Forecast of the MMSE score up to 6 years ahead, with cross-cohort replications

Etienne Maheux (1, 2) , Igor Koval (1, 2) , Damiano Archetti (3) , Alberto Redolfi (3) , Stanley Durrleman​ (1, 2)

1 ARAMIS - Algorithms, models and methods for images and signals of the human brain = Algorithmes, modèles et méthodes pour les images et les signaux du cerveau humain [ICM Paris]
2 ICM - Institut du Cerveau = Paris Brain Institute
3 IRCCS - Centro San Giovanni di Dio, Fatebenefratelli, Brescia

Most of the studies that focus on the Alzheimer’s disease (AD) prediction forecast a dementia label: cognitively normal, mild cognitive impairments or AD. However, these labels cannot always be compared across cohorts, they are not perfectly accurate and they describe the disease thanks to a limited number of stages (only 3) contrary to what is, in practice, a continuous progression. For these reasons, we predict the Mini-Mental State Examination (MMSE), a cognitive test whose scale presents 30 discrete values. Predicting this measure, assessed in multiple AD-related cohorts, allows better staging of the disease progression at the individual level, in particular in the ADNI, AIBL and PharmaCOG cohorts.

Mots clés: Cognitive decline forecast; Personalized prediction; Longitudinal study; Cross-cohort replication

Domaines: Bio-informatique [q-bio.QM]; Machine Learning [stat.ML]; Statistiques [math.ST]; Neurosciences

Etienne Maheux, Igor Koval, Damiano Archetti, Alberto Redolfi, Stanley Durrleman​. Forecast of the MMSE score up to 6 years ahead, with cross-cohort replications. CompAge 2020 - Computational approaches for ageing and age-related diseases, Sep 2020, Paris / Virtual, France. ⟨hal-03137878⟩