International Journal on Advanced Science, Engineering and Information Technology, Vol. 12 (2022) No. 5, pages: 2083-2088, DOI:10.18517/ijaseit.12.5.15047

The Spatial and Flexible Model for Low-Birth-Weight Prediction

Waego Hadi Nugroho, Agus Dwi Dwi Sulistyono, Atiek Iriany, Novi Nur Aini

Abstract

The case of Low Birth Weight (LBW) in Indonesia is still high. Based on data from the Central Agency on Statistics (BPS), the case of LBW in East Java is still high, with about 20,836 people in 2016 and 14,882 people in 2017. Many factors trigger LBW, especially women's condition and nutritional intake during pregnancy. This study aims to establish a location-based LBW prediction model using a spatial and flexible model. This research was conducted using a Geographically Weighted Regression (GWR) model in East Java and a flexible model with a genetic programming approach. Endogenous variables (Y) portrayed LBW cases in East Java and exogenous variables were the Percentage of Early Marriage (X1), Human Development Index (X2), Number of Midwives (X3), K1 Visit (X4), K4 Visit (X5), Consumption of Fe 30 (X6), and Consumption of Fe 90 (X7). Based on the analysis results using the GWR model, global equation models, and local models 38 models were obtained with R2 = 82.06%. Meanwhile, based on the results of the analysis with a flexible model with a deep learning approach, the model was obtained with R2 = 84.8%. From this study, it can be concluded that the GWR model and the flexible model have the same level of accuracy. However, the flexible model can show the non-linear effect of the variables X1 and X3.

Keywords:

Geographically weighted regression; genetic programming; low birth weight; flexible model.

Viewed: 92 times (since abstract online)

cite this paper     download