The project aimed to implement a digital processing algorithm of biomedical signals to estimate the chances of survival for patients admitted to the ICU. A large variety of types of signals and biomedical parameters, coming from various sensors, were analyzed. A classification model based on artificial intelligence was implemented to estimate the chances of survival. The data for this project was taken from the WiDS Datathon 2020 which focused on patient health through data from MIT’s GOSSIS (Global Open Source Severity of Illness Score) initiative. https://www.kaggle.com/competitions/widsdatathon2020/overview
This project was presented as the final bachelor's diploma thesis of the author. Copyright © 2023, Elena Briana BOERU. All rights reserved.
Personal contributions include, but are not limited to:
• Extensive research of the State of The Art;
• Using a highly diverse and comprehensive database and its successful handling, without corrupting or invasively modifying the data;
• The development of personal prediction models that use regularization techniques, more layers with a varied number of electrons and multiple activation techniques;
• Formulation and validation of a hypothesis regarding the necessary attributes required to obtain optimal performances, which enables a new precondition for future works developed on the same topic;
• Successfully presenting an intermediate stage of the project in front of an accredited evaluation commission, implementing the feedback given by this entity and obtaining the third place in the Student Scientific Communication Session.