Background: Gross motor development during the first year of life reflects early brain maturation and neuromotor pathway integrity. High-risk infants, especially those exposed to prenatal, perinatal, or neonatal complications, are more likely to demonstrate delayed milestone acquisition. Factors such as birth asphyxia, neonatal seizures, meconium aspiration, small-for-gestational-age (SGA) status, prolonged oxygen support, low Apgar scores, multiple gestation, consanguinity, and periventricular leukomalacia (PVL) are known contributors to early motor impairment. Early prediction of motor outcomes enables timely physiotherapy intervention and improved developmental surveillance.

Objective: To identify significant prenatal, perinatal, and neonatal predictors of 12-month gross motor development in high-risk infants and to develop a regression-based predictive model.

Methodology: A cross-sectional analytical study was conducted on 284 high-risk infants attending the NICU follow-up clinic of a tertiary-care hospital. Eighteen clinically relevant maternal, perinatal, and neonatal factors were recorded from medical records. Gross motor development at 12 months was assessed using the ASQ-3 Gross Motor domain. Multivariate linear regression was performed in two steps: an initial model identifying 14 significant variables, followed by a final model yielding 10 independent predictors. The model incorporated the constant and unstandardized coefficients of these predictors to generate a predictive formula for 12-month gross motor scores.

Results: Ten factors significantly predicted 12-month gross motor performance: history of miscarriages, consanguineous marriage, birth asphyxia, meconium aspiration, SGA status, oxygen support days, neonatal seizures, PVL, low Apgar score, and multiple gestation. The final model explained 33.9% of score variance. Model validation against actual ASQ-3 scores demonstrated a sensitivity of 68.80%, specificity of 60.57%, positive predictive value of 52.08%, negative predictive value of 75.71%, and overall accuracy of 63.73%.