Isil Ozkan
Research Subjects
Isil Ozkan has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Abdullah Omer Ebeoglu
Department of Cardiology, Istanbul University-Cerrahpaşa, Institute of Cardiology, Turkey
1 shared publication - Hasan Ali Barman
Department of Cardiology, Istanbul University-Cerrahpasa, Institute of Cardiology, Istanbul, Turkiye
1 shared publication - Huseyin Uvet
Yildiz Technical University
1 shared publication - Ibrahim Cem Balci
Yildiz Technical University
1 shared publication - Ipek Aydin
Department of Cardiology, Istanbul University-Cerrahpasa, Institute of Cardiology , Cerrahpasa Mah., Org. Abdurrahman Nafiz Gurman Cd., Bina No. 24, Istanbul 34098
1 shared publication - Irem Sayin
Yildiz Technical University,Department of Mechatronics Engineering,Istanbul,Turkey
1 shared publication
Latest Publications
- 22 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 449-453
Heart Rate-Based Prediction of Daily Activity Levels Using Wearable Devices for Home Rehabilitation Process of Cardiac Patients
This study aims to contribute to the literature by developing an advanced Long Short-Term Memory (LSTM) model that classifies and predicts the duration of patients' daily physical activities (resting, light, moderate, intense) using heart rate data. This approach is an innovative step towards personalizing cardiac rehabilitation programs and accurately monitoring patients' daily activity levels. The application of this model aims to allow healthcare professionals to objectively track patients' activity levels. Data collection was conducted at the Istanbul University - Cerrahpaşa, Cardiology Institute exercise test laboratory during the years 2023–2024. Heart rate (bpm) of the patients was recorded through a Garmin watch worn on the wrist during exercise tests conducted using the Modified Bruce Protocol. The study included patients aged between 18 and 85 with medium-risk pre-test probability for coronary artery disease. The collected data underwent preprocessing, including synchronization of heart rate data with exercise test records and removal of inconsistent or illogical data points. LSTM model was chosen for analyzing sequential data related to heart rate over time. The model consisted of two LSTM layers and one output layer, achieving a prediction accuracy of 59.1% on the test dataset. Further efforts are underway to enhance the model's performance within the project scope. This research underscores the importance of objective activity tracking and highlights the potential to fill existing gaps in this field. The developed model, based on heart rate data collected via wearables, is expected to serve as an objective tool for monitoring cardiac rehabilitation processes and daily activity levels. By evaluating data collected through wearables, the model facilitates the calculation of individuals' durations of rest, light, moderate, and high-intensity activities throughout the day, contributing to personalized health care practices.
- 22 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 449-453
Heart Rate-Based Prediction of Daily Activity Levels Using Wearable Devices for Home Rehabilitation Process of Cardiac Patients
This study aims to contribute to the literature by developing an advanced Long Short-Term Memory (LSTM) model that classifies and predicts the duration of patients' daily physical activities (resting, light, moderate, intense) using heart rate data. This approach is an innovative step towards personalizing cardiac rehabilitation programs and accurately monitoring patients' daily activity levels. The application of this model aims to allow healthcare professionals to objectively track patients' activity levels. Data collection was conducted at the Istanbul University - Cerrahpaşa, Cardiology Institute exercise test laboratory during the years 2023–2024. Heart rate (bpm) of the patients was recorded through a Garmin watch worn on the wrist during exercise tests conducted using the Modified Bruce Protocol. The study included patients aged between 18 and 85 with medium-risk pre-test probability for coronary artery disease. The collected data underwent preprocessing, including synchronization of heart rate data with exercise test records and removal of inconsistent or illogical data points. LSTM model was chosen for analyzing sequential data related to heart rate over time. The model consisted of two LSTM layers and one output layer, achieving a prediction accuracy of 59.1% on the test dataset. Further efforts are underway to enhance the model's performance within the project scope. This research underscores the importance of objective activity tracking and highlights the potential to fill existing gaps in this field. The developed model, based on heart rate data collected via wearables, is expected to serve as an objective tool for monitoring cardiac rehabilitation processes and daily activity levels. By evaluating data collected through wearables, the model facilitates the calculation of individuals' durations of rest, light, moderate, and high-intensity activities throughout the day, contributing to personalized health care practices.