Mapping global research on artificial intelligence in physical therapy: a bibliometric analysis from 1990 to 2023

The application of artificial intelligence (AI) in physical therapy has garnered increasing interest in recent years. We aimed to explore the current state of research on AI applications in physical therapy using bibliometric methods. A comprehensive literature search was conducted in Scopus (1990–2023). Two independent reviewers assessed titles, abstracts, and full documents. Inclusion criteria consisted of documents addressing AI applicability in physical therapy. Bibliometric analysis was conducted using VOSviewer and the R package Bibliometrix. A total of 805 studies were retrieved. After applying exclusion criteria and screening, 460 documents published across 317 journals were included, showing an annual growth rate of 16.7%. The average document age was 5.1 years. Contributions came from 1974 authors, with the University of Toronto being the most prolific institution. Research originated from 65 countries, led by the USA, followed by China, India, Germany, and Canada. Key themes included ‘machine learning’, ‘rehabilitation’, ‘physiotherapy’, ‘artificial intelligence’, ‘physical therapy’, and ‘deep learning’. The number of publications on AI in physical therapy has grown significantly. Despite this, there is a notable gap in international collaboration, with research primarily centred in high- and upper-middle-income countries. Findings provide valuable insights into underexplored topics representing potential areas.