Clinical Translation of Artificial Intelligence-Driven Gait Analysis Using Plantar Pressure and Ground Reaction Force
Junxiao Yang1, Chunli Dong1, Xiuping Zhang1
1School of Nursing, Jining Medical University, Jining 272000, China.
Bioengineering (Basel, Switzerland)
|July 28, 2026
Summary
Artificial intelligence (AI) gait analysis shows promise for rehabilitation but faces translation challenges. More robust, validated, and interpretable AI models are needed for clinical use.
Area of Science:
- Biomedical Engineering
- Rehabilitation Science
- Artificial Intelligence in Medicine
Background:
- AI-driven gait analysis using plantar pressure and ground reaction force (GRF) offers potential digital biomarkers for rehabilitation.
- Clinical translation of these AI tools remains uncertain due to various gaps.
Purpose of the Study:
- To conduct a scoping review and evidence map of AI-driven gait analysis in clinical settings.
- To summarize current applications, assess evidence maturity, and identify translational challenges.
Main Methods:
- Searched major databases (PubMed, Web of Science, Embase, Scopus) up to May 2026.
- Included original clinical studies using AI with plantar pressure or GRF signals for disease recognition, assessment, prediction, monitoring, or decision support.
Main Results:
- Fifteen studies were included, with evidence concentrated in Parkinson's disease (PD) recognition and freezing of gait prediction.
- Evidence for other conditions (e.g., knee osteoarthritis, fall risk) is less mature.
- Translation is hindered by small sample sizes, limited validation, lack of calibration, and insufficient real-world testing.
Conclusions:
- Future research must focus on prospective, externally validated, interpretable, and calibrated AI models.
- Clinical embedding and real-world workflow testing are crucial for successful implementation in rehabilitation.
