Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
DISCENTE: NATASHA RUSTY SILVA CARMO
DATA: 20/08/2021
HORA: 10:00
LOCAL: On-line no Google Meet: link meet.google.com/ydt-aqwg-qng
TÍTULO: machine learning techniques for detecting hypoglycemic events using electrocardiograms
PALAVRAS-CHAVES: machine learning. biosignal processing. hypoglycemia. Diabetes type-1. D1namo dataset
PÁGINAS: 65
GRANDE ÁREA: Ciências Exatas e da Terra
ÁREA: Ciência da Computação
SUBÁREA: Sistemas de Computação
ESPECIALIDADE: Arquitetura de Sistemas de Computação
RESUMO:

Background Machine learning methods have long been employed to automatically analyzeelectrocardiogram signals. In the past ten years, most studies have used a limited number of opendatabases to test their results, most of which were collected in clinical settings. The growth in thenumber of fitness trackers and other wearable devices that collect large amounts of data everyday offer a new potential to use data analysis to derive information that can improve the qualityof life for many people. Recently, an open database was released with data (electrocardiogram,respiratory rate, motion data, food intake annotations and blood glucose) from patients with type1 diabetes. It gives the opportunity to explore the potential of this data to predict hypoglycemicevents through a noninvasive method.

Methods: The study uses pre-processing techniques to clean the data and extract features fromphysiological signals in the dataset and verify how they correlate with blood glucose. Timeand frequency domain features are derived from the signal for the analysis. Automatic machinelearning is employed to determine the best classification model. The results are compared againsta 1D Convolutional Neural Network approach that automatically extracts features from individualheart beats. The final models are evaluated in regards to performance metrics (accuracy, specificityand sensitivity) with respect to their ability to predict hypoglycemic events.

Results A 10-fold cross-validation provided the following percentage values for accuracy,precision and sensitivity, respectively: 86:89  2:8, 87:03  2:7 and 86:90  2:8 for the decisiontree model and 93:00  2:3, 93:08  2:2 and 93:00  2:3 for 1D CNN. The statistical evaluationof the mean accuracy for both models from an unpaired T test returned a p-value lower than0.0001, meaning that the distributions are significantly different and 1D CNN model outperformsthe decision tree model.

Discussion and Conclusion The small number of positive samples for hypoglycemia and highdata imbalance pose a challenge to classification. It is necessary to have reasonable numberof samples from both classes to achieve classification metrics that are suitable for medicalapplications. When this condition is satisfied, data acquired from a wearable device under normalliving conditions has shown to be suitable for the task of classifying hypoglycemic events.

MEMBROS DA BANCA:
Presidente - 1692341 - EDWARD DAVID MORENO ORDONEZ
Interno - 2027113 - ANDRE BRITTO DE CARVALHO
Externo à Instituição - VICTOR MANUEL RODRIGUES ALVES