Mental stress detection using bioradar respiratory signals

Highlights

• Mental stress in humans can be identified using bioradars.

• The use of respiratory patterns leads to an accurate assessment of mental health.

• Respiratory patterns can be acquired using a non-contact concealed mechanism.

• Recurrence Quantification Analysis features provide useful information when applied to respiratory patterns.

Abstract

Stress detection techniques seek to provide an accurate assessment of mental health. This paper presents a new stress detection technique based on signals taken by a bioradar. The main advantage of this approach is its non-invasive nature since it uses a non-contact concealed mechanism that does not require the direct interaction between the person and the measuring device. In addition to being one of the first solutions based exclusively on respiratory signals, the novelty of the research also lies in the use of Recurrence Quantification Analysis (RQA) features on respiratory recordings. The RQA features, traditionally applied for heart rate measurements, allowed reaching a precision of 94,4% after the leave-one-subject-out-cross-validation using a multi-layer perceptron.

Introduction

Mental stress is one of the fundamental problems of today’s society. The high workloads and the variability of modern life, among other factors, disturb people’s mental health, causing different effects. Previous studies have reported mental stress as one of the major contributing factors leading to various diseases such as depression [1], stroke [2], sleep disorders [3], heart attack and cardiac arrest [4,5].

Although there are numerous techniques for the treatment and prophylaxis of this condition, most assume that an expert examiner, the patient or people close to him, detect symptoms of stress. However, sometimes individuals who are under stress show no signs of their condition, or exhibit signs which are hard to notice.

Stress detection techniques are helpful tools in such situations since they facilitate the monitoring of stress levels. These methods are generally based on the processing of physiological signals such as: blood pressure [6], heart rate [6], heart rate variability (HRV) [7], skin conductance [8,9], level of cortisol [10,11], pupil diameter [12] and facial gestures [13], among others. The objective of these methods is to detect stress with a high level of accuracy, causing a minimum of discomfort to the patient.

This paper introduces a new method for the detection of mental stress using only respiratory patterns taken by a bioradar, avoiding the introduction of more complicated vital signs. One of the main advantages of the approach is its non-invasive nature, since the signals are taken by a non-contact method that establishes no direct physical interaction between the individual and the measuring device. Non-contact methods have experienced a growing interest by the scientific community thanks to the high acceptance of patients [14].

Even when samples were taken by a fixed laboratory equipment, there are modern implementations of the technology in small size devices [15,16]. In addition, the processed signals contained variations in the mean level and exhibited marked differences in the breathing rate from one subject to another, which supports the idea that the solution may be applicable in real environments.

Also, it should be noted that, until present work, heartbeat signal analysis had been mainly used for detecting mental stress by means of Doppler radar. For example, in [17] it was shown that the mean heartbeat frequency could be used to estimate examineeś reaction to mental stress. However, the accurate detection of heartbeat patterns using bioradars may be quite challenging, if not an impossible task, for subjects with body mass index greater than 25, whereas respiratory patterns can be reliably detected. This makes the proposed method applicable to a wider range of users.