ALTOIDA‐iADL for the diagnosis of Mild Cognitive Impairment and early Alzheimer's disease

Abstract

Background Diagnostic accuracy for the early detection of mild cognitive impairment (MCI) is critical both in the clinical and research settings. Our aim was to evaluate the diagnostic performance of the ALTOIDA‐iADL test in subjects with non‐degenerative MCI and prodromal (pAD) and mild (mAD) Alzheimer's disease. Methods ALTOIDA‐iADL is a 10‐minute administrable cognitive test, which assesses activities of daily living in the form of an augmented virtual reality game. The task consists of placing and finding virtual objects in a real environment and provides a final score (the NeuroMotor Index; NMI). The NMI is obtained by weighting multi‐modal information such as hands’ micromovements, walking bouts and speed, reaction time and navigation trajectory (among others), and represents the overall outcome of the individual task performance. Fifty‐one participants were included and classified according to cerebrospinal fluid (CSF) AD biomarkers: MCI (n = 22; age: 68.2; MMSE: 26.5), pAD (n = 15; age: 69.4; MMSE: 24.0) and mAD (n = 14; age: 70.6; MMSE: 20.7). Results The NMI allowed differentiating between subjects with absence (Aβ‐) and presence (Aβ+) of abnormalities in the amyloid biomarker (p < 0.01). Also, differences were found between the MCI group and the pAD (p < 0.01) and mAD (p < 0.01) groups (Fig. 1). ROC curves showed good diagnostic accuracy of the NMI in the discrimination between the Aβ‐ and Aβ+ (AUC = 0.777; p < 0.01), MCI and pAD (AUC = 0.781; p < 0.01) and MCI and mAD (AUC = 0.772; p < 0.01). The NMI did not discriminate between the pAD and mAD (AUC = 0.557; p = 0.61) groups (Fig. 2). The NMI correlated with CSF NfL levels (r = ‐.456; p < 0.05) and the MMSE score (r = .432; p < 0.01), showing an association with the degree of cognitive impairment. Conclusions ALTOIDA‐iADL is useful in the differential diagnosis between patients with non‐degenerative MCI and prodromal and mild Alzheimer's disease. Its performance is related to the degree of impairment in cognitive screening tests and with biomarkers of axonal damage/neurodegeneration.

Machine Learning and Digital Biomarkers Can Detect Early Stages of Neurodegenerative Diseases

“…The necessity of a good user experience in cognitive screening tools is crucial, not only for user engagement but also for the accuracy and reliability of the data collected. One notable project in this domain is the Altoida initiative, which focuses on identifying digital biomarkers for mild cognitive impairment [ 79 ].…”