Exploring different feature selection methods and developing more advanced models with higher reliability

The gap

Evidence profile

Sourced from the future-work section and future work of the source papers, classified as general, drawn from work published between 2025 and 2026, spanning 2 journals. Those papers have been cited 15 times in total.

Supporting evidence — 3 representative gaps

• Groundwater Potential Mapping Using Optimized Decision Tree-Based Ensemble Learning Model with Local and Global Explainability (2025) · Water · cited 15× · doi

  exploring different feature selection methods and developing more advanced models with higher reliability, - using different explainability strategies to increase the explainability of ML models, - comparing results with groundwater levels derived from satellite imagery, - assessing the impact of LULC change, land surface temperature, and climate change on groundwater potential

  generalfuture-work section
  Keywords: exploring different feature selection methods developing advanced models
• Groundwater Potential Mapping Using Machine Learning Techniques: Current Trends and Future Perspectives (2026) · Water · doi

  ML has demonstrated strong potential in advancing GPM by improving predictive accuracy and integrating multi-source geospatial data. However, to transition from ex- perimental research toward operational and decision-support tools for groundwater man- agement, several future directions must be pursued. These directions address current methodological gaps and emerging technologies that can enhance the robustness, inter- pretability, and transferability of ML-based groundwater models. 5.1. Integration of Explainable Artificial Intelligence (XAI) Future GPM studies should emphasize model transparency and interpretability through the integration of explainable artificial intelligence (XAI) techniques. Meth- ods such as SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model- agnostic Explanations), Partial Dependence Plots (PDP), and Accumulated Local Effects (ALE) can provide quantitative insights into how input variables influence model pre- dictions. These tools not only enhance scientific understanding of groundwater systems but also build confidence among policymakers and hydrogeologists by connecting ML outcomes with hydrogeological principles. Standardizing XAI frameworks across GPM studies will help improve model interpretability and support decision-making processes in groundwater management. 5.2. Development of Hybrid and Physics-Informed Models One promising future direction is the integration of ML with physically based hydro- logical models, forming hybrid or physics-informed learning frameworks. While traditional ML models excel at pattern recognition, they lack explicit representation of groundwater flow and recharge dynamics. Combining ML with numerical models (e.g., MODFLOW, SWAT) or embedding physical constraints within the learning process can enhance model reliability and extrapolation capacity. For example, coupling Random Forest or XGBoost with Darcy’s law-based simulation results can provide physically meaningful predictions that respect mass balance and aquifer properties. Such hybrid models represent a key step toward scientifically interpretable and policy-relevant GPM frameworks. Recent advances in groundwater potential mapping research highlight several emerg- ing directions that may significantly improve prediction reliability and practical applicabil- ity. One promising development is the integration of physics-informed machine learning, in which hydrological or hydrogeological constraints are incorporated into data-driven models to improve their physical consistency and interpretability. Another important research direction involves uncertainty quantification, which aims to evaluate the reliability and robustness of model predictions by analyzing the influence of data quality, model parameters, and algorithmic assumptions. In addition, the transferability of groundwater potential models across different hydrogeological regions remains an important challenge, https://doi.org/10.3390/w18080947 Water

  generalfuture workevidence 5/5
  Keywords: models groundwater model integration potential future directions enhance based interpretability hydrogeological frameworks improve hybrid physics
• AI-Integrated LC–MS/MS Bioanalytical Platforms for Matrix-Effect Mitigation: Advances in Real-Time Monitoring, Sample Preparation, and Validation Challenges (2026) · Journal of Pharmaceutical Innovation · doi

  As analytical technologies continue to evolve, the integration of artificial intelligence into pharmaceutical bioanalysis is expected to move beyond supportive roles toward fully auton- omous and adaptive systems [38, 334]. While current advance- ments have demonstrated significant potential, the next phase of development will focus on improving transparency, scalabil- ity, sustainability, and system-level integration [335–337]. Explainable AI (XAI) One of the most pressing challenges in AI-driven bioanal- ysis is the lack of transparency in model decision-making Page 15 of 27 604 [30]. Traditional machine learning models, particularly deep learning systems, often function as “black boxes,” making it difficult to understand how outputs are generated. This lack of interpretability poses a barrier to regulatory acceptance and limits user confidence [189, 297]. Explainable AI (XAI) aims to address this issue by devel- oping models that provide clear and interpretable insights into their decision processes [189]. In the context of phar- maceutical analysis, XAI can help analysts understand how specific variables influence outcomes such as impurity detection or quantification accuracy [285]. This improved transparency is expected to play a key role in bridging the gap between advanced AI technologies and regulatory requirements [338]. Miniaturized and AI-assisted analytical workflows may additionally support reduced solvent and sample consump- tion [38, 261]. However, future development priorities in regulated bioanalysis will primarily depend on validation reliability, model transparency, and regulatory compliance rather than sustainability alone [24, 37].

  generalfuture workevidence 5/5
  Keywords: transparency regulatory analytical technologies integration bioanalysis expected systems development sustainability explainable lack model decision making

Questions about this gap