Validation of LoggerOne system’s 5-lead ECG

• Autor(a) principal: Daniel, Lisana Moniz de Sousa
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.6/13884
• topic: 5 Derivações; Ecg; Extração de Características de Ecg; Processamento de Sinal; Validação; Domínio/Área Científica::Engenharia e Tecnologia::Bioengenharia

Validation studies are becoming increasingly crucial as technology advances and the number of electrocardiography (ECG) wearables grows. Existing validation studies, however, are limited and particular to the approach in question. The primary purpose of this dissertation is to provide a standard validation method that can be used for a wide range of validation studies. To that purpose, the LoggerOne system will be compared against the COMEN system, an established gold standard. ECG signals will be recorded using both devices at the same time, with the assistance of an ECG signal simulator. The simulator used is a CE certified device that can simulate both pediatric and adult ECG signals, as well as different types of arrhythmia’s. The recorded signals will be processed using Python’s ECG and data manipulation toolboxes. Prior to validation, signals from both devices will be normalized, and a number of filters and algorithms for feature extraction for each lead will be tested. The optimized algorithms will then extract relevant signal properties such as the P and T wave, QRS complex, RR, PR, ST, and QT intervals, and ST and PR segments. This intervals and segments from each device will be analyzed using statistical analysis methods such as Bland-Altman limits of agreement (LoA), intraclass correlation coefficient (ICC), and morphological comparison to validate the device. This work results can be divided into two components: algorithm optimization results and validation results. Regarding algorithm results, only some leads presented P wave bounds and the T onset with values of sensitivity (Se) and positive predictive value (+P) below 95% , whereas most results present values above 98%. Standard deviation (SD) values for P bounds and QRS onset exceeded the reference values in some leads, whereas all other values remained within the range of reference. The mean values remained below the reference value imposed. Regarding validation results, almost all results for each lead show good agreement and reliability, except for intervals and segments computed using the T onset.