"Unfortunately, most prediction research focuses on model development and there are relatively few external validation studies." It's tedious and you don't learn much from it (but this is my job) https://t.co/Pcu68iljRS
RT @gscollins: @modeRNepi Disappointingly, the number of events often not reported in non-ML reviews too (https://t.co/XxNBA6nBmR, https://…
@modeRNepi Disappointingly, the number of events often not reported in non-ML reviews too (https://t.co/XxNBA6nBmR, https://t.co/y722hXIAjE, https://t.co/aNrXReSQmI + many more) despite being a key item in the TRIPOD reporting statement (https://t.co/5Zbnf
@amancayork @Richard_D_Riley @laure_wynants @MaartenvSmeden Existed well before covid (e.g., https://t.co/hZTKuT51la, https://t.co/mt1nznp9Wz, https://t.co/Gcptj2Gl38, https://t.co/XxNBA6nBmR, https://t.co/aNrXReSQmI) - but amplified during covid arguably
@randyboyes @PWGTennant I can point you to review after review showing how poor GLMs are used in prediction regardless of whether ML or stats (https://t.co/aNrXReSQmI, https://t.co/26dCCXV5oT, https://t.co/hZTKuT51la, https://t.co/g7P3pUU0di, + many more),
@flavioclesio @BenVanCalster @zacharylipton Paper is highlighting methodological flaws in these ML/stat comparisons, and once you account for that, claims of superiority disappears. Plenty of reviews highlight models developed using stats also poor (https: