Real-time physiological stress monitoring would be a relevant addition to virtual reality (VR) training for high-risk professions, such as the military. VR is highly suitable for the implementation of such monitoring due to the controlled environment and the already used wearables. However, physiological stress measurements suffer from distortion due to physical activity. Therefore, we tested whether we can use accelerometry to correct non-invasively measured heart rate (HR) for physical activity in 23 soldiers who performed three room-clearing VR scenarios. These scenarios were dynamic, in that soldiers moved around in the VR environment by walking around in the real environment. In contrast to uncorrected HR, and HR corrected by subtracting baseline HR measured when walking, the accelerometry-corrected HR was able to significantly predict the participants’ self-reported stress in the scenarios, p = 0.047, R ² = 0.11. Whereas uncorrected HR significantly predicted self-reported physical demand, p = 0.028, R ² = 0.09, the accelerometry-corrected HR did not. All HR measures significantly predicted self-reported mental effort, which was most strongly the case for uncorrected HR, p < 0.001 R ² = 0.42. These findings, in combination with the methods’ low sensitivity to motion artifacts and non-invasiveness, are very promising for its use to monitor stress in real-time during dynamic VR training scenarios.

Early detection of exposure to a toxic chemical, e.g., in a military context, can be life-saving. We propose to use machine learning techniques and multiple continuously measured physiological signals to detect exposure, and to identify the chemical agent. Such detection and identification could be used to alert individuals to take appropriate medical counter measures in time. As a first step, we evaluated whether exposure to an opioid (fentanyl) or a nerve agent (VX) could be detected in freely moving guinea pigs using features from respiration, electrocardiography (ECG) and electroencephalography (EEG), where machine learning models were trained and tested on different sets (across subject classification). Results showed this to be possible with close to perfect accuracy, where respiratory features were most relevant. Exposure detection accuracy rose steeply to over 95% correct during the first five minutes after exposure. Additional models were trained to correctly classify an exposed state as being induced either by fentanyl or VX. This was possible with an accuracy of almost 95%, where EEG features proved to be most relevant. Exposure detection models that were trained on subsets of animals generalized to subsets of animals that were exposed to other dosages of different chemicals. While future work is required to validate the principle in other species and to assess the robustness of the approach under different, realistic circumstances, our results indicate that utilizing different continuously measured physiological signals for early detection and identification of toxic agents is promising.