‘Timed Up-and-Go Dual-Task’ gait parameters from video recordings as an aid for early diagnosis of dementia disorder

This is a continuation of a VR-funded study in which we investigate if a mobility-test (TUG) combined with a verbal task (TUGdt) can support early identification of dementia.

We aim at further exploration of the TUGdt measures by extraction and analyses of gait parameters based on video-recorded tests.

Hence, we develop a program for valid and reliable semi-automatic gait-parameter extraction to investigate which TUGdt gait parameters best predict cognitive dementia over 2-, 4- and 8-year periods.

We will use previous TUG and TUGdt data collected at memory clinics, i.e. baseline data on 298 patients undergoing memory assessment.

Subjects without manifest dementia at baseline (n=212) are followed by re-assessments 2-years and 4-years post baseline and by medical records reviews 2, 4 and 8-years post baseline. Additional data on cognitively healthy controls (n=166) will be used for comparisons between groups. Reliability will be estimated for patients and healthy controls using repeated assessments data.

Statistical analysis will include logistic regression models or Cox regression models with and without adjustment for gender, age and education.

Sensitivity, specificity, area under the receiving operating characteristics curve (c-statistics), will be estimated.The results will determine the validity of the TUGdt-test for early identification of dementia disorder, which will bring new knowledge to the research area and may also provide a basis for clinical implementation.