“…25) These parameter values were compared with the traditional in vivo-derived input parameters (Figure 1). When our previous machine-learning system 25) (in which Fa was derived from apparent permeabilities estimated from 17 chemical descriptors 21) using LightGBM, ka was derived from 65 chemical descriptors 25) using ridge regression, V1 was derived from 64 chemical descriptors 25) using ridge regression, and CLh,int was derived from 17 chemical descriptors 25) using LightGBM) was applied to the expanded panel of 355 chemicals, the correlation coefficients of the resulting Fa•Fg, ka, V1, and CLh,int values, respectively, were 0.28 (p < 0.01, Figure 1A), 0.36 (p < 0.01, 1B), 0.56 (p < 0.01, Figure 1C), and 0.85 (p < 0.01, Figure 1D). Moreover, the average absolute fold errors for Fa•Fg, ka, V1, and CLh,int were 2.02, 2.32, 2.18, and 2.56, respectively, for the previous estimation systems.…”