Sensor-based explainable detection of cognitive decline
Introduction to the pilot
Context: Mild Cognitive Impairment (MCI) can progress to dementia, resulting in impaired cognitive function as well as decreased work, social, and relational abilities.
Problem: Sporadic visits with medical experts can not accurately assess the cognitive decline, which often consists of subtle behavioral changes.
Solution: We are developing a solution for continuous behavioral remote monitoring at home, so to identify digital biomarkers for early detection of cognitive decline.
Objective: To enable clinicians inspecting potential indicators of cognitive decline and improving their diagnosis.
SERENADE will be based on digital biomarkers evaluating mobility, activities of daily living, and behavior, defined as quantitative data collected and measured from digital devices (e.g., environmental sensors in the home, and wearable devices) in order to detect symptomatic and functional changes for MCI patients.
Each patent will be monitored thanks to a combination of unobtrusive environmental sensors (installed in the home infrastructure), mobile, and wearable devices. This sensing infrastructure generates a continuous sensor data stream that is stored in a gateway inside the home.
AI algorithms will analyze collected sensor data to identify and monitor human activities, sleep and movement patterns considered as relevant indicators of behavioral changes.
SERENADE will include a graphical interface for clinicians, that will show the output of our AI algorithm in a clear and explainable way as a decision support tool for their diagnosis.
The pilot’s goal is to integrate the decision support system in a telemedicine platform. This will facilitate the storage and management of sensible data, its analysis in an anonymous form, and the interaction with the regional health service, including booking and accounting of the proposed service.
SelfAct: Personalized Activity Recognition based on Self-Supervised and Active Learning, 20th EAI International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services, November 14-17, 2023 Melbourne, Australia., 2023 Publication
DOMINO: A Dataset for Context-Aware Human Activity Recognition using Mobile Devices, UNIMI Dataverse, V3, 2023 Dataset
DOMINO: A Dataset for Context-Aware Human Activity Recognition using Mobile Devices, In Proceedings of the 24th International Conference on Mobile Data Management (MDM), IEEE Computer Society, 2023 Publication
SmartFABER: Recognizing Fine-grained Abnormal Behaviors for Early Detection of Mild Cognitive Impairment, Artificial Intelligence in Medicine, Elsevier, 2016 Publication
Combining Public Human Activity Recognition Datasets to Mitigate Labeled Data Scarcity, In 2023 IEEE International Conference on Smart Computing (SMARTCOMP), 2023 Publication
The MARBLE dataset: Multi-Inhabitant Activities of Daily Living combining Wearable and Environmental Sensors Data, UNIMI Dataverse, V1, 2021 Dataset
Probabilistic Knowledge Infusion through Symbolic Features for Context-Aware Activity Recognition, Pervasive and Mobile Computing, Elsevier, 2023. (DOI: 10.1016/j.pmcj.2023.101780), 2023 Publication