Autors: Karageorgiev I., Ahmed-Shieva, S. A., Shakev, N. G., Minev I.
Title: Unsupervised Categorization of Hysteresis Loops: A Comparative Study of Expert-Driven Feature Engineering vs. Deep 1D-Convolutional Autoencoders †
Keywords: artificial intelligence, autoencoders, feature engineering, HDBSCAN, hysteresis, mechanical ventilation, signal processing, UMAP

Abstract: Hysteresis defines a system’s state as a function of its historical trajectory, manifesting as path-dependent loops where variables diverge based on the direction of change. Intrinsic to both information storage and irreversible thermodynamic dissipation, this phenomenon serves as a critical fingerprint for characterizing complex behaviors and dynamic system responses. The automated classification and labeling of dense hysteresis signals is essential for systems engineering, electronics, bio-signal processing, and meteorology. This study presents a comprehensive comparison between two unsupervised learning pipelines, to which a main objective is to discover the capabilities for phenotype discovery between automated feature extraction and expert-driven feature engineering. These methodologies are examined by clustering 7736 respiratory cycles from critical care patients who have undergone invasive mechanical ventilation. Methodology I utilizes the extraction of 17 geometric and spectral features, reduced via Principal Component Analysis (PCA) to ease the choice of high variance parameters for subsequent clustering. Methodology II proposes a Deep Learning (DL) framework utilizing a 1D-convolutional autoencoder (CAE) for automated feature discovery. Both methods utilize UMAP and HDBSCAN for the final clustering stage. After evaluation both models performed with high trustworthiness (0.9974; 0.9853) and Silhouette Scores of 0.6441 and 0.6415. Here, both methodologies obtained a surprisingly low noise ratio (1.5% and 0.74%) by the aformentioned HDBSCAN. The strategy of narrowing discrete features allows for better model interpretability, and the resulting similarity measures reasoned trough a Davies–Bouldin Index are higher. However, a peculiar difference was observed in favor to the automated extraction (CAE), where the method assessed a better quality of clustering when compared for dispersion between clusters and the dispersion of reference clusters.

References

  1. intwentytwelve Modelling-Patient-Ventilator-Asynchrony Available online: https://github.com/intwentytwelve/Modelling-patient-ventilator-asynchrony (accessed on 12 May 2026)
  2. Karageorgiev I. Ahmed-Shieva S. Shakev N. Minev I. Conveying Respiratory Care Towards Sub-symbolic Models: Data-Centric Approaches to Mitigate VILI and Patient-Ventilator Asynchrony Proceedings of the 2025 International Conference Automatics and Informatics (ICAI) Varna, Bulgaria 9–11 October 2025 283 287
  3. Zhou C. Chase J.G. Sun Q. Knopp J. Tawhai M.H. Desaive T. Möller K. Shaw G.M. Chiew Y.S. Benyo B. Reconstructing asynchrony for mechanical ventilation using a hysteresis loop virtual patient model Biomed. Eng. Online 2022 21 16 10.1186/s12938-022-00986-9 35255922
  4. Stewart G. Al-Khassaweneh M. An implementation of the HDBSCAN* clustering algorithm Appl. Sci. 2022 12 2405 10.3390/app12052405
  5. Abdi H. Williams L.J. Principal component analysis Wiley Interdiscip. Rev. Comput. Stat. 2010 2 433 459 10.1002/wics.101
  6. McInnes L. Healy J. Melville J. Umap: Uniform manifold approximation and projection for dimension reduction arXiv 2018 1802.03426
  7. Van der Maaten L. Hinton G. Visualizing data using t-SNE J. Mach. Learn. Res. 2008 9 2579 2605
  8. Ige A.O. Sibiya M. State-of-the-art in 1d convolutional neural networks: A survey IEEE Access 2024 12 144082 144105 10.1109/access.2024.3433513

Issue

Engineering Proceedings, vol. 150, 2026, Switzerland, https://doi.org/10.3390/engproc2026150001

Вид: публикация в международен форум, публикация в издание с импакт фактор, публикация в реферирано издание, индексирана в Scopus