Some limitations should be noted: The validation dataset is synthetic, parameterised from published maintenance studies rather than live operational data. Sustainability metrics are treated as deterministic, whereas real-world values shift with supplier practices and transport conditions. Real-time condition monitoring data, which could improve both demand forecasts and circular economy classifications, is not yet incorporated. Four directions remain for future work. First, the framework needs testing with live inventory data in collaboration with actual airline operators. Second, sensor-based condition data could support dynamic, condition-driven replenishment rather than relying solely on historical patterns. Third, a distributed ledger system would enable trusted peer-to-peer transactions for reusable and repairable parts across MRO networks. Fourth, interpretability tools are necessary to reveal the sustainability logic embedded in model recommendations—without this transparency, maintenance managers might be slow to adopt. Collectively, these developments would help transition the framework from proof-of-concept to a fully operational system, creating a supply chain that is both more resilient and environmentally less damaging.

Several important limitations need acknowledgment: The dataset used for validation consists of synthetic data based on parameters from published maintenance research, not actual operational information from live systems. The sustainability measurements are handled as fixed values, but in reality these numbers fluctuate based on how suppliers operate and shipping circumstances. The system also lacks real-time monitoring data from sensors, which would enhance both prediction accuracy for parts demand and categorization within circular economy principles. Future research should pursue four key areas. The framework requires validation using actual inventory information through partnerships with working airline companies. Incorporating sensor data could enable replenishment decisions based on current component conditions instead of depending entirely on past usage patterns. A blockchain-based system could facilitate secure direct exchanges of reusable and repairable components between maintenance organizations. Finally, tools for explaining model decisions are essential to show maintenance professionals the sustainability reasoning behind recommendations - without clear explanations, managers may resist adopting the system. Together, these improvements would advance the framework beyond its current experimental stage toward full operational deployment, resulting in supply chains that perform better while reducing environmental impact.