Goal: This paper proposes a reliable stress and relaxation level estimation algorithm that is implemented in a portable, low-cost hardware device and executed in real time. The main objective of this work is to offer an affordable and “ready-to-go” solution for medical and personal environments, in which the detection of the arousal level of a person is crucial. Methods: To achieve meaningful identification of stress and relaxation, a fuzzy algorithm based on expert knowledge is built according to parameters extracted from physiological records. In addition to the heart rate, parameters extracted from the galvanic skin response and breath are employed to extend the results. Moreover, this algorithm achieves accurate results with a restricted computational load and can be implemented in a miniaturized low-cost prototype. The developed solution includes standard and actively shielded electrodes that are connected to an Arduino device for acquisition, while parameter extraction and fuzzy processing are conducted with a more powerful Raspberry Pi board. The proposed solution is validated using real physiological registers from 42 subjects collected using BIOPAC MP36 hardware. Additionally, a real-time acquisition, processing and remote cloud storage service is integrated via IoT wireless technology. Results: Robust identification of stress and relaxation is achieved, with F1 scores of 91.15% and 96.61%, respectively. Moreover, processing is performed using a 20-second sliding window; thus, only a small frame of context is required. Significance: This work presents a reliable solution for identifying stress and relaxation levels in real time, which can lead to the production of low-cost commercial devices for use in medical and personal environments.

Relaxation helps to reduce physical, mental and emotional pressure. Relaxation techniques generally enable a person to obtain calmness and well-being by reducing stress, anxiety or anger. When a person becomes calm the body reacts physiologically, producing the so-called Relaxation Response (RResp) which affects the organism in a positive manner, no matter if it is during a state of relaxation or in the middle of a stressful period. The goal of this work is to design a system capable of identifying automatically the RResps of a subject by analyzing a single physiological signal, the Galvanic Skin Response (GSR). To do so, a team composed of psychologists, neurologists and engineers designed two experiments for inducing RResps in the participants while their GSR signals were being collected. The team analyzed the data and identified three different levels of RResp that can be quantified using only two easily calculated GSR features. Moreover, the use of the surface produced by GSR and its linear approximation is totally novel. Finally, the data were classified using Decision Tree strategies for each of the experiments and, after seeing that the obtained trees were similar, the team synthesized them in a single classification system. The F1 values obtained by the generalized classifier scored between 0.966 and 1.000 for the data collected in both experiments.

A brain-computer interface (BCI), based on motor imagery EEG, uses information extracted from the electroencephalography signals generated by a person who intends to perform any action. One of the most important issues of current research is how to detect automatically whether the user intends to send some message to a certain device. This study presents a proposal, based on a hierarchical structured system, for recognising intentional and non-intentional mental tasks on a BCI system by applying machine learning techniques to the EEG signals. First-level clustering is performed to distinguish between intentional control (IC) and non-intentional control (NC) state patterns. Then, the patterns recognised as IC are passed on to a second stage where supervised learning techniques are used to classify them. In BCI applications, it is critical to correctly classify NC states with a low false positive rate (FPR) to avoid undesirable effects. According to the literature, we selected a maximum FPR of 10%. Under these conditions, our proposal achieved an average test accuracy of 66.6%, with an 8.2% FPR, for the BCI competition IIIa dataset. The main contribution of this paper is the hierarchical approach, based on machine learning paradigms, which performs intentional and non-intentional discrimination and, depending on the case, classifies the intended command selected by the user.