Neuromorphic Edge Artificial Intelligence Architecture for Real-time Surgical Decision Support: Integrating Spiking Neural Networks with Hybrid Symbolic-neural Reasoning

Background: Real-time surgical decision-making demands systems that are both efficient and clinically interpretable. Conventional deep learning models often lack the adaptability and low latency performance required in intraoperative environments, while purely rule-based systems fail to process complex and high-dimensional data streams effectively. Methods: We present a neuromorphic edge artificial intelligence (AI) framework that integrates spiking neural networks (SNNs) with hybrid symbolic-neural reasoning for real-time surgical decision support. The system processes multimodal intraoperative data including endoscopic video, bio signals and robotic instrument telemetry into sparse spike-based representations. A three-layer SNN featuring biologically plausible learning (spike-timing-dependent plasticity) and dendritic compartmentalisation performs adaptive decision-making, while a hybrid reasoning module applies formal surgical safety constraints using answer set programming. Results: Deployed on Intel’s Loihi 2 neuromorphic chip in a simulated surgical setting, the system achieved sub-50 ms response times for critical events and reduced energy consumption by 94% compared to GPU-based models. It maintained zero safety violations across 1000 test cases and outperformed baseline convolutional neural network in adaptability, detecting novel complications within 3–10 exposures. Context-sensitive features, enabled by dendritic learning, reduced false positives by 63% during surgical phase transitions. Conclusion: This neuromorphic AI system demonstrates a significant step towards safe, efficient and explainable decision support in surgery. By combining low-power, spike-based processing with verifiable symbolic logic, the framework bridges data-driven adaptability and clinical reliability laying a foundation for the next-generation intelligent operating rooms.