A Robust Deep Learning Model for Predicting Gestational Age Based on Neonates Electroencephalogram (EEG)
Abdolzadehgan, Donya (2024-06-18)
A Robust Deep Learning Model for Predicting Gestational Age Based on Neonates Electroencephalogram (EEG)
(18.06.2024)
Julkaisu on tekijänoikeussäännösten alainen. Teosta voi lukea ja tulostaa henkilökohtaista käyttöä varten. Käyttö kaupallisiin tarkoituksiin on kielletty.
avoin
Julkaisun pysyvä osoite on:
https://urn.fi/URN:NBN:fi-fe2024072561834
https://urn.fi/URN:NBN:fi-fe2024072561834
Tiivistelmä
The early stages of life play a pivotal role in shaping neurodevelopment, and understanding the factors influencing neonatal brain maturation is crucial for assessing long-term outcomes. Traditional methods, such as Ultrasound Scans (USS) and the Ballard Score, present limitations in accessibility and objectivity, especially in resource-constrained settings. This research proposes a novel approach to predicting gestational age in neonates by leveraging the time-series data of 16-channel EEG recordings and a hybrid CNN-LTM architecture. Unlike existing studies that have primarily focused on specific aspects of prenatal maternal conditions, this research shifts the focus to the direct assessment of neurodevelopment using EEG data. By predicting gestational age, the study endeavours to contribute valuable insights into the dynamic changes occurring in the neonatal brain.
Inspired by the success of deep learning in various medical imaging and time-series analysis tasks, this research seeks to harness the power of neural networks for predicting brain age. The study aims to evaluate the suggested model's efficiency by evaluating its performance against traditional CNN and LSTM models, as well as assessing its predictive capabilities in the context of existing literature on gestational age prediction. The model was trained and evaluated using a dataset of EEG recordings from neonates with chronological age 0-5 days, with performance metrics including MAE, RMSE, and R^2. Results demonstrate the model's ability to accurately predict gestational age, with strong correlations between predicted and actual values (MAE=3.16 days, RMSE=4.38 days, and R^2= 0.75). Advantages of the proposed approach include robust performance and potential utility in clinical settings, while limitations such as interpretability and generalizability are also acknowledged. Future research directions include exploring additional data modalities and addressing model limitations to further advance the field of gestational age prediction. Overall, this study contributes to the development of accurate and reliable predictive models for neonatal care.
Inspired by the success of deep learning in various medical imaging and time-series analysis tasks, this research seeks to harness the power of neural networks for predicting brain age. The study aims to evaluate the suggested model's efficiency by evaluating its performance against traditional CNN and LSTM models, as well as assessing its predictive capabilities in the context of existing literature on gestational age prediction. The model was trained and evaluated using a dataset of EEG recordings from neonates with chronological age 0-5 days, with performance metrics including MAE, RMSE, and R^2. Results demonstrate the model's ability to accurately predict gestational age, with strong correlations between predicted and actual values (MAE=3.16 days, RMSE=4.38 days, and R^2= 0.75). Advantages of the proposed approach include robust performance and potential utility in clinical settings, while limitations such as interpretability and generalizability are also acknowledged. Future research directions include exploring additional data modalities and addressing model limitations to further advance the field of gestational age prediction. Overall, this study contributes to the development of accurate and reliable predictive models for neonatal care.