Answers are grounded in this paper and cite their sources.
The rapid development of AI in the analysis of halal foods and the management of pilgrimage presents opportunities for AI decision support for holistic and sustainable approaches, yet it can also introduce uncertainties about model reliability, ethical considerations for governance, and responsible deployment. In this study, a hybrid AI-based five-step framework is presented, which combines computer vision, natural language processing (NLP), a simulation-based pilgrim sustainability index (PSI), and a governance architecture in line with the sustainable development goals (SDGs). Seven convolutional neural networks were tested for halal/haram visual classification and four models, namely logistic regression, multinomial Naive Bayes, LSTM, and BiLSTM tested for ingredient text classification. The held-out test, bootstrap confidence intervals, paired statistical tests, regularisation, near-duplicate auditing, and group-aware partitioning and repeated seed validation were used to evaluate the reliability of models. When evaluated in a group-aware manner, the robust BiLSTM came across with 98.84% accuracy and AUC of 0.9969. The PSI was assessed for 10,000 synthetic pilgrim profiles and found to have a mean of 0.4848 (SD = 0.1730), while threshold and weighting sensitivity analysis showed dependence of categorical outcomes on specification choices. The results demonstrated support for the conceptual simulation based approach of HaTIS and lack of evidence of the operational validation process before deployment, but revealed a need for obtaining real world data and external validation before starting the process of deployment.
No takes yet. Share an insight, caveat, or question.
Rajak et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: