JMIR Publications

ABSTRACT

Background:

Artificial Intelligence (AI) has already revolutionized the analysis of image, text, and tabular data, bringing significant advances across many medical sectors. Now, by combining with Wearable Inertial Measurement Units (IMUs), AI could transform healthcare again by opening new opportunities in patient care and medical research.

Objective:

This systematic review aims to evaluate the integration of AI with Wearable IMUs in healthcare, identifying current applications, challenges, and future opportunities. The focus will be on the types of AI models used, the characteristics of the datasets, and the potential for expanding and enhancing the use of this technology to improve patient care and advance medical research.

Methods:

Our aim is to identify studies that leverage AI to analyze IMU data, targeting specific medical conditions. The inclusion criteria specify that studies must be published in PubMed or Web of Science and focus on AI-assisted analysis of IMU data related to a particular medical condition. Exclusion criteria include studies lacking clear descriptions of the AI models used or those not employing IMU data. A systematic methodology, following PRISMA guidelines, was employed to answer three core questions: Which medical fields are most actively researching AI and IMU data? Which AI models are being utilized in the analysis of IMU data within these medical fields? And what are the characteristics of the datasets used for AI training in this context? The results were synthesized using a narrative approach, presenting findings in a clear and concise manner.

Results:

A total of 122 studies were analyzed, revealing key trends in AI-IMU data analysis. Notably, the median dataset size was 50 participants, highlighting limitations for AI methods that rely on large datasets. Machine Learning models dominated (76%) over Deep Learning models, while Supervised Learning was favored by 93% of studies, leaving Unsupervised Learning underutilized at 7%. A significant proportion (77%) conducted studies in clinical settings, rather than real-life scenarios, limiting practical applicability. The focus on neurological issues (65%) suggests potential for AI-IMU data analysis to impact many other medical areas like musculoskeletal, cardiorespiratory, or psychology.

Conclusions:

In conclusion, the review calls for a collaborative effort to address the highlighted challenges, including improvements in data collection, increasing dataset sizes, a move that inherently pushes the field towards the adoption of more complex Deep Learning models, and the expansion of the application of AI on IMU data methodologies across various medical fields. This approach aims to enhance the reliability, generalizability, and clinical applicability of research findings, ultimately improving patient outcomes and advancing medical research.