Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence

Jannicke Moe, Anders Læsø Madsen, Kristin A. Connors, Jane M. Rawlings, Scott E. Belanger, Wayne G. Landis, Raoul Wolf, Adam D. Lillicrap

Research output: Working paperResearch

Abstract

A Bayesian network was developed for predicting the acute toxicity intervals of chemical substances to fish, based on information on fish embryo toxicity (FET) in combination with other information. This model can support the use of FET data in a Weight-of-Evidence (WOE) approach for replacing the use of juvenile fish. The BN predicted correct toxicity intervals for 69%-80% of the tested substances. The model was most sensitive to components quantified by toxicity data, and least sensitive to components quantified by expert knowledge. The model is publicly available through a web interface. Further development of this model should include additional lines of evidence, refinement of the discretisation, and training with a larger dataset for weighting of the lines of evidence. A refined version of this model can be a useful tool for predicting acute fish toxicity, and a contribution to more quantitative WOE approaches for ecotoxicology and environmental assessment more generally.
Original languageEnglish
PublisherCold Spring Harbor Laboratory Press
DOIs
Publication statusSubmitted - 30 Aug 2019

Fingerprint

toxicity
fish
embryo
chemical substance
ecotoxicology
environmental assessment
fish toxicity

Cite this

Moe, J., Madsen, A. L., Connors, K. A., Rawlings, J. M., Belanger, S. E., Landis, W. G., ... Lillicrap, A. D. (2019). Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence. Cold Spring Harbor Laboratory Press. https://doi.org/10.1101/750935
Moe, Jannicke ; Madsen, Anders Læsø ; Connors, Kristin A. ; Rawlings, Jane M. ; Belanger, Scott E. ; Landis, Wayne G. ; Wolf, Raoul ; Lillicrap, Adam D. / Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence. Cold Spring Harbor Laboratory Press, 2019.
@techreport{96fd98022add444694c7b7ebc465d19f,
title = "Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence",
abstract = "A Bayesian network was developed for predicting the acute toxicity intervals of chemical substances to fish, based on information on fish embryo toxicity (FET) in combination with other information. This model can support the use of FET data in a Weight-of-Evidence (WOE) approach for replacing the use of juvenile fish. The BN predicted correct toxicity intervals for 69{\%}-80{\%} of the tested substances. The model was most sensitive to components quantified by toxicity data, and least sensitive to components quantified by expert knowledge. The model is publicly available through a web interface. Further development of this model should include additional lines of evidence, refinement of the discretisation, and training with a larger dataset for weighting of the lines of evidence. A refined version of this model can be a useful tool for predicting acute fish toxicity, and a contribution to more quantitative WOE approaches for ecotoxicology and environmental assessment more generally.",
author = "Jannicke Moe and Madsen, {Anders L{\ae}s{\o}} and Connors, {Kristin A.} and Rawlings, {Jane M.} and Belanger, {Scott E.} and Landis, {Wayne G.} and Raoul Wolf and Lillicrap, {Adam D.}",
year = "2019",
month = "8",
day = "30",
doi = "10.1101/750935",
language = "English",
publisher = "Cold Spring Harbor Laboratory Press",
type = "WorkingPaper",
institution = "Cold Spring Harbor Laboratory Press",

}

Moe, J, Madsen, AL, Connors, KA, Rawlings, JM, Belanger, SE, Landis, WG, Wolf, R & Lillicrap, AD 2019 'Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence' Cold Spring Harbor Laboratory Press. https://doi.org/10.1101/750935

Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence. / Moe, Jannicke; Madsen, Anders Læsø; Connors, Kristin A.; Rawlings, Jane M.; Belanger, Scott E.; Landis, Wayne G.; Wolf, Raoul; Lillicrap, Adam D.

Cold Spring Harbor Laboratory Press, 2019.

Research output: Working paperResearch

TY - UNPB

T1 - Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence

AU - Moe, Jannicke

AU - Madsen, Anders Læsø

AU - Connors, Kristin A.

AU - Rawlings, Jane M.

AU - Belanger, Scott E.

AU - Landis, Wayne G.

AU - Wolf, Raoul

AU - Lillicrap, Adam D.

PY - 2019/8/30

Y1 - 2019/8/30

N2 - A Bayesian network was developed for predicting the acute toxicity intervals of chemical substances to fish, based on information on fish embryo toxicity (FET) in combination with other information. This model can support the use of FET data in a Weight-of-Evidence (WOE) approach for replacing the use of juvenile fish. The BN predicted correct toxicity intervals for 69%-80% of the tested substances. The model was most sensitive to components quantified by toxicity data, and least sensitive to components quantified by expert knowledge. The model is publicly available through a web interface. Further development of this model should include additional lines of evidence, refinement of the discretisation, and training with a larger dataset for weighting of the lines of evidence. A refined version of this model can be a useful tool for predicting acute fish toxicity, and a contribution to more quantitative WOE approaches for ecotoxicology and environmental assessment more generally.

AB - A Bayesian network was developed for predicting the acute toxicity intervals of chemical substances to fish, based on information on fish embryo toxicity (FET) in combination with other information. This model can support the use of FET data in a Weight-of-Evidence (WOE) approach for replacing the use of juvenile fish. The BN predicted correct toxicity intervals for 69%-80% of the tested substances. The model was most sensitive to components quantified by toxicity data, and least sensitive to components quantified by expert knowledge. The model is publicly available through a web interface. Further development of this model should include additional lines of evidence, refinement of the discretisation, and training with a larger dataset for weighting of the lines of evidence. A refined version of this model can be a useful tool for predicting acute fish toxicity, and a contribution to more quantitative WOE approaches for ecotoxicology and environmental assessment more generally.

U2 - 10.1101/750935

DO - 10.1101/750935

M3 - Working paper

BT - Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence

PB - Cold Spring Harbor Laboratory Press

ER -

Moe J, Madsen AL, Connors KA, Rawlings JM, Belanger SE, Landis WG et al. Development of a hybrid Bayesian network model for predicting acute fish toxicity using multiple lines of evidence. Cold Spring Harbor Laboratory Press. 2019 Aug 30. https://doi.org/10.1101/750935