COVID-19 Waves and Cardiac Health: An Investigative Analysis of Creatine Phosphokinase Levels and Troponin Status Using Machine Learning.

Background: This study explores the correlation between creatine phosphokinase (CPK) levels and cardiac troponin status with eight waves of COVID-19 and identifies the most significant biomarker for assessing disease severity. Methods: Participants were selected based on confirmed COVID-19 diagnoses using RT-PCR testing. Machine learning modeling with the PyCaret autoML library established a benchmark for classification models using variables such as age, gender, serum troponin, CPK, and COVID-19 waves. Rigorous evaluation metrics were employed to assess model performance. Results: The analysis included 1975 COVID-19 patients. Patient demographics showed a shift in age and gender distribution across different waves, with later waves characterized by younger patients and a greater proportion of females. Mortality rates varied, peaking at 34.5% in the third wave and dropping to 0% in the eighth wave. CPK levels differed significantly among waves, with the third wave having the highest levels and later waves showing the lowest levels. However, troponin positivity rates did not differ significantly among waves. An extra trees classifier model achieved an overall accuracy, micro-average area under curve (AUC), sensitivity, and specificity of 0.79, 0.65, 0.79, and 0.89, respectively. CPK was identified as the most important predictor of patient outcome, followed by COVID-19 wave, age, and gender, while troponin status had the least importance. Conclusion: These findings shed light on the potential relationship between CPK, troponin, and different waves of COVID-19 and their impact on disease severity. This understanding could significantly contribute to future research and clinical practices, aiding in the management and mitigation of COVID-19's cardiac implications.