LAUSR

Key Points Question What items are collectively requested by model reporting guidelines and does documentation for deployed clinical predictive models report this information? Findings This systematic review combined 15 model reporting guidelines and identified 220 distinct requested items. A review of the documentation of 12 deployed models from a single vendor found that the median item completion rate was 39%, and although commonly requested items were highly reported, at least half of the documentation could have provided more information on reliability (including external validation) and fairness. Meaning These findings suggest that items collectively requested by guidelines represent a substantial reporting burden, and additions regarding reliability and fairness may improve their documentation.