PARAMETRIC FEATURE INTERACTION AND SENSITIVITY ANALYSIS IN MULTI-HAZARD DISASTER MODELING FOR AI-BASED DECISION SUPPORT SYSTEMS

https://doi.org/10.59982/18294359-26.1-pf-35

Abstract

This study examines one specific structural issue for AI-based Emergency Decision Support Systems (DSSs) concerning foundational disaster modeling, which fails to adequately capture the intricate interplay of multi-faceted hazard parameters. Although contemporary DSSs increasingly incorporate features of cognitive reasoning and adaptability, the modeling of the layers of disasters is still mostly linear or unstructured.

The study proposes a parametric interaction approach that models disasters as non-linear spaces with specific coupling and sensitivity coefficients. This approach is designed to capture other mechanisms, apart from enhanced prediction, such as aggradations, structural destabilization, and risk-increasing interactions in a multi-hazard environment. The framework also helps to understand the coherence of interactions among parameters enabling the system to distinguish isolated signal anomalies from structurally significant escalations.

The author then offers a mathematically structured layer of DSSs that contains no decision biases and can be integrated with AI-based DSS. Further, the approach is backed by empirical evidence derived from structured disaster datasets, from which he is able to enhance the robustness, interpretability and, reliability of a system under conditions of uncertainty.

Keywords: Multi-hazard modeling, Parametric interaction, Sensitivity analysis, Structural risk, Nonlinear systems, Uncertainty modeling

PAGES: 419-429

DOWNLOAD FULL ARTICLE