Introduction: Blood gas analysis provides essential information on acid-base and respiratory status, but interpretation requires the simultaneous assessment of several laboratory variables and their physiological relationships. This work presents an explainable biomedical decision-support pipeline designed to assist clinicians in structured blood gas interpretation while preserving transparency of the underlying reasoning.
Methods: The prototype processes laboratory parameters including pH, pCO₂, pO₂, HCO₃⁻, anion gap, and glucose. Explicit physiological and rule-based logic is applied to identify patterns consistent with respiratory acidosis, respiratory alkalosis, metabolic acidosis, and metabolic alkalosis. Oxygenation abnormalities can additionally be flagged for further clinical assessment. The system architecture separates data input, rule evaluation, interpretation generation, explanation, and clinician-facing output into distinct functional components. Rather than generating an isolated classification, the pipeline provides the detected acid-base pattern together with the laboratory variables and rules contributing to that interpretation.
Results and Discussion: A working prototype integrates laboratory-data processing, rule-based reasoning, structured interpretation, and explanation within a unified workflow. The current implementation generates interpretable outputs for the targeted acid-base disorders and presents them through a clinician-facing interface. The modular design enables individual reasoning rules and output components to be inspected and refined independently. The system is intended to support, rather than replace, professional clinical assessment. Prospective clinical validation has not yet been performed.
Conclusions: The proposed pipeline demonstrates how explicit physiological reasoning can be translated into an explainable bioengineering workflow for automated blood gas interpretation. Its modular architecture provides a basis for subsequent technical refinement and clinical evaluation.
Keywords: blood gas analysis; clinical decision support; explainability; biomedical engineering; rule-based interpretation