Related Experiment Video
Updated: Mar 27, 2026
08:45
Measuring the Kinematics of Daily Living Movements with Motion Capture Systems in Virtual Reality
Published on: April 5, 2018
8.1K
Artificial intelligence in Prechtl's General Movements Assessment: A systematic review and meta-analysis
Zhanna Zhussupova1, Amin Tamadon2, Natalya Chagay3
1Department of Neurology, West Kazakhstan Marat Ospanov Medical University, Aktobe, Kazakhstan.
Journal of Neonatal-Perinatal Medicine
|March 25, 2026
Summary
AI-assisted General Movements Assessment (GMA) shows high accuracy in predicting cerebral palsy (CP) and classifying expert labels. However, significant heterogeneity and bias limit certainty, highlighting the need for standardized protocols and validation.
Area of Science:
- Neurology
- Artificial Intelligence
- Developmental Pediatrics
Background:
- Cerebral palsy (CP) diagnosis relies on expert assessment, which can be subjective and resource-intensive.
- General Movements Assessment (GMA) is a valuable tool for early detection of neurological abnormalities.
- AI offers potential for objective and scalable analysis of GMA.
Purpose of the Study:
- To evaluate the performance of AI-assisted GMA in predicting later CP diagnosis.
- To assess AI-assisted GMA's accuracy in classifying expert-rated GMA labels.
- To analyze heterogeneity and risk of bias in AI-assisted GMA studies.
Main Methods:
- Systematic review and meta-analysis following PRISMA 2020 guidelines.
- Inclusion of 105 studies for qualitative synthesis and 28 for quantitative synthesis.
- Random-effects meta-analysis of proportions with logit transformation to estimate pooled accuracy.
Main Results:
- Pooled diagnostic accuracy for CP prediction was 0.884 (95% CI: 0.838-0.918).
- Pooled classification accuracy for expert-rated GMA labels was 0.848 (95% CI: 0.761-0.908).
- Substantial heterogeneity and very low certainty of evidence were observed across analyses.
Conclusions:
- AI-assisted GMA demonstrates high pooled performance for CP diagnosis prediction and expert-label classification.
- Significant heterogeneity and risk of bias necessitate caution and further research.
- Standardized protocols, high-quality data, and transparent validation are crucial for clinical adoption of AI-enabled GMA.
