Background: Age-related mobility decline is often unavoidable, resulting in wearable mobility devices to augment mobility patterns and behaviors. Artificial intelligence (AI) and machine learning (ML) are emerging technologies that have been integrated into mobility wearable devices to enhance older adults’ mobility. This state-of-the-art review aimed to identify the range and breath of studies that have applied AI/ML in wearable mobility devices, and their associated hardware and software features.

 

Methods: We followed Grant and Booth’s state-of-the-art Framework to search seven databases using librarian guided search terms such as “artificial intelligence”, “machine learning”, “mobility”, “wearables”, and “older adults”. Using predefined inclusion and exclusion criteria, two reviewers independently conducted title/abstract and full screening, and extracted data in COVIDENCE. We mapped AI systems across the five categories of machine learning – supervised learning (SL), unsupervised learning (USL), semi-supervised learning (SSL), reinforcement learning (RL) and deep learning (DL).

 

Results: We included 14 studies conducted in North America (n=6), Asia (n=4), Europe (n=3) and multiple continents (n=1), across several study designs including 7 validation, 4 prospective observational, 2 cross-sectional, and one case series design. The types of mobility wearable devices (hardware) included accelerometers (n=8), smartwatches or belts (n=3), insole or body sensors (n=2) and an inertia measurement unit (n=1), which were worn on the trunk and lower limbs (n=5 each), upper limbs (n=3) and head (n=1). Across the five categories of AI/ML software systems, 13 studies utilised SL – a form of ML where the algorithm is trained on labelled data to map an input data to an output sample, while one study employed DL – a subset of ML that uses neural networks with multiple layers to identify patterns and make predictions. Other AI systems, including USL, SSL and RL were not utilised in mobility wearable devices in any of the studies.

 

Conclusions: This review highlights a growing but still emerging body of literature on the use of AI-powered wearable mobility devices for older adults, with pre-dominant focus on SL and DL used mostly in validation studies. The limited use of other advanced AI/ML such as USL, SSL and RL and limited real-world studies underscores a significant opportunity for innovation and more research in this field. Advancing the application of diverse AI methodologies in mobility wearables in real world contexts could transform mobility support and promote independence and healthy ageing among older adults.

Chukwuebuka Prince Onyekere is a Nigerian-trained physiotherapist and currently a PhD student at the School of Kinesiology and Health Science, York University, Canada. Chukwuebuka is also an ageing researcher affiliated with the Emerging Researchers and Professionals in Ageing- African Network.

 

Chukwuebuka’s research focuses broadly on older adults with specific interest in mobility and social connectedness among older adults and how culturally informed technologies and artificial intelligence can be used to improve older adults’ mobility and social connectedness.