QXAI: Explainable AI Framework for Quantitative Analysis in Patient Monitoring Systems

Black-box AI models in healthcare require explanation tools: The paper demonstrates that traditional deep learning models used for patient monitoring lack transparency, which can force clinicians to make decisions without understanding why the AI system made specific predictions—a significant concern in medical settings where trust and accountability are essential.

Explainability can be added to existing AI systems: The QXAI framework shows that post-hoc explainability methods (like Shapley values) can be applied to already-trained models to reveal which patient data features most influence predictions of vital signs like heart rate, making it possible to retrofit existing systems with transparency.

Attention mechanisms improve both performance and interpretability: By incorporating attention-based models (specifically BiLSTM), the framework achieved state-of-the-art results while simultaneously making the AI's decision-making process more transparent—suggesting that explainability and accuracy don't have to be trade-offs.

Different explanations serve different purposes in clinical practice: The paper distinguishes between "global explanations" (which features matter overall) and "local explanations" (why a specific prediction was made for an individual patient), both of which are necessary for clinicians to understand and validate AI recommendations in real patient monitoring scenarios.