Analysis of heart rate variability to understand the effect of cannabis consumption on Indian male paddy-field workers
Analysis of heart rate variability to understand the effect of cannabis consumption on Indian male paddy-field workers
复制标题
DOI：
10.1016/j.bspc.2020.102072
复制
发表时间：
2020-09-01
影响因子：
5.1
通讯作者：
Pal, Kunal
中科院分区：
文献类型：
--
作者：
Nayak, Suraj K.;Pradhan, Bikash K.;Pal, Kunal
由于大麻产品的欣快作用，全球范围内大麻产品的消费量日益增加。许多研究报告了吸食大麻的人心血管疾病的发病率甚至死亡率。然而，人们并没有太多关注大麻引起的自主神经系统（ANS）活动的改变，而这有助于心血管疾病的早期诊断。目前的研究通过心率变异性 (HRV) 分析调查了 200 名印度男性志愿者因食用大麻（一种大麻产品）而导致的 ANS 活动的变化。结果表明，食用大麻的人群心率变异性降低，交感神经优势增强，副交感神经活动相应减少，这可能导致各种心血管疾病。这些推论可以作为建议人们停止消费大麻的证据。该研究进一步提出了一种机器学习模型，用于自动识别大麻消费人群。 HRV 参数采用基于权重的特征排序和降维方法，为机器学习模型选择合适的输入。在比较朴素贝叶斯（NB）、广义线性模型（GLM）、线性回归（LR）、快速大裕度（FLM）、深度学习（DL）、决策树（DT）、随机森林（RF）、梯度提升树（GBT）和支持向量机（SVM）的性能后，最终选择GBT模型作为最佳模型。 (C) 2020 Elsevier Ltd. 保留所有权利。
The consumption of cannabis-based products is increasing worldwide day-by-day because of their euphoric effects. Numerous studies have reported the incidence of cardiovascular diseases and even mortality in people consuming cannabis. However, not much attention has been paid to understand the cannabis-induced alteration in the autonomic nervous system (ANS) activity, which can help in the early diagnosis of cardiovascular diseases. The current study investigated the alteration in the ANS activity of 200 Indian male volunteers due to the consumption of bhang (a cannabis-based product) using heart rate variability (HRV) analysis. The results suggested a reduction in the variability of the heart rate, increased sympathetic dominance, and a corresponding reduction in the parasympathetic activity in the bhang consuming population, which may lead to various cardiovascular diseases. These inferences can act as evidence for counseling people to stop consuming cannabis. The study further proposes a machine learning model for automated identification of the bhang consuming population. The HRV parameters were subjected to weight-based feature ranking and dimension reduction methods to select suitable inputs for the machine learning models. After comparing the performances of the Naive Bayes (NB), Generalized Linear Model (GLM), Linear Regression (LR), Fast Large Margin (FLM), Deep Learning (DL), Decision Tree (DT), Random Forest (RF), Gradient Boosted Tree (GBT), and Support Vector Machine (SVM), a GBT model was finally chosen as the best model. (C) 2020 Elsevier Ltd. All rights reserved.