Blood glucose (BG) monitoring devices play an important role in diabetes management, offering real time BG measurements, which can be analyzed to discover new knowledge. In this paper we present a multi-patient and multivariate deep learning approach, based on Long-Short Term Memory (LSTM) artificial neural networks, for building a generalized model to forecast BG levels on a short-time prediction horizon. The proposed framework is evaluated on a clinical dataset of 17 patients, receiving care at the IRCCS Policlinico San Matteo hospital in Pavia, Italy. BG profiles collected by a flash glucose monitoring system were analyzed together with information collected by an activity tracker, including heart rate, sleep, and physical activity. Results suggest that a model with good prediction performance can be obtained and that a combination of HR and lifestyle monitoring signals can help to predict BG levels.
Scheda prodotto non validato
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo
Titolo: | Deep Learning Applied to Blood Glucose Prediction from Flash Glucose Monitoring and Fitbit Data |
Autori: | BELLAZZI, RICCARDO [Supervision] |
Data di pubblicazione: | 2020 |
Serie: | |
Abstract: | Blood glucose (BG) monitoring devices play an important role in diabetes management, offering real time BG measurements, which can be analyzed to discover new knowledge. In this paper we present a multi-patient and multivariate deep learning approach, based on Long-Short Term Memory (LSTM) artificial neural networks, for building a generalized model to forecast BG levels on a short-time prediction horizon. The proposed framework is evaluated on a clinical dataset of 17 patients, receiving care at the IRCCS Policlinico San Matteo hospital in Pavia, Italy. BG profiles collected by a flash glucose monitoring system were analyzed together with information collected by an activity tracker, including heart rate, sleep, and physical activity. Results suggest that a model with good prediction performance can be obtained and that a combination of HR and lifestyle monitoring signals can help to predict BG levels. |
Handle: | http://hdl.handle.net/11571/1365515 |
ISBN: | 978-3-030-59136-6 978-3-030-59137-3 |
Appare nelle tipologie: | 2.1 Contributo in volume (Capitolo o Saggio) |