EMOTIONAL EXHAUSTION IN CYBER DEFENCE TEAMS EXPLAINED THROUGH HR-DRIVEN PREDICTIVE MODELLING

Abstract

Emotional exhaustion is a burgeoning issue that we are seeing across cyber defence teams, as cyber defence professionals are continuously faced with high-level cognitive demands, high-stakes vigilance, and unpredictable workloads to manage. This paper outlined the study to examine how you could use predictive modelling to forecast emotional exhaustion in a cybersecurity environment grounded in human resource (HR) analytics and psychological assessment. Specifically, by synthesising and linking multiple data sources like workload logs, absenteeism, and exit interview records with psychometrics assessments from the Maslach Burnout Inventory (MBI), NASA Task Load Index (NASA-TLX), and PANAS, slotted these within a machine learning framework using algorithms like logistic regression and random forest, seek to build an explainable model. The logistic regression and random forest models could effectively classify staff members based on their low, moderate, and high-risk profiles for feelings of burnout with a high degree of predictive accuracy while preserving the model's explainability using SHAP and LIME. Two models were able to harness key predictors that were related to the frequency of overtime and perceived task overload, cueing an emotional affect imbalance. The model predicts where organisations can forecast emotional exhaustion and immediately act with an adaptive HR intervention for the management of emotional exhaustion, and improve the sustainability of the workforce overall. The key contributions are the potential practicality and ethical implications of incorporating psychological knowledge with HR data to help enable decision-making related to mental health in a more informed way within mission-critical domains like cybersecurity.