Supplementary Fig. 1. Statistical Classification Analysis Results. Box and whisker plots displaying mean performance metrics returned in the assessment of in vitro AC 50 values and dosimetry-adjusted oral equivalent values in the prediction of the 40 discretized in vivo toxicity endpoints. Metrics assessed: AUC, balanced accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Eighty-four SAS statistical classification models were used.* Detailed descriptions and parameters of the SAS models are provided in Supplementary Table 8; full names for the in vivo endpoints are provided in Supplementary Table 7. * RatCast_OE endpoints CHR_Rat_SpleenPathology and MGR_Rat_LitterSize only include 83 models in the CVMC (both are missing Discriminant Analysis model #7 as it produced floating point errors which caused the CVMC to fail).
In Vitro Oral Equivalent Dose – Discretized In Vivo Response Data (Yes/No) In Vitro Nominal AC 50 and LEC Concentrations – Discretized In Vivo Response Data (Yes/No)
In Vitro Oral Equivalent Dose – Discretized In Vivo Response Data (Yes/No) In Vitro Nominal AC 50 and LEC Concentrations – Discretized In Vivo Response Data (Yes/No)
In Vitro Oral Equivalent Dose – Discretized In Vivo Response Data (Yes/No)
In Vitro Nominal AC 50 and LEC Concentrations – Discretized In Vivo Response Data (Yes/No)
In Vitro Oral Equivalent Dose – Discretized In Vivo Response Data (Yes/No) In Vitro Nominal AC 50 and LEC Concentrations – Discretized In Vivo Response Data (Yes/No)
In Vitro Oral Equivalent Dose – Discretized In Vivo Response Data (Yes/No) In Vitro Nominal AC 50 and LEC Concentrations – Discretized In Vivo Response Data (Yes/No)