Clinical validation of an abridged AIMS: Streamlining motor screening in the first-year infant

Area of Science:

• Pediatrics
• Developmental Pediatrics
• Machine Learning in Healthcare

Background:

• The Alberta Infant Motor Scale (AIMS) is a standard tool for assessing infant motor development.
• Streamlining the AIMS assessment process can improve efficiency in early intervention (EI) services.
• Video analysis and machine learning offer potential for automated or semi-automated developmental screening.

Purpose of the Study:

• To validate an abridged version of the AIMS, the "salient set," for efficient infant motor screening.
• To assess the accuracy of machine learning models in predicting full AIMS scores using the salient set.
• To evaluate the sensitivity and specificity of the salient set for identifying infants needing EI referral.

Main Methods:

• Retrospective analysis of 21 infant videos manually tagged by occupational therapists using the 15-item salient set.
• Support vector regressors (SVRs) were trained on a larger dataset (n=102) to predict full AIMS scores from salient set data.
• Concurrent validity was assessed by correlating salient set predictions with full AIMS scores.

Main Results:

• The SVR model demonstrated strong concurrent validity with the full 58-item AIMS (Pearson correlation: 0.99).
• The salient set achieved high screening sensitivity (1.0) and specificity (0.895).
• Evaluation time was reduced by 67% using the abridged salient set.

Conclusions:

• The salient set of the AIMS, when analyzed with machine learning, provides a valid and efficient method for infant motor screening.
• This approach accurately identifies infants requiring early intervention services.
• The salient set represents a significant advancement in applying machine learning to developmental assessment, reducing screening time while maintaining accuracy.