Guideline for the clinical practice of prevention and management of hypertensive disorders of pregnancy in primary health care

Guevara E.; Villagaray-Pacheco N.; Perez-Aliaga C.; Caytuiro H.; Vilchez Requejo R.; Guevara Gonzalez P.E.; Riesco de la Vega G.S.; Mercedes Asmat Aquije J.M.; Charri J.C.; Jumpa Armas D.V.; Salvador-Salvador S.; Timaná-Ruiz R.; Huaringa-Marcelo J. · Anales de la Facultad de Medicina (2025) · DOI: 10.15381/anales.v86i4.31486

Resumen

The complexity of monsoonal systems in Indonesia, driven by large-scale ocean-atmosphere interactions and local climate variability, requires the practical use of models for accurate rainfall prediction. This study presents an enhanced forecasting model for the Indonesian Monsoon Index (IMI) that integrates the Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Network (ANN) methodologies. We utilized ERA5, the European Centre for Medium-Range Weather Forecasts v.5 global reanalysis, zonal wind data at 850 and 200 hPa, and 30 years of precipitation data validated against in situ observations. This data was used to develop a modified IMI and assess its applicability in two selected regions: the North Coast of Java and East Kalimantan Province. Seasonal ARIMA modeling indicates robust short-term predictive capabilities (R² = 0.90) for five-month forecasts despite diminished performance in the presence of non-linear patterns or sudden climatic transitions. To address these limitations, a hybrid ANN-ARIMA model was adopted, which improved forecast accuracy by up to 8 months while increasing correlation (R² = 0.91) and reducing the risk of overfitting. This hybrid model effectively captures irregular seasonal variations, outperforming the conventional ARIMA, and provides more advanced tools for long-term forecasting under variable climatic conditions. In addition, the analysis revealed a notable connection between rainfall anomalies on the North Coast of Java and specific phases of the monsoon index, highlighting a significant influence of monsoons on local weather patterns. The improved predictive capabilities of the hybrid model is valuable for planning and decision-making in agriculture, water management, and disaster preparedness. © (2025), (Universidad Nacional Mayor de San Marcos). All rights reserved.