Heart Rate Information-Based Machine Learning Prediction of Emotions Among Pregnant Women

• DOI: 10.3389/fpsyt.2021.799029
• Publication Date: 2022-01-27T15:56:38Z
• Authors: Xue Li; Chiaki Ono; Noriko Warita; Tomoka Shoji; Takashi Nakagawa; Hitomi Usukura; Zhiqian Yu; Yuta Takahashi; Kei Ichiji; Norihiro Sugita; Natsuko Kobayashi; Saya Kikuchi; Yasuto Kunii; Keiko Murakami; Mami Ishikuro; Taku Obara; Tomohiro Nakamura; Fuji Nagami; Takako Takai; Soichi Ogishima; Junichi Sugawara; Tetsuro Hoshiai; Masatoshi Saito; Gen Tamiya; Nobuo Fuse; Shinichi Kuriyama; Masayuki Yamamoto; Nobuo Yaegashi; Noriyasu Homma; Hiroaki Tomita
• Keywords: Matthews correlation coefficient

In this study, the extent to which different emotions of pregnant women can be predicted based on heart rate-relevant information as indicators autonomic nervous system functioning was explored using various machine learning algorithms. Nine indicators, including coefficient variation R-R interval (CVRR), standard deviation all NN intervals (SDNN), and square root mean squared differences successive (RMSSD), were measured a rate monitor (MyBeat) four "happy," positive emotion "anxiety," "sad," "frustrated," negative self-recorded smartphone application, during 1 week starting from 23rd 32nd weeks pregnancy 85 women. The k-nearest neighbor (k-NN), support vector (SVM), logistic regression (LR), random forest (RF), naïve bayes (NB), decision tree (DT), gradient boosting trees (GBT), stochastic descent (SGD), extreme (XGBoost), artificial neural network (ANN) methods applied predict information. To emotions, RF also showed modest area under receiver operating characteristic curve (AUC-ROC) 0.70. CVRR, RMSSD, SDNN, high frequency (HF), low (LF) mostly contributed predictions. GBT displayed second highest AUC (0.69). Comprehensive analyses revealed benefits prediction accuracy beneficial establish models indicators. results implicated LF, HF important parameters for