Dual-step Nearest Neighborhood Prediction Method of Individual Childhood Growth

New dual-step child growth prediction model is developed based on regular longitudinal measurements to evaluate the expected child growth development and the growing patterns. Based on percentiles on reference growth charts used by pediatricians as the first step, a measure of distances between the reference curves and child growth data are utilized to tentatively estimate the future child growth expectation. The second-step method is implemented to predict the future growth to follow up the normal growth trends by using neighborhood-based children dataset in the same age group. Scoring nearest neighborhood-based curve matching is the technique for matching the nearby curves to represent the predicted values. The results demonstrated that the dual-step prediction methods estimate future individual child growth accurately with average errors less than 1.2 cm. Comparison results from the reference line showed with average errors less than 1.5 cm. obtained from 12-month future growth forecasting.

[1]  S. van Buuren,et al.  Response conversion for the Health Monitoring Program , 2004 .

[2]  G. G. Stokes.  "J." , 1890, The New Yale Book of Quotations.

[3]  T J Cole,et al.  Growth charts for both cross-sectional and longitudinal data. , 1994, Statistics in medicine.

[4]  T J Cole,et al.  Presenting information on growth distance and conditional velocity in one chart: practical issues of chart design. , 1998, Statistics in medicine.

[5]  M. de Onís,et al.  Worldwide practices in child growth monitoring. , 2004, The Journal of pediatrics.

[6]  H. Jung,et al.  Near final height in Korean children referred for evaluation of short stature: clinical utility and analytical validity of height prediction methods , 2018, Annals of pediatric endocrinology & metabolism.

[7]  T. Marshall,et al.  The use of short-term increments in weight to monitor growth in infancy. , 1992, Annals of human biology.

[8]  T. Reinehr,et al.  A New Model of Adult Height Prediction Validated in Boys with Constitutional Delay of Growth and Puberty , 2019, Hormone Research in Paediatrics.

[9]  Naomi S. Altman,et al.  Quantile regression , 2019, Nature Methods.

[10]  Dietrich von Rosen,et al.  Does noise reduction matter for curve fitting in growth curve models? , 2009, Comput. Methods Programs Biomed..

[11]  W. Johnson.  Analytical strategies in human growth research , 2014, American journal of human biology : the official journal of the Human Biology Council.

[12]  M. Segal,et al.  Trees and tracking. , 1993, Statistics in medicine.

[13]  R. Koenker,et al.  Quantile regression methods for reference growth charts , 2006, Statistics in medicine.