Based on the overall strategy for the OSCOFAI work plan, the work to be performed in the project will be realized through a set of work packages outlined below.
WP1. Actigraphy-based sleep-wake data associated with cognitive function
Duration: M01 - M24
Objectives: Examine associations of sleep-wake data from actigraphy with cognitive function.
Description of Work: Related to Aim1. Associate 7-day sleep-wake traditional summary actigraphy measures provided by the device software with baseline and changes in cognitive function over time (global composite z scores and individual composite domains z scores).
Objectives: Examine associations of sleep-wake data from actigraphy with cognitive function.
Description of Work: Related to Aim1. Associate 7-day sleep-wake traditional summary actigraphy measures provided by the device software with baseline and changes in cognitive function over time (global composite z scores and individual composite domains z scores).
Tasks
Task 1.1. Identify summary actigraphy measures, including total sleep time, sleep latency, wake time after sleep onset, sleep–wake ratio, and sleep efficiency to use in modeling- Dataset creation
Task 1.2. Investigate association between actigraphy sleep-wake measures and cognitive function cross-sectionally and prospectively.
Task 1.2. Investigate association between actigraphy sleep-wake measures and cognitive function cross-sectionally and prospectively.
Deliverables
D1.1 Actigraphy-based summary measures dataset
D1.2 Actigraphy-based sleep-wake measures associations with cognitive function.
D1.2 Actigraphy-based sleep-wake measures associations with cognitive function.
WP2. Actigraphy data associated with risk of MCI using advanced AI/ML techniques
Duration: M01 - M24
Objectives: Examine cross-sectional and prospective associations of features extracted from actigraphy time series with MCI incidence using advanced signal processing and machine learning methods.
Description of Work: Related to Aim2. Associate 7-day sleep-wake information with the risk of MCI using advanced ML - AI methods and actigraphy-extracted features.
Objectives: Examine cross-sectional and prospective associations of features extracted from actigraphy time series with MCI incidence using advanced signal processing and machine learning methods.
Description of Work: Related to Aim2. Associate 7-day sleep-wake information with the risk of MCI using advanced ML - AI methods and actigraphy-extracted features.
Tasks
Task 2.1. Actigraphy signal processing and AI - ML computational pipeline development
Task 2.2. Feature engineering using the developed pipeline
Task 2.3. ML models development and evaluation using actigraphy extracted features to predict MCI risk
Task 2.2. Feature engineering using the developed pipeline
Task 2.3. ML models development and evaluation using actigraphy extracted features to predict MCI risk
Deliverables
D2.1 Feature sets extracted from the actigraphy signal
D2.2 Predictive models of MCI risk based on actigraphy-extracted features
D2.2 Predictive models of MCI risk based on actigraphy-extracted features
WP3. Mediating effects of potential mechanisms
Duration: M01 - M24
Objectives: Examine whether plasma, CSF, and neuroimaging biomarkers mediate the relationships between sleep-wake cycle and cognitive function/impairment.
Description of Work: Related to Aim 3A. Assess the mediating role of plasma, CSF and neuroimaging biomarkers on the relationships between the sleep-wake information from actigraphy and cognitive function/impairment.
Objectives: Examine whether plasma, CSF, and neuroimaging biomarkers mediate the relationships between sleep-wake cycle and cognitive function/impairment.
Description of Work: Related to Aim 3A. Assess the mediating role of plasma, CSF and neuroimaging biomarkers on the relationships between the sleep-wake information from actigraphy and cognitive function/impairment.
Tasks
Task 3.1. Plasma Aβ, total and p-tau and GlycA analyses
Task 3.2. Mediation analyses
Task 3.2. Mediation analyses
Deliverables
D3.1 Biomarker measures dataset
D3.2 Mediation analyses
D3.2 Mediation analyses
WP4. Design and develop an AI tool predictive of cognitive impairment and associated outcomes of mediating biological mechanisms using actigraphy data
Duration: M01 - M24
Objectives: Identify actigraphy-based features predictive of outcomes of the individual mechanisms believed to correlate with cognitive decline. Determine the best combination of a subset of those features for predicting accurately cognitive decline risk.
Description of Work: Related to Aim3B. We will first use feature engineering and machine learning methods to identify good predictors for individual mechanism endpoints first. We will then check across mechanisms to determine the best combination of such features and ML models that best predict MCI-related endpoints when working synergistically.
