ResearchBenchmarkseGFR trajectory forecasting on CKD case study (val)Follow7.31MAECMWM7.282527.468017.65357.83899May 21, 2026Evaluation ResultsMethodMethodLinksMAERMSERelative MAE ChangeRelative RMSE ChangeCMWMProtocol=Dynamic-50% h...Protocol=Dynamic-50% history rollout, Sample size (n)=1,020, Trainable parameters=0.79 M2026.057.319.8518.5911.48GPT-5.5Protocol=Dynamic-50% h...Protocol=Dynamic-50% history rollout, Sample size (n)=1,0202026.057.99711.128——