WP7 (final fatigue/sleep endpoints) together with WP4 (deriving intermediate predictors) is responsible for the analysis of the extant datasets as well as the data from the FS and the CVS. It will perform the analysis of the extant datasets, analysis of the feasibility study (FS), analysis of the clinical validation study (CVS), statistics and longitudinal disease modelling, ML/AI approaches. WP7 will be also in charge of designing the statistical plan for the clinical studies.
Clinical study design. A sample size estimation based on the variability observed during the FS will be conducted to help specify the CVS sample size and design. If feasible, longitudinal models to compute virtual clinical trials will also be used. Population statistical model. Mixed effect population as well as semi-parametric models will be used for the analysis of the CVS. They will characterize and predict the average population response and inter-individual variabilities based on the longitudinal data. Diagnostic and predictive models will be implemented to describe the temporal dynamics of fatigue, sleep and selected ADL for the overall population. ML/AI approaches.
EFPIA support: EFPIA partners will assist in identifying literature on relevant ADL/HRQoL parameters of interest and, if possible, provide internal data allowing the design of a first series of mixed effect population models based on the recorded observations of digital endpoints. EFPIA partners will also provide execution and regulatory constraints for the clinical studies (enrolment, cost, timeline, etc.) and comment on desired statistical power. Overall input on the statistical aspects of clinical study design will be provided.
Objectives
O7.1 Variability assessment of measures produced by selected devices during feasibility study.
O7.2 Sample size study and design specification for the Clinical Validation study.
O7.3 Statistical population modelling for longitudinal observations of digital endpoints
O7.4 Develop non-device-specific AI/ML and modelling methods for fatigue, sleep disturbances and other ADL/HRQoL based on the multimodal data sources
O7.5 Prepare Data Analytic Packages of digital endpoints for fatigue, sleep and a selected ADL for submission to EMA ± FDA for scientific advice
Deliverables
D7.1 Variability assessment of data collected during FS
D7.2 Sample size and design study for CVS
D7.3 An AI methods toolbox for robust multi-variate time-series analysis of personal data and ML methods for heterogeneous data sources aggregation
D7.4 Library of population models with selected digital endpoints as longitudinal observations
D7.5 Data Analytic Package for the final submission for scientific advice from the EMA ± FDA
