Fig. 7

Histogram comparisons of the absolute errors for the 2020 a mortality, b disability, and c dispensing rate predictions between the fully trained filter and the filter trained on progressively less data. The error distribution of the fully trained filter is colored in blue, whereas the error distribution of the filter trained on less data is colored in red. The top row showcases comparisons for the filter trained on four years of data, iteratively progressing downwards to the last row showcasing comparisons for the filter trained on a single year of data. Each histogram features a red arrow highlighting the maximum error from the filter trained on less data. Figures are best viewed online in color