Embodied General AI Require Decision Support System (DSS): An Embedded Design for Electrocardiogram (ECG) Interpretation

In this book chapter, the authors present a single-lead-based electrocardiogram (ECG) interpretation employing wavelet transformation (WT) and a decision support system (DSS). It includes the delineation of the ECG signal by identifying the QRS complex, peaks, onset and offset of the waves followed by interval detection. We conceptualized and evaluated the proposed algorithm on the PhysioNet STAFF II dataset. This database is single-lead and, unlike other databases, provides noiseless ECG signals for evaluation. The detector obtained a positive prediction of more than 99% involving more than 980,000 beats. The DSS was developed, and a comparison was drawn between the predicted ECG findings of the patients in comparison to those provided by the cardiologists. A remarkable improvement was achieved with minor prediction inaccuracies for those where T waves were not correctly identified. A rigorous sample set was collected and tested to analyse the prediction accuracy. As the wavelet transform generated the least error, the interpretation of patients’ records was also close to perfect.