based and AI-supported risk assessment.

The OPRA concept has been introduced in public ONTOX and ASPIS materials, including the 2025 ASPIS Open Symposium presentation on the ONTOX AI-supported probabilistic risk assessment approach and the recent ECETOC–VHP4Safety–ONTOX workshop on AI-enabled digital infrastructures.

While OPRA is still under development and not yet the subject of a dedicated peer-reviewed publication, it builds on the growing peer-reviewed scientific literature on probabilistic risk assessment in toxicology. OPRA concept is explored in several case studies, including a PFOA case study, which is planned to be finalised by October 2026 and submitted to OECD. 

Relevant ONTOX-related literature: 

Gao Y, Han K, Mughal Z, Nagyal M, Mikhail J, Fakhrzadeh L, Hartung T and Luechtefeld T. ToxTransformer – a conditional causal transformer for large-scale toxicity prediction. In preparation.

Hartung T, Luechtefeld T and Maertens A. The role of AI in overcoming animal testing. In: Van Norman G and Krebs C. Moving Toward Nonanimal Approaches in Medical Research & Testing. Academic Press (Elsevier). 2027, Cambridge, MA, USA. pp. 143-158. DOI: 10.1016/b978-0-443-26514-3.00017-5.

Hartung T. TwAIns: how artificial intelligence transforms biological digital twins from cell to population, and why validation is the limiting factor. Frontiers in Artificial Intelligence, revised.

Hartung T, Kopańska K, Maertens A, Whaley P and Hoffmann S. Causality analysis of toxicological mechanisms in networked systems such as Adverse Outcome Pathway networks. Archives of Toxicology, 2026, in press. Doi: 10.1007/s00204-026-04386-z

Kopańska K, Leist M and Hartung T. Uncertainty in toxicological assessments in the era of artificial intelligence. In preparation

Kopańska K and Hartung T. Next Generation Validation for Next Generation Risk Assessment. Frontiers in Toxicology 2026, 8:1790669. doi: 10.3389/ftox.2026.1790669.

Kopańska K, Rovida C, Antignac E, Barton-Maclaren T, Corgreave I, Escher SE, Herzler M, Hollnagel HM, Kuchovská E, Maertens A, Malinowska JM, Patlewicz G and Hartung T. From thresholds to odds: the role of thresholds of toxicological concern (TTC) in modern risk assessment. In preparation.

Luechtefeld T and Hartung T. Evidence-based AI: from trailblazer to trustblazer? Front. Artif. Intell. 2026, 9:1818128. doi: 10.3389/frai.2026.1818128.

Maertens A, Golden E, Luechtefeld TH, Hoffmann S, Tsaioun K, Hartung T. Probabilistic risk assessment – the keystone for the future of toxicology. ALTEX. 2022;39(1):3-29. doi: 10.14573/altex.2201081. PMID: 35034131; PMCID: PMC8906258. 

Maertens A, Antignac E, Benfenati E, Bloch D, Fritsche E, Hoffmann S, Jaworska J, Loizou G, McNally K, Piechota P, Roggen EL, Teunis M, Hartung T. The probable future of toxicology – probabilistic risk assessment. ALTEX. 2024;41(2):273-281. doi: 10.14573/altex.2310301. Epub 2024 Jan 12. PMID: 38215352.