In:
Journal of Applied Toxicology, Wiley, Vol. 35, No. 11 ( 2015-11), p. 1361-1371
Abstract:
Supervised learning methods promise to improve integrated testing strategies. We curated a skin sensitization data set and evaluated the information from in silico , in chemico and in vitro assays by recursive variable selection. Cell assays consistently ranked high in the feature elimination. A hidden Markov model taking advantage of monotonicity between local lymph node assay and dose is shown to outperform dose‐naïve models on all data sets and strongly reduce extreme sensitizers classified as non‐sensitizers.
Type of Medium:
Online Resource
ISSN:
0260-437X
,
1099-1263
Language:
English
Publisher:
Wiley
Publication Date:
2015
detail.hit.zdb_id:
1475015-6