Care pathways represent the temporal sequences of clinical events that define real-world patient journeys within complex healthcare systems. Recent advances in artificial intelligence have enabled the analysis of these pathways through sequence analytics, uncovering latent patterns beyond traditional guideline-based approaches. This narrative review synthesizes literature to examine three pillars of AI-enabled care pathway analytics: clustering methods that group similar patient trajectories, deviation detection techniques that identify meaningful variations from expected flows, and interpretability frameworks that support transparency and clinician trust.Drawing on process mining, sequence analysis, and explainable AI, the review highlights how electronic health record data can be transformed into actionable insights for clinical decision-making. Clustering approaches reveal hidden patient subgroups across domains such as oncology, cardiology, mental health, and critical care. Deviation detection methods expose bottlenecks, workarounds, and non-adherence associated with adverse outcomes and inefficiencies. Interpretability frameworks link algorithmic outputs to clinical logic, improving trust and adoption in healthcare settings.Cross-study evidence shows that while clustering and deviation detection methods have advanced significantly, their integration with interpretability remains limited, constraining large-scale implementation. The review proposes an integrative systems perspective that positions care pathway sequence analytics as a foundational component of AI-enabled healthcare infrastructure, encompassing data pipelines, model inference, intervention orchestration, and governance. Overall, AI-driven pathway analytics offers the potential to move healthcare from reactive, guideline-based care toward proactive, personalized, and continuously learning systems.
The integration of artificial intelligence into clinical decision support systems offers improved diagnostic accuracy and efficiency, but the opacity of many machine learning models raises concerns about trust, accountability, and regulatory compliance. Explainable artificial intelligence (XAI) has been proposed to address this by making model predictions interpretable to clinicians; however, its true clinical value remains uncertain, and evaluation has not kept pace with methodological development. This systematic review aimed to identify XAI methods used in clinical decision support systems, assess how they are evaluated with clinicians, and determine whether explanations improve diagnostic accuracy, trust, mental models, and efficiency. Following PRISMA guidelines, we searched PubMed, Web of Science, IEEE Xplore, ACM Digital Library, and Scopus for studies published between 2017 and 2024. Eligible studies included original research evaluating XAI in clinical decision support systems with clinician participants and reporting quantitative or qualitative outcomes. Risk of bias was assessed using adapted QUADAS-2 and ROBIS tools, and findings were synthesized narratively with subgroup analyses. From 2,847 records, 68 studies were included. The most common XAI methods were SHAP-based feature attribution (38%), saliency or heatmap methods (29%), concept-based approaches such as TCAV (15%), and counterfactual or example-based explanations (12%). Radiology was the dominant field (54%), followed by dermatology (18%) and pathology (12%). Evaluation approaches were highly inconsistent, with few validated instruments and most studies relying on Likert-scale trust measures or qualitative feedback. Only 16% of studies showed improved diagnostic accuracy with explanations, 67% showed no significant effect, and 17% reported reduced accuracy due to over-reliance or misinterpretation. Although 82% of studies reported increased clinician trust, trust rarely correlated with actual diagnostic performance. Overall, while XAI methods are widely studied in clinical decision support, their evaluation is inconsistent and their benefits are limited. Explanations tend to increase clinician trust without reliably improving diagnostic accuracy, and may sometimes worsen performance, highlighting a trust–accuracy gap that poses important safety concerns for clinical deployment.