Comprehensive Summary
The heart and brain are connected in a complex and dynamic network, where the relay of signals between the two is known as autonomic regulation. This study investigated the ability of electrocardiogram (ECG) signals, which measure heart rhythm, to predict the onset of seizures in epileptic individuals. Over 1,000 second-long ECG recordings from healthy and epileptic individuals were processed through the Pan and Tompkins algorithm for QRS complex detection, which removed noise, normalized signal amplitude, compared signal morphology, and classified them by the Multi-Layer Perceptron (MLP) neural network configuration. In individuals with epilepsy, ECG recordings revealed cardiac signals with increases in heart rate and deformations in cycle frequency and amplitude. The median-based method for variance, skewness, and kurtosis yielded a 98.01% accuracy of ECG signals to proactively and correctly detect brain activity fluctuations, 97.01% sensitivity (to rule out false negatives/precisely identify healthy individuals), and 98.01% specificity (to confirm diagnoses). The use of MLP and median-based methods was highlighted as the most effective system compared to previous studies and their classification methods. In conclusion, these results indicate a clear relationship between cardiac cycles and brain activity in epileptic individuals.
Outcomes and Implications
With over fifty million people affected by epilepsy, utilizing ECG as a tool for detecting seizures provides patients with a more accessible, affordable, and reliable method of brain activity surveillance. Additionally, continuous monitoring and proactive detection afford patients time to respond before an epileptic episode. The proposed next steps are to conduct a study across a more diverse patient population, allowing for the formation of a more resilient network. The authors also highlight the need for establishing an adaptive sampling mechanism based on physical activity, such as multimodal sensors to recognize periods of intense activity and ignore those sudden cardiac changes, incorporating patient history, and the overall implementation and durability of the ECG as a wearable device.