Assessing ECG Interpretation Expertise in Medical Practitioners Through Eye Movement Data and Neuromorphic Models

Abstract

This study introduces an innovative method for assessing ECG interpretation abilities in medical professionals via eye-tracking data. We examine eye movement patterns from five separate groups of cardiology practitioners utilizing a combination of neuromorphic computing models, including Spiking Neural Networks (SNN), Spiking Convolutional Neural Networks (SCNN), Recurrent Spiking Neural Networks (RSNN), and Spiking Convolutional Long Short-Term Memory (SCLSTM). Utilizing eye movement data, we analyze the skill levels of practitioners in diverse medical positions, including consultants, nurses, and technicians, during ECG evaluations. Our proposed work combines spiking neuron activations with convolutional and recurrent architectures to analyze spatial and temporal gaze patterns that reflect clinical expertise. The suggested SNN, SCNN, RSNN, and SCLSTM models attained accuracies of 84.35%, 93.04%, 94.68%, 99.76% respectively, exceeding standard machine learning approaches in both precision and recall for identifying expertise levels based on visual attention patterns. This paradigm has the potential to construct skill evaluation tools in medical education, specifically for ECG interpretation training, thereby addressing prevalent difficulties related to inconsistent ECG diagnosis methods.

Keywords

• Spiking neural networks (SNN)
• spiking convolutional neural networks (SCNN)
• recurrent spiking neural networks (RSNN)
• spiking convolutional long short-term memory (SCLSTM)
• random forest
• ECG