04 Jun Wearable Sensors and Artificial Intelligence for Ecological Knee Osteoarthritis Assessment: Development and Feasibility of a Hybrid Digital Phenotyping Framework
J. Mapinduzi , K. Daniels , O. Kossi , J. Verbrugghe and B. Bonnechère
Abstract
Osteoarthritis (OA) is a highly prevalent musculoskeletal disorder and a major cause of
disability, posing growing challenges for healthcare systems worldwide. Conventional
supervised clinical assessments provide valuable insights but are largely limited to crosssectional snapshots and often fail to reflect the variability of real-world functioning, physical activity patterns, and symptom fluctuations experienced by individuals with OA, especially those with knee OA. This perspective introduces a multisensor digital phenotyping framework for smart knee OA assessment, integrating supervised laboratory evaluations with unsupervised continuous monitoring in daily living environments using wearable sensors, smart insoles, activity trackers, and mobile devices. Feasibility was tested in
40 participants (20 knee OA patients, 20 controls). Raw data from questionnaires, electronic goniometry, dynamometry, force plate, connected insoles, and seven-day home
monitoring were harmonized via a standardized pipeline aligned with the ICF framework. The pipeline employed anomaly detection, missing data imputation, z-score normalization, and cloud-based storage. This framework is envisioned to facilitate advanced
data integration and machine-learning-ready analytics, enabling longitudinal monitoring,
pattern recognition, and individualized health profiling. By conceptually bridging crosssectional and continuous sensing modalities, this approach has the potential to enhance
ecological validity, support earlier identification of functional decline, and inform datadriven clinical decision-making. Key methodological, technological, and ethical challenges—including data quality, interpretability, privacy, digital literacy, and clinical
adoption—are also highlighted. Overall, this paper underscores the promise of AI-enabled multisensor digital phenotyping to advance smart, personalized, and precision
healthcare for individuals with knee OA.