Using machine learning algorithms to determine the emotional maladjustment of a person by his rhythmogram

The purpose of this study is to explore the feasibility of identifying emotional maladjustment using machine learning algorithms. Methods. Electrocardiogram data were gathered using an event-telemetry approach, employing a software and hardware setup comprising a compact wireless ECG sensor (HxM; Zephyr Technology, USA) and a smartphone equipped with specialized software.For constructing the classifier, the following algorithms were employed: logistic regression, easy ensemble, and gradient boosting. The performance of these algorithms was assessed using the f1 metric. Results. It is demonstrated that employing dynamic spectra of the original signals enhances the classification accuracy of the model compared to using the original rhythmograms. Conclusion. A method is proposed for automatically determining the level of emotional maladaptation based on an individual's cardiorhythmogram. Information from a portable heart sensor, worn by an individual, is transmitted via Bluetooth to a mobile device. Here, the level of emotional maladaptation is assessed through a pre-trained neural network algorithm. When considering a neural network algorithm, it is recommended to employ a classifier trained on spectrograms.