Abstract:
Accurate disease diagnosis is the cornerstone of healthcare, and predicting cardio-vascular events and cerebrovascular complications in hypertensive patients is crucial for preventing the progression of heart disorders. Existing gold-standard methods can forecast vascular event likelihood, but often fall short in clinical efficacy due to limitations in accuracy and generalizability. This study proposes a novel machine-learning approach combining Heart Rate Variability (HRV) and Wavelet Transform (WT) to identify hypertensive patients at higher risk of cardiac incidents. We extracted HRV features from patient data in the afternoon (12:00 pm-18:00 pm) using time-domain and frequency-domain analysis, and then applied the daubuche (db1) wavelet transform to extract energy features. A random under-sampling boosting (RUSboost) model was trained using various feature combinations, achieving maximum accuracy with the inclusion of the energy and demographic features. Our results demonstrate the potential of integrating Wavelet Transform characteristics, Heart Rate Variability, and Machine Learning models to predict vascular events in hypertension patients. This approach offers a simple yet effective prediction strategy that supports the practitioner decision-making and surpasses existing methodologies. By improving risk stratification, our method can facilitate early interventions, reduce cardiovascular morbidity and mortality, enhance patient outcomes, and ultimately lead to better healthcare resource allocation and improved patient care.
