Project Abstract

Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection. Globally, it is one of the major causes of in-hospital mortality and morbidity, especially in intensive care units (ICU), leading to prolonged hospitalizations and escalated healthcare costs. The urgency for early identification of sepsis is rooted in its time-critical nature, where any delay in detection can hinder timely medical intervention and adversely affect the health of a patient. Currently, hospitals utilize criteria-based scoring methods of sepsis identification such as Systemic Inflammatory Response Syndrome (SIRS) