Abstract
<title>Abstract</title> <p> <bold>Background:</bold> Population ageing is accelerating globally, placing increasing pressure on healthcare systems to detect chronic and neurodegenerative diseases at the earliest possible stage. Traditional episodic clinical assessments are ill-suited to capturing the subtle, day-to-day behavioural and physiological changes that precede formal diagnosis. Digital biomarkers represent a paradigm shift toward continuous, passive, and ecologically valid health monitoring. <bold>Objective:</bold> This scoping review synthesises evidence on four primary digital biomarker domains including language production, typing behaviour, smartphone interactions, and wearable sensor measurements, and evaluates their clinical utility for early detection of cognitive decline, frailty, Parkinson's disease, stroke risk, and late-life depression in adults aged 60 years and older. <bold>Methods:</bold> A systematic literature search was conducted across PubMed, Embase, IEEE Xplore, and ACM Digital Library (2015 to 2024) following PRISMA-ScR guidelines. Studies were eligible if they enrolled adults aged 60 years or older, used at least one passive digital biomarker modality, and reported a clinical outcome. A total of 87 studies met eligibility criteria after full-text screening. <bold>Results:</bold> Language-based biomarkers demonstrated diagnostic accuracy for cognitive decline with area under the receiver-operating-characteristic curve (AUC) values of 0.72 to 0.91. Keystroke dynamics identified Parkinsonian motor features with sensitivity of 78 to 86%. Smartphone passively-sensed variables predicted depressive episodes with AUC 0.74 to 0.88. Wearable-derived gait and heart-rate variability measures predicted frailty and cardiovascular events with AUC 0.79 to 0.94. Multimodal fusion models consistently outperformed single-modality classifiers. <bold>Conclusion:</bold> Multimodal digital biomarkers collected unobtrusively via everyday devices offer a scalable, cost-effective complement to conventional clinical assessment in older adults. Rigorous prospective validation, standardised interoperability frameworks, and privacy-preserving architectures are essential before widespread clinical deployment. </p>