A Hybrid Model for Supporting Auditors' Professional Judgment in Going Concern Evaluation Using Traditional Techniques and AI-Based Big Data Analytics

This study proposes a hybrid model that integrate the Altman Z-score— A traditional financial distress prediction Techniques -with six AI based Big Data Analytics (Deep Neural Networks (DNN), Recurrent Neural Networks (RNN), Support Vector Machines (SVM), Random Forests (RF), Decision Trees (DT), and K-Nearest Neighbors (KNN) to enhance the professional judgment of external auditors in evaluating an entity’s going-concern status. The model was empirically tested on a sample of 144 non-financial firms listed on the Egyptian Stock Exchange from 2018 to 2023. The findings indicate that although the Altman Z-score provides valuable insights into assessing an entity’s going-concern status, the Hybrid Model consistently outperforms the predictive performance of both the standalone Altman model and individual AI-based Big Data Analytics (BDA) techniques. The traditional Altman model achieves an accuracy of 84%. All Hybrid models exceed this baseline, with the Decision Tree (DT) model performing best at 94%, followed by the Deep Neural Network (DNN) at 92%, and the Recurrent Neural Network (RNN) at 91%, indicating that Hybrid models provide more reliable overall classifications. Also, Statistical tests, including McNemar, Phi, Cramer’s V, Kappa, -2log likelihood, and Nagelkerke R Square, consistently supported the effectiveness of the Hybrid Model. These findings highlight the potential of hybrid models to significantly elevate the quality of auditors’ professional judgment and decision-making in going-concern evaluations.