APPLICATION OF ARTIFICIAL INTELLIGENCE IN MOBILE APPLICATION FOR INFANT MOVEMENT CLASSIFICATION

File
Yung-Jen Hsu, Po-Nien Tsao, Ting-An Yen, Wei-Chih Liao, Wei-J Chen, Suh-Fang Jeng, Yu-Ching Hsiao
Purpose:

This study aimed to examine the parental perception of infant movements and developed an action recognition model incorporating an application (APP) to identify movements in preterm and full-term infants.

Methods:

This study included 37 preterm infants and 54 full-term infants and prospectively followed up their motor development using the Alberta Infants Motor Scale (AIMS) (58 movements) from 4 to 18 months of age. The APP “Baby Go,” based on the AIMS movement items, was developed for their parents to upload their child’s movement videos at home for subsequent machine learning and movement identification. The app contains an education module to guide parents regarding good-quality video recording. Parental perception of infant movements was examined using the agreement of parents’ classification of movements compared with the physiotherapist’s assessment results. Home video files were further annotated by the physiotherapist and served as the standards to examine the accuracy of the AI algorithm.

Results:

Parents accurately classified supine and prone movements (agreement >65%) but were less likely to classify sitting and standing movements. Missing body parts and unstable camera movement were common factors affecting the quality of home videos, with 27.7% of annotated results showing missing body parts and 27.4% of annotated results showing unstable camera movement. The parents uploaded 1,027 home videos via the app. The overall accuracy of the action recognition model in classifying 31 movements was 0.77, precision was 0.66, recall was 0.66, and F-score was 0.65. 

Conclusion(s):

Parents accurately perceived infant movements mainly occurring in supine and prone positions. The overall action recognition model of home videos achieved moderate accuracy. Future work needs to increase the number of home videos and the heterogeneity of the sample to enhance the accuracy of the action recognition model and the generalizability of the results.

Implications:

The findings provide insightful information regarding the refinement of AI for infant motor assessment via a mobile app to help early identify infants with a risk of neurodevelopmental disabilities.

Funding acknowledgements:
This study was funded by the National Science and Technology Council (MOST 110-2314-B-002-055-MY3)
Keywords:
Artificial intelligence
Home videos
Infant movement screening
Primary topic:
Paediatrics
Second topic:
Innovative technology: information management, big data and artificial intelligence
Did this work require ethics approval?:
Yes
Name the institution and ethics committee that approved your work:
National Taiwan University Hospital Research Ethics Committee
Provide the ethics approval number:
202012089INB
Has any of this material been/due to be published or presented at another national or international conference prior to the World Physiotherapy Congress 2025?:
No

Back to the listing