Accurately predicting which individuals with Mild Cognitive Impairment (MCI) will progress to dementia is a major clinical challenge. This study investigated whether baseline cognitive screeners (CS) and magnetic resonance imaging measures rated with visual rating scales (VRS) can meaningfully predict progression to dementia. Additionally, it aimed to determine which combination of these tools yields the most efficient prognostic model. In a longitudinal cohort of 113 MCI patients, Firth-corrected Cox proportional hazards regression models were used to estimate hazard ratios for baseline performance on CS (7-Minute Screen [7MS], Frontal Assessment Battery [FAB], and Montreal Cognitive Assessment [MoCA]) and VRS (Fazekas scale, Global Cortical Atrophy scale [GCA], and Right Medial Temporal Atrophy scale [MTAR]) in relation to time-to-conversion to dementia. The Akaike Information Criterion (AIC) was employed for model selection to identify the most informative set of predictors. Results indicated significant associations for the 7MS, FAB, MoCA, GCA, and MTAR with dementia progression; however, the Fazekas scale's utility was limited. The combination of the 7MS, FAB, and GCA yielded the most efficient predictive model. While FAB and GCA relevance was anticipated, 7MS's predictive utility proved notably strong. The 7MS, FAB, MoCA, GCA, and MTAR each demonstrate independent predictive value for MCI-to-dementia conversion. Crucially, a combined assessment model of the 7MS, GCA, and FAB offers a clinically feasible strategy for identifying high-risk individuals. These findings highlight the importance of integrating cognitive and structural imaging for improved early prognostication in memory clinic settings.

Keywords: Mild Cognitive Impairment, dementia, cognitive screeners, visual rating scales, cox proportional hazard model, Akaike Information Criterion