Martin Karl Moser

Mr. Martin Karl Moser

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
6 August 2024 in Sensors
Verified
Journal Article
Open access
PDF file
MMMartin Karl Moser
Martin Karl Moser
MEMaximilian Ehrhart
Maximilian Ehrhart
Bernd ReschBernd Resch
Bernd Resch
Stress has various impacts on the health of human beings. Recent success in wearable sensor development, combined with advancements in deep learning to automatically detect features from raw data, opens several interesting applications related to detecting emotional states. Being able to accurately detect stress-related emotional arousal in an acute setting can positively impact the imminent health status of humans, i.e., through avoiding dangerous locations in an urban traffic setting. This work proposes an explainable deep learning methodology for the automatic detection of stress in physiological sensor data, recorded through a non-invasive wearable sensor device, the Empatica E4 wristband. We propose a Long-Short Term-Memory (LSTM) network, extended through a Deep Generative Ensemble of conditional GANs (LSTM DGE), to deal with the low data regime of sparsely labeled sensor measurements. As explainability is often a main concern of deep learning models, we leverage Integrated Gradients (IG) to highlight the most essential features used by the model for prediction and to compare the results to state-of-the-art expert-based stress-detection methodologies in terms of precision, recall, and interpretability. The results show that our LSTM DGE outperforms the state-of-the-art algorithm by 3 percentage points in terms of recall, and 7.18 percentage points in terms of precision. More importantly, through the use of Integrated Gradients as a layer of explainability, we show that there is a strong overlap between model-derived stress features for electrodermal activity and existing literature, which current state-of-the-art stress detection systems in medical research and psychology are based on.

16 August 2023 in IEEE Sensors Journal
Verified
Journal Article
Open access
MMMartin Karl Moser
Martin Karl Moser
Bernd ReschBernd Resch
Bernd Resch
MEMaximilian Ehrhart
Maximilian Ehrhart
Accurately measuring a person’s level of stress can have a wide variety of impacts, not only for human health, but also for the perceived feeling of safety when going after daily habits, such as walking, cycling, or driving from one place to another. While there is a vast amount of research done on stress and the related physiological responses of the human body, there is no go-to method when it comes to measuring acute stress in a live setting. This work proposes an advancement of the rule-based stress detection algorithm proposed in [1], to identify moments of stress (MOS) more reliably, through an adaptation and an individualization of the rules proposed in the original paper. The proposed algorithm leverages electrodermal activity and skin temperature, both recorded by the Empatica E4 wristband, for the assessment of an individual’s stress when exposed to an audible stimulus. The algorithm achieves an average recall of 81.31%, with a precision of 46.23%, and an accuracy of 92.74%, measured on 16 test subjects. The trade-off between precision and recall can be controlled by adjusting the MOS threshold that needs to be reached for a MOS to be detected.

9 July 2024 in Preprints.org
Verified
Preprint
Open access
PDF file
MMMartin Karl Moser
Martin Karl Moser
MEMaximilian Ehrhart
Maximilian Ehrhart
Bernd ReschBernd Resch
Bernd Resch
Stress has various impacts on the health of human beings. Recent success in wearable sensor development, combined with advancements in deep learning to automatically detect features from raw data, opens several interesting applications related to detecting emotional states. Being able to accurately detect stress-related emotional arousal in an acute setting can positively impact the imminent health status of humans, i.e., through avoiding dangerous locations in an urban traffic setting. This work proposes an explainable deep learning methodology for the automatic detection of stress in physiological sensor data, recorded through a non-invasive wearable sensor device, the Empatica E4 wristband. We propose an Long-Short Term-Memory (LSTM) network, extended through a Deep Generative Ensemble of conditional GANs (LSTM DGE) , to deal with the low data regime of sparsely labeled sensor measurements. As explainability is often a main concern of deep learning models, we leverage Integrated Gradients (IG) to highlight the most essential features used by the model for prediction and to compare the results to state-of-the-art expert-based stress detection methodologies in terms of precision, recall and interpretability. Results show that our LSTM DGE outperforms the state-of-the-art algorithm by 3 percentage points in terms of recall, and 7.18 percentage points in terms of precision. More importantly, through the use of Integrated Gradients as a layer of explainability, we show that there is a strong overlap between model-derived stress features for electrodermal activity, and existing literature, which current state-of-the-art stress detection systems in medical research and psychology are based on.

