Fig. 2
From: Assessing stroke severity using electronic health record data: a machine learning approach
![Fig. 2](http://media.springernature.com/full/springer-static/image/art%3A10.1186%2Fs12911-019-1010-x/MediaObjects/12911_2019_1010_Fig2_HTML.png)
NIHSS Distribution. Distribution of (a) real NLP-extracted NIHSS scores and (b) model-imputed NIHSS scores for the hold-out test set (n = 1033). Model-imputed scores were rounded down to nearest integer