EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Wearable Sensor-Based Human Activity Recognition: Performance and Interpretability of Dynamic Neural Networks

Dalius Navakauskas iD, Martynas Dumpis iD

Sensors · 16.07.2025 · 25 (14) · 4420

DOI: 10.3390/s25144420 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 16.07.2025. Verfasst von Dalius Navakauskas und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electronic Systems, Vilnius Gediminas Technical University, Plytines g. 25-234, LT-10105 Vilnius, Lithuania. Hauptthemen laut MEDLINE: Wearable Electronic Devices, Neural Networks, Computer, Human Activities. Die Arbeit wurde bisher 28-mal zitiert (Crossref) und nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: New Generation Lithuania / Ministry of the Economy and Innovation of the Republic of Lithuania, Economic Revitalization and Resilience Enhancement Plan “New Generation Lithuania”.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Human Activity Recognition (HAR) using wearable sensor data is increasingly important in healthcare, rehabilitation, and smart monitoring. This study systematically compared three dynamic neural network architectures—Finite Impulse Response Neural Network (FIRNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU)—to examine their suitability and specificity for HAR tasks. A controlled experimental setup was applied, training 16,500 models across different delay lengths and hidden neuron counts. The investigation focused on classification accuracy, computational cost, and model interpretability. LSTM achieved the highest classification accuracy (98.76%), followed by GRU (97.33%) and FIRNN (95.74%), with FIRNN offering the lowest computational complexity. To improve model transparency, Layer-wise Relevance Propagation (LRP) was applied to both input and hidden layers. The results showed that gyroscope Y-axis data was consistently the most informative, while accelerometer Y-axis data was the least informative. LRP analysis also revealed that GRU distributed relevance more broadly across hidden units, while FIRNN relied more on a small subset. These findings highlight trade-offs between performance, complexity, and interpretability and provide practical guidance for applying explainable neural wearable sensor-based HAR.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Wearable Electronic DevicesNeural Networks, ComputerHuman ActivitiesHumansAlgorithmsBiosensing Techniques

Stichwörter der Autor:innen

GRULSTMdynamic neural networksfinite impulse response neural networkhuman activity recognitionlayer-wise relevance propagationwearable sensors

Autor:innen und Institutionen

  • Department of Electronic Systems, Vilnius Gediminas Technical University, Plytines g. 25-234, LT-10105 Vilnius, Lithuania.

ORCID-Profile: Dalius Navakauskas, Martynas Dumpis

Förderung

  • New Generation Lithuania / Ministry of the Economy and Innovation of the Republic of Lithuania
  • Economic Revitalization and Resilience Enhancement Plan “New Generation Lithuania”

Zitieren

APA

Navakauskas, D., & Dumpis, M. (2025). Wearable Sensor-Based Human Activity Recognition: Performance and Interpretability of Dynamic Neural Networks. Sensors, 25(14), 4420. https://doi.org/10.3390/s25144420

Vancouver

Navakauskas D, Dumpis M. Wearable Sensor-Based Human Activity Recognition: Performance and Interpretability of Dynamic Neural Networks. Sensors. 2025;25(14):4420. doi:10.3390/s25144420

BibTeX
@article{Navakauskas2025,
  author = {Navakauskas, Dalius and Dumpis, Martynas},
  title = {Wearable Sensor-Based Human Activity Recognition: Performance and Interpretability of Dynamic Neural Networks},
  journal = {Sensors},
  year = {2025},
  volume = {25},
  number = {14},
  pages = {4420},
  doi = {10.3390/s25144420}
}
Kostenlos starten →