1 January 2023 in SSRN Electronic Journal
Verified
Preprint
Open access
PDF file
JMJan Mikuni
Jan Mikuni
MDMargot Dehove
Margot Dehove
LDLinda Dörrzapf
Linda Dörrzapf
MMMartin Karl Moser
Martin Karl Moser
Bernd ReschBernd Resch
Bernd Resch
HLHelmut Leder
Helmut Leder
Promoting urban well-being is a significant societal issue in the context of rapid urbanization. Past research has highlighted that interaction with urban green spaces, such as parks and forests, is key in promoting urban well-being. However, there is limited knowledge regarding the potential in promoting well-being from non-nature elements. In the present study, we explored whether interacting with art could enhance well-being in urban street contexts. In our field experiment, we built two interventions on urban streets, decorating them with either laminated art prints or green elements. We measured subjective and physiological well-being before and after the interaction with the interventions. With this paradigm, we assessed if, not only green, but also artistic interventions can improve well-being. Our results showed that, after interacting with the artistic intervention in an urban environment, the participants reported reduced feelings of anxiety, stress, and negative mood as they did with the green intervention. Further, our results indicate that improvements in well-being were linked to participants’ evaluations of the testing location (restorativeness), of aesthetic quality of the intervention (e.g., beauty, meaningfulness), and of their overall experience (e.g., enjoyment). These findings have significant implications in promoting urban well-being and city planning, as they highlight the potential of art as a novel tool for enhancing urban well-being.

10 November 2025 in Environment and Planning B: Urban Analytics and City Science
Verified
Journal Article
Open access
MMMartin Karl Moser
Martin Karl Moser
SSSebastian Schmidt
Sebastian Schmidt
DGDavid Ruben Max Graf
David Ruben Max Graf
Merve KeskinMerve Keskin
Merve Keskin
SGShaily Gandhi
Shaily Gandhi
BRBernd Resch
Bernd Resch
The complexity of environmental factors experienced in active mobility presents unique challenges for the design of sustainable urban mobility environments. Particularly, active mobility modes are frequently associated with increased stress and unsafety. Most studies apply qualitative assessment methods for evaluating cyclists’ stress levels and subjective cycling experiences. Quantitative approaches are either limited in sample size, or conducted over short periods of time. This study introduces a transferable methodology that combines physiological measurements from wearable sensors with openly available spatial data to assess environmental stressors in urban cycling. A field study was conducted in Osnabrück, Germany, and involved 89 participants, 1,780 cycling trips, and 2,104,109 geo-referenced data points. Stress levels were quantified through processed Electrodermal Activity (EDA) measurements to identify Moment of Stress (MOS) along mapped road segments. We derived features from OpenStreetMap (OSM), Sentinel-2 Remote Sensing (RS), and Mapillary Street View Imagery (SVI) to characterise spatial elements of the built and natural environment. Using feature importance methods on top of a Random Forest (RF) Machine Learning (ML) model, we identified key environmental aspects associated with cyclists’ stress. Results show that the availability of cycling infrastructure, traffic regulations and other road users, are of higher importance than the availability of green space, when it comes to predicting the stress potential of individual road segments. The proposed methodology offers a multi-faceted and extensible approach to evaluate environmental characteristics related to stress, providing information for creating safer and more comfortable cycling environments. While our approach investigated spatiotemporal stress factors in cycling, the use and the availability of open data sources restricts the feature set that can be derived and evaluated in a particular region. We encourage future research to apply and extend this approach in diverse urban contexts, incorporating temporally dynamic features to support evidence-based mobility planning.

23 December 2025 in ACM Transactions on Spatial Algorithms and Systems
Verified
Journal Article
Open access
DGDavid Ruben Max Graf
David Ruben Max Graf
MMMartin Karl Moser
Martin Karl Moser
Shaily GandhiShaily Gandhi
Shaily Gandhi
MBMarco Badici
Marco Badici
Bernd ReschBernd Resch
Bernd Resch
Sustainable and active urban mobility are critical for addressing the pressing challenges of climate change, urban congestion, and public health in cities. External conditions caused by the spatiotemporal nature of urban environments pose stressful factors hindering the adoption of active modes of transportation, and are mainly evaluated through qualitative feedback, or labour-intensive workflows processing multi-modal information. To investigate stress caused by static and dynamic elements of the natural and built environment, experienced while cycling in urban areas, we propose a multi-modal, scalable methodology that combines wearable sensors and cyclist’s POV videos. By spatially aggregating EDA based stress responses, captured by wearable sensors, we are able to identify spatial clusters of stress, which are used as label class to alleviate issues concerning spatial autocorrelation. Combining this with CV models to automatically process synchronously recorded cyclist’s point of view videos, captured by low-cost camera devices mounted on the bikes (chest of the cyclist or handlebar of the bicycle), we derive spatiotemporal elements of the natural and built environment and link them to measured stress. We show that the confidence threshold for semantic segmentation can be used to handle obstructed areas, caused by improper camera positioning, and that CV-based urban features are sufficient to predict stressful urban areas. Our model demonstrates strong performance in classifying spatial clusters of stress, achieving an accuracy of 0.745 and a recall of 0.615 when distinguishing between hot spots (areas with concentrated stress measurements) and cold spots (regions with clustered non-stress measurements).