PhD Studentship: Digital Movement Biomarkers of Mood and Fatigue in Older Adults: Using Computer Vision, Artificial Intelligence and Functional Movement Analysis to Detect Anxiety, Depression and Fatigue in People With and Without Multiple longterm conditions

Digital Movement Biomarkers of Mood and Fatigue in Older Adults: Using Computer Vision, Artificial Intelligence and Functional Movement Analysis to Detect Anxiety, Depression and Fatigue in People with and without Multiple Long-Term Conditions (MLTCs) Depression, anxiety and fatigue are among the most common and disabling symptoms experienced by older adults living with multiple long-term conditions (MLTCs), including stroke, Parkinson's disease, frailty, arthritis and mild cognitive impairment. These symptoms are associated with reduced mobility, poorer rehabilitation outcomes, loss of independence, increased healthcare utilisation and lower quality of life.

However, detecting changes in mood and fatigue within these populations remains challenging. Symptoms often overlap with physical impairments, cognitive changes and disease-related functional limitations, making accurate identification difficult and delaying intervention. As the prevalence of MLTCs continues to rise, improving the objective detection and monitoring of mood-related symptoms has become a major healthcare and research priority. Current assessment approaches rely primarily on self-report questionnaires and clinical interviews conducted during infrequent healthcare encounters. These methods are subjective, vulnerable to reporting bias and often fail to capture subtle fluctuations that occur between appointments. There is therefore a critical need for innovative and scalable technologies capable of objectively assessing psychological wellbeing.

Emerging evidence suggests that depression, anxiety and fatigue influence movement behaviour, affecting posture, balance, movement variability, transfers and overall functional performance. Advances in computer vision and artificial intelligence (AI) now provide a unique opportunity to quantify these movement signatures using standard 2D video recordings. Markerless pose estimation methods can extract detailed skeletal movement information from low-cost RGB cameras without wearable sensors or specialist laboratory equipment, creating opportunities for large-scale deployment across healthcare, community and home settings.

This PhD aims to develop and validate digital movement biomarkers capable of detecting depression, anxiety and fatigue from functional movement recordings in older adults with and without MLTCs. A major strength of the project is the availability of an existing, richly characterised dataset that has already been collected. The dataset comprises more than 1,000 functional movement assessments from adults aged 60 years and older, including healthy individuals and people living with stroke, Parkinson's disease, frailty, arthritis and mild cognitive impairment. Associated measures include demographics, clinical characteristics, functional outcomes, mood assessments and fatigue scales, providing a unique foundation for the development and validation of novel digital biomarkers. The project will first apply state-of-the-art computer vision approaches, including OpenPose, MediaPipe and MoveNet, to extract biomechanical and behavioural features from videos of posture, balance, transfers, walking and activities of daily living.

Extracted variables will include spatiotemporal characteristics, joint kinematics, movement smoothness, symmetry, postural alignment and inter-limb coordination. Traditional statistical techniques will establish interpretable baseline relationships between movement patterns and measures of depression, anxiety and fatigue. Nonlinear movement analyses, including entropy measures, fractal dynamics and recurrence based metrics, will then be used to identify subtle neurobehavioural signatures that may not be captured through conventional biomechanical approaches. Building on these features, the research will develop a hierarchical AI framework for digital biomarker discovery. Explainable machine learning models, including Random Forests, XGBoost and LightGBM.

Advert information