DeStress: Deep Learning for Unsupervised Identification of Mental Stress in Firefighters from Heart-rate Variability (HRV) Data

TL;DR

This study evaluates unsupervised deep learning methods to detect mental stress in firefighters using heart rate variability data, finding autoencoder-based approaches effectively distinguish stressed from normal states validated by physiological markers.

Contribution

It introduces and compares autoencoder-based unsupervised methods for stress detection from HRV data, demonstrating their effectiveness over traditional clustering with engineered features.

Findings

Autoencoder methods outperform K-Means with engineered features.

Convolutional autoencoders reliably identify stress-related clusters.

Clusters correlate with established physiological stress markers.

Abstract

In this work we perform a study of various unsupervised methods to identify mental stress in firefighter trainees based on unlabeled heart rate variability data. We collect RR interval time series data from nearly 100 firefighter trainees that participated in a drill. We explore and compare three methods in order to perform unsupervised stress detection: 1) traditional K-Means clustering with engineered time and frequency domain features 2) convolutional autoencoders and 3) long short-term memory (LSTM) autoencoders, both trained on the raw RRI measurements combined with DBSCAN clustering and K-Nearest-Neighbors classification. We demonstrate that K-Means combined with engineered features is unable to capture meaningful structure within the data. On the other hand, convolutional and LSTM autoencoders tend to extract varying structure from the data pointing to different clusters with different sizes of clusters. We attempt at identifying the true stressed and normal clusters using the HRV markers of mental stress reported in the literature. We demonstrate that the clusters produced by the convolutional autoencoders consistently and successfully stratify stressed versus normal samples, as validated by several established physiological stress markers such as RMSSD, Max-HR, Mean-HR and LF-HF ratio.