A Study to Evaluate the Ability of Speech- and Language-based Digital Biomarkers to Detect and Characterise Prodromal and Preclinical Alzheimer's Disease in a Clinical Setting - FUTURE Extension Study.
Overview
- Phase
- Not Applicable
- Intervention
- Not specified
- Conditions
- Alzheimer Disease
- Sponsor
- Novoic Limited
- Enrollment
- 67
- Locations
- 1
- Primary Endpoint
- The agreement between the change in the PACC5 composite between baseline and +12 months and the corresponding regression model, trained on baseline speech data, predicting in all four Arms, as measured by the coefficient of individual agreement (CIA).
- Status
- Terminated
- Last Updated
- last year
Overview
Brief Summary
The primary objective of the study is to evaluate whether a set of algorithms analysing acoustic and linguistic patterns of speech, can predict change in PACC5 between baseline and +12 month follow up across all four Arms, as measured by the coefficient of individual agreement (CIA) between the change in PACC5 and the corresponding regression model, trained on baseline speech data to predict it. Secondary objectives include (1) evaluating whether similar algorithms can predict change in PACC5 between baseline and +12 month follow up in the cognitively normal (CN) and MCI populations separately; (2) evaluating whether similar algorithms trained to regress against PACC5 scores at baseline, still regress significantly against PACC5 scores at +12 month follow-up, as measured by the coefficient of individual agreement (CIA) between the PACC5 composite at +12 months and the regression model, trained on baseline speech data to predict PACC5 scores at baseline; (3) evaluating whether similar algorithms can classify converters vs non-converters in the cognitively normal Arms (Arm 3 + 4), and fast vs slow decliners in the MCI Arms (Arm 1 + 2), as measured by the AUC, sensitivity, specificity and Cohen's kappa of the corresponding binary classifiers. Secondary objectives include the objectives above, but using time points of +24 months and +36 months; and finally to evaluate whether the model performance for the objectives and outcomes above improved if the model has access to speech data at 1 week, 1 month, and 3 month timepoints.
Investigators
Eligibility Criteria
Inclusion Criteria
- •Subjects are fully eligible for and have completed the AMYPRED (Amyloid Prediction in early stage Alzheimer's disease from acoustic and linguistic patterns of speech) study.
- •(See https://clinicaltrials.gov/ct2/show/NCT04828122)
- •Subject consents to take part in FUTURE extension study.
Exclusion Criteria
- •Subject hasn't completed the full visit day in the AMYPRED study.
Outcomes
Primary Outcomes
The agreement between the change in the PACC5 composite between baseline and +12 months and the corresponding regression model, trained on baseline speech data, predicting in all four Arms, as measured by the coefficient of individual agreement (CIA).
Time Frame: 1 year
Baseline speech data is speech data collected during the first 14 days of the study for each participant.
Secondary Outcomes
- The agreement between the changes in the PACC5 composite between baseline and +12, +24, +36 months and the corresponding regression models, trained on baseline speech data, to predict them in the CN Arms (3 and 4), as measured by the CIA.(3 years)
- The agreement between the changes in the PACC5 composite between baseline and +12, +24, +36 months and the corresponding regression models, trained on baseline speech data, predicting them in the MCI Arms (Arms 1 and 2), as measured by the CIA.(3 years)
- The AUC, sensitivity, specificity and Cohen's kappa of the binary classifier distinguishing between converters vs non-converters in the cognitively normal (CN) Arms (Arms 3 and 4).(3 years)
- The outcomes above where the model has access to speech data at 1 month, 3 month time points, in addition to baseline speech data.(3 years)
- The agreement between the PACC5 composite at +12, +24, +36 months and the corresponding regression models, trained on baseline speech data and +12, +24, +36 month speech data, as measured by the coefficient of individual agreement (CIA).(3 years)
- The agreement between the PACC5 composite at +12, +24, +36 months and the corresponding regression models, trained on baseline speech data, predicting in all four Arms based on +12, +24, +36 month speech data, as measured by the CIA.(3 years)
- The AUC, sensitivity, specificity and Cohen's kappa of the binary classifier distinguishing between fast decliners vs slow decliners in the MCI Arms (Arms 1 and 2) at +12, +24, +36 months.(3 years)
- The agreement between changes in the PACC5 composite between baseline and +24, +36 months and the corresponding regression models, trained on baseline speech data, predicting in all 4 Arms, as measured by the coefficient of individual agreement (CIA).(3 years)