- Toward electrocardiogram acquisition via a novel wearable device and analysis via artificial intelligence
- Neri, Luca <1982>
Subject
- ING-INF/06 Bioingegneria elettronica e informatica
Description
- There is a fast-growing interest in several aspects of wearables and artificial intelligence algorithms applied to an electrocardiogram (ECG) as monitoring and diagnostic tools, showing the potential to augment the patient’s data analysis and enhance healthcare. This Ph.D. thesis aimed to improve the features of a novel smart t-shirt by developing new algorithms applied to the ECG for signal processing and disease detection and prediction and testing its performance compared to gold standard devices to understand its potential and possible limitations. The advantages and limitations of combining wearable devices with artificial intelligence were studied when acquiring and analyzing the ECG. The most significant limitations are the relatively low use of wearables for research purposes due to the preference for publicly available databases and the lack of standardization in using performance measurements to compare research results. Due to the importance of the QRS complex detection for ECG analysis and feature extraction, we developed and tested a novel algorithm for its detection. We obtain a more efficient and accurate solution to serve wearable and mobile applications better. An innovative artificial intelligence algorithm for predicting sudden cardiac arrest was developed. Using a public dataset, a deep learning model could predict which patients will suffer cardiac arrest. The promising results are valid with both 12 leads and a single-lead ECG. Finally, a clinical trial revealed the features and limitations of the ECG signal acquired by the smart t-shirt when compared with a gold-standard Holter monitor, helping to address the next developments. Our research highlighted the potential of wearable devices in combination with dedicated signal processing and the application of artificial intelligence for detecting and predicting diseases. Future research will be dedicated to improving wearables ECG signal acquisition and artificial intelligence algorithms and collecting more specific and high-quality datasets.
Date
- 2024-03-27
- info:eu-repo/date/embargoEnd/2025-03-31
Type
- Doctoral Thesis
- PeerReviewed
Format
- application/pdf
Identifier
urn:nbn:it:unibo-30292
Neri, Luca (2024) Toward electrocardiogram acquisition via a novel wearable device and analysis via artificial intelligence, [Dissertation thesis], Alma Mater Studiorum Università di Bologna. Dottorato di ricerca in Scienze e tecnologie della salute