Predictive Modeling with Artificial Intelligence to Mitigate Variability and Cost Overruns in Andean Public Infrastructure

Magno Chavarry Vallejos C.; Samuel Támara Rodríguez J.; Janet Chavarría Reyes L.; Clotilde Panana Holgado E.; Cristina Chiok Guerra A.; Cesar Coral Jamanca J.; Eloy Solorzano Poma J. · Proceedings of the LACCEI international Multi-conference for Engineering, Education and Technology (2026) · DOI: 10.18687/laccei2026.1.1.864

Resumen

This study sought to create a framework with Artificial Intelligence (AI) to address an urgent problem: the instability, unpredictability and high costs in the management of public infrastructure in the Andean region. To this end, a quantitative approach was adopted, analyzing 88 representative Peruvian state projects. The measurement tool used showed high reliability, ensuring the robustness of the data. The results are eye-opening and highlight two critical gaps. First, 76% of the works do not consider sustainability criteria in their execution. Second, and even more significantly, 88% lack any technological tool that allows predicting risks or optimizing resources. This double lack generates a vicious circle of risk and uncertainty, where cost overruns and unforeseen events are the norm, ultimately affecting the quality and impact of the infrastructure that reaches the communities. A significant correlation (rs=0.78) was demonstrated between the absence of machine learning models and the uncontrolled variability of costs and deadlines. It is concluded that the integration of AI is imperative to move towards precision engineering that prioritizes environmental protection and climate resilience under the framework of SINAGERD 2050. The main contribution lies in a decision-making support system capable of optimising resources, reducing the carbon footprint and ensuring compliance with international standards (ISO), consolidating technology as the axis of a symbiosis between industrial productivity and the preservation of ecosystems.

Instituciones: Universidad Ricardo Palma; Universidad Nacional Santiago Antúnez de Mayolo