Explainable Multidisciplinary Artificial Intelligence for Sustainable Innovation and Intelligent Decision-Making Across Emerging Research Domains

Artificial intelligence (AI) is transforming scientific research, environmental sustainability, healthcare, and intelligent infrastructure. However, conventional AI approaches often lack interpretability, cross-domain adaptability, and transparent decision-making. This study proposes an explainable multidisciplinary AI framework integrating predictive analytics, explainable AI (XAI), and multi-criteria decision analysis to support sustainable innovation across diverse research domains. The proposed framework combines Random Forest, XGBoost, and Explainable Boosting Machines with SHAP-based feature attribution and multi-criteria decision analysis. A comparative evaluation design considers environmental sustainability, healthcare, and intelligent infrastructure as representative application domains. Models are assessed using accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (AUROC), explanation fidelity, computational latency, and decision consistency. Baseline comparisons include conventional machine learning without explanations, explainable models without decision optimization, and the integrated framework. Sustainability indicators, interpretability, computational efficiency, and decision robustness are incorporated into a unified evaluation protocol. Performance is compared using consistent data-splitting procedures, domain-specific metrics, and sensitivity analysis. The proposed framework establishes a measurable evaluation strategy for determining whether integrated explainability and decision optimization improve predictive reliability and practical decision support. Quantitative outcomes will include differences in predictive performance, explanation fidelity, latency, and decision consistency across models and domains. Qualitative analysis will examine stakeholder interpretability, cross-domain adaptability, transparency, and the relevance of generated recommendations. Numerical improvements will be reported only after empirical experimentation. This research advances a unified approach to explainable, multidisciplinary AI by connecting predictive intelligence with transparent, sustainability-oriented decision-making. Its comparative design supports reproducible assessment of accuracy, interpretability, efficiency, and robustness. The framework offers a foundation for responsible AI deployment across emerging research domains while highlighting the need for domain-specific validation, representative datasets, and human oversight.