Qianwen Wu, Johar Permana, Taufani Chusnul Kurniatun, Suryadi Suryadi
To address the issues of strong subjectivity and dimensional fragmentation in the evaluation of leaders’ crisis response ability, this paper constructs a multimodal assessment model that integrates affective computing and behavior recognition. First, drawing on crisis management theory and emotional intelligence theory, deconstruct leaders’ crisis response ability into three core dimensions—emotional stability, cognitive flexibility, and decisiveness—and establish a mapping between these dimensions and multimodal data, including facial expressions, speech, physiological signals, eye movements, and behavioral logs. Secondly, a hybrid network model is designed to achieve quantitative assessment across dimensions through multimodal feature encoding, Bi-LSTM temporal modeling, and a dimensional interaction attention mechanism. Experiments show that the model achieves Mean Absolute Errors of 4.12, 4.58, and 4.35 for emotional stability, cognitive flexibility, and decision-making decisiveness, respectively. These values represent a 17.4% reduction in overall MAE compared to the best-performing baseline (MulT). The consistency rate between the comprehensive evaluation index and expert rating levels reaches 86.7%, outperforming comparative methods such as MFN and MulT. The research demonstrates that the proposed model can effectively capture the multidimensional dynamic interactions during crisis response, providing an interpretable and highly robust quantitative tool for evaluating leaders’ capabilities. © 2026 by author(s).
Universitas Pendidikan Indonesia, Bandung, 40154, Indonesia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock