Copyright (c) 2026 Documentos de Trabajo ECACEN
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
AI-Based Multilayer Urban Risk Prediction Model for Vulnerable Road Users in Bogotá: The MURIM Model and Its Integration with Real-Time Navigation Platforms”
This paper proposes the Multilayer Urban Risk Intelligence Model (MURIM), a spatiotemporal predictive framework designed to estimate dynamic risk levels for intersections and road segments in Bogotá. Unlike conventional heatmap-based systems, MURIM introduces the concept of Traffic-Condition-Dependent Urban Vulnerability Modeling, whereby risk is not solely a function of geographic location, but rather emerges from the interaction among location, real-time traffic conditions, and the user's exposure profile. The model integrates five layers criminal security, congestion exposure, climate vulnerability, social disruptions, and user profile into a Dynamic Urban Risk Score (DURS) with machine-learning-calibrated weights. Three user profiles with distinct exposure windows are identified: logistics operators, nighttime service drivers, and female drivers during daytime hours. For the latter group, the study's central finding is that the primary risk is not nighttime-related but occurs during the day, concentrated in periods of immobilization at traffic lights and in traffic congestion. The paper also proposes a technical architecture for integration with Google Maps Platform and Waze for Cities, leveraging a data infrastructure that is already fully available in Bogotá.
