The potential of low-cost tests for decision support in neurodegenerative diseases

Rationale: There is a great need to detect onset of dementia as early as possible. This can help to provide treatments earlier, which may reduce costs and prolong maintenance of quality of life. For this, we need novel low-cost biomarkers that can be measured during daily living.
Methods: We concentrated on: web-based cognitive testing using the Muistikko test, computer games, and gait analysis. We compared the relationship of these measures with standard neuropsychological assessments used in clinical practice. As there is no efficient composite score for separating healthy, MCI and dementia cases, we developed a global cognitive score (GCS) composed of age, sex, MMSE, and different cognitive tests. Then, regression models were developed to estimate GCS from Muistikko- and computer-game features respectively. For the gait analysis, we studied time- and frequency features from 3D-accelerometry and performed correspondence analysis with established cognitive measures. Over 300 patients from four memory clinics in Europe (Amsterdam, Copenhagen, Kuopio and Perugia) participated in the context of the PredictND EU project.
Results: Both Muistikko test and game-based features show a good correspondence with the GCS. Additionally, gait-extracted features, such as speed, variance and regularity show correlations with different levels of cognitive impairment.
Conclusions: Low-cost measurements, that can be done easily during daily living, contain valuable information related to cognitive impairment. As such, they hold promise to give indications of risk of cognitive decline in early informal settings, thereby enrichening the existing set of tools we have at hand for decision support in dementias.