The purpose of this preliminary study is to apply AI technology on a sample of infants aged 4 to 18 months to develop an action tracking and recognition algorithm for infant motor screening and to determine the accuracy of the captured movements during the Alberta Infant Motor Scale (AIMS) assessment using an experienced physical therapists' assessment results as the reference.

Background and Purpose. Although there is an increase in the public awareness of early intervention for children with developmental disorder in Taiwan, the number of children reported for early intervention by the Ministry of Health and Welfare is limited, particularly for those aged under two years, outside of hospital settings, and in remote areas. This has highlighted the need of early screening for infants who are at risk or have developmental disorders. While motor development has a potential impact on the emergence of abilities in other domains in children at later age, motor screening may serve as the cornerstone to help detect signs of developmental dysfunction. Artificial Intelligence (AI), based on machine learning of big data, may be an alternative for assisting healthcare professionals to efficiently screen children's development and to help plan for further diagnostic assessment. The purpose of this preliminary study is to apply AI technology on a sample of infants aged 4 to 18 months to develop an action tracking and recognition algorithm for infant motor screening and to determine the accuracy of the captured movements during the Alberta Infant Motor Scale (AIMS) assessment using an experienced physical therapists' assessment results as the reference. Method. This study will recruit 50 infants (40 preterm infants and 10 term infants age 4-18 months (corrected age for preterm infants) from the National Taiwan University Children's Hospital. Each infant will be evaluated by a physical therapist for their gross motor development during prone, spine, sitting and standing positions using the AIMS assessment. The whole assessment procedure will be video recorded by five cameras. The data processing of movement video records will consist of selection of movement records, establishment of a pose estimation model, and establishment of an action recognition model. The accuracy of the pose and action recognition model in identifying infants' movements will be examined using the physical therapist's results as the gold standard. The results of this study will provide preliminary data to help establish the best and appropriate action recognition model of infant motor screening for future validation on a large sample.

Inclusion Criteria:

• The inclusion criteria for term infants are: gestational age 37-42 weeks, birth weight >2,500 grams, and no congenital/genetic abnormalities.
• The inclusion criteria for preterm infants are: gestational age <37 weeks, birth weight <2,500 grams, and no congenital/genetic abnormalities.
• Their mothers are older than 20 years of age, have no history of alcohol or drug abuse, and are married or live with fathers.

Exclusion Criteria:

• No

The inclusion criteria for preterm infants are: gestational age <37 weeks, birth weight <2,500 grams, and no congenital/genetic abnormalities. Their mothers are older than 20 years of age, have no history of alcohol or drug abuse, and are married or live with fathers.