Assessment of the Impact of Alcohol Consumption Patterns on Heart Rate Variability by Machine Learning in Healthy Young Adults.

Autonomic nervous system (ANS) dysfunction is present in early stages of alcohol abuse and increases the likelihood of cardiovascular events. Given the nonlinear pattern of dynamic interaction between sympathetic nervous system (SNS) and para sympathetic nervous system (PNS) and the complex relationship with lifestyle factors, machine learning (ML) algorithms are best suited for analyzing alcohol impact over heart rate variability (HRV), because they allow the analysis of complex interactions between multiple variables. This study aimed to characterize autonomic nervous system dysfunction by analysis of HRV correlated with cardiovascular risk factors in young individuals by using machine learning. Total of 142 young adults (28.4 ± 4.34 years) agreed to participate in the study. Alcohol intake and drinking patterns were assessed by the AUDIT (Alcohol Use Disorders Identification Test) questionnaire and the YAI (Yearly Alcohol Intake) index. A short 5-min HRV evaluation was performed. Post-hoc analysis and machine learning algorithms were used to assess the impact of alcohol intake on HRV. Binge drinkers presented slight modification in the frequency domain. Heavy drinkers had significantly lower time-domain values: standard deviation of RR intervals (SDNN) and root mean square of the successive differences (RMSSD), compared to casual and binge drinkers. High frequency (HF) values were significantly lower in heavy drinkers ( = 0.002). The higher low-to-high frequency ratio (LF/HF) that we found in heavy drinkers was interpreted as parasympathetic inhibition. Gradient boosting machine learner regression showed that age and alcohol consumption had the biggest scaled impact on the analyzed HRV parameters, followed by smoking, anxiety, depression, and body mass index. Gender and physical activity had the lowest impact on HRV. In healthy young adults, high alcohol intake has a negative impact on HRV in both time and frequency-domains. In parameters like HRV, where a multitude of risk factors can influence measurements, artificial intelligence algorithms seem to be a viable alternative for correct assessment.