Objectives: Identify actigraphy-based features predictive of outcomes of the individual mechanisms believed to correlate with cognitive decline. Determine the best combination of a subset of those features for predicting accurately cognitive decline risk.
Description of Work: Related to Aim3B. We will first use feature engineering and machine learning methods to identify good predictors for individual mechanism endpoints first. We will then check across mechanisms to determine the best combination of such features and ML models that best predict MCI-related endpoints when working synergistically.
Tasks
Task 4.1. Build effective predictive models linking actigraphy features to individual biological mechanisms believed to mediate cognitive decline
Task 4.2. Combine features identified in Task 4.1 to build a parsimonious ML model with input the smallest possible number of features that is highly predictive of cognitive decline
Task 4.3. Develop an “AI Advisor” tool that can process unseen patient actigraphy data and predict the risk of cognitive decline considering both cross-sectional and longitudinal data.
Task 4.2. Combine features identified in Task 4.1 to build a parsimonious ML model with input the smallest possible number of features that is highly predictive of cognitive decline
Task 4.3. Develop an “AI Advisor” tool that can process unseen patient actigraphy data and predict the risk of cognitive decline considering both cross-sectional and longitudinal data.
Deliverables
D4.1 Predictive models linking actigraphy features to individual biological mechanism endpoints associated with cognitive decline
D4.2 AI Advisor software tool for assisting clinicians in assessing cognitive decline status based on the objective actigraphy data of the patient
D4.2 AI Advisor software tool for assisting clinicians in assessing cognitive decline status based on the objective actigraphy data of the patient
WP5. Dissemination and communication
Duration: M01 - M24
Objectives: Disseminate and communicate our results to the broader scientific community, patients and their organizations and the society in general.
Description of Work: Create/apply a dissemination plan addressing key stakeholders via institutional websites, social media, press releases, scientific publications, and participation in forums, conferences, and symposia. We also communicate results to diverse audiences, including non-experts (patients, general society), through the website created in the context of the current WP as well as through social media. We will also communicate and spread the results through Athens Alzheimer’s Association addressed to people with dementia, their relatives and health care professionals interested in Alzheimer’s disease; this will also enhance the recruitment.
Objectives: Disseminate and communicate our results to the broader scientific community, patients and their organizations and the society in general.
Description of Work: Create/apply a dissemination plan addressing key stakeholders via institutional websites, social media, press releases, scientific publications, and participation in forums, conferences, and symposia. We also communicate results to diverse audiences, including non-experts (patients, general society), through the website created in the context of the current WP as well as through social media. We will also communicate and spread the results through Athens Alzheimer’s Association addressed to people with dementia, their relatives and health care professionals interested in Alzheimer’s disease; this will also enhance the recruitment.
Tasks
Task 5.1. Develop a dissemination plan including a set of materials (e.g., leaflets, posters, news, videos, etc.) for dissemination of the project at conferences and through patient organization websites and social media
Task 5.2. Create a project website with information on project goals and results; it will also include an internal wiki type part to facilitate and enhance internal scientific work
Task 5.3. Compiled Scientific journal publications/conference proceedings and / or press releases on selected results
Task 5.2. Create a project website with information on project goals and results; it will also include an internal wiki type part to facilitate and enhance internal scientific work
Task 5.3. Compiled Scientific journal publications/conference proceedings and / or press releases on selected results
Deliverables
D5.1Dissemination/communication plan
D5.2 Project Website
D5.3 Scientific journal publications/conference proceedings and/or press releases (at least two)
D5.2 Project Website
D5.3 Scientific journal publications/conference proceedings and/or press releases (at least two)
WP6. Project management
Duration: M01 - M24
Objectives: Ensure that the project meets its objectives and reporting requirements within scheduled timeframes.
Description of Work: Planning and monitoring the project progress by the PI.
Objectives: Ensure that the project meets its objectives and reporting requirements within scheduled timeframes.
Description of Work: Planning and monitoring the project progress by the PI.
Tasks
Task 6.1. Project coordination including bi-monthly research team meetings. The PI will be responsible for planning and monitoring the progress of the work, ensuring the timely submission of the deliverables and all progress reports. For that purpose the PI will organize bi-monthly progress meetings with all participants. The minutes of these meetings will be available to all participants.
Deliverables
D6.1 Interim Progress report
D6.2 Final project report
D6.2 Final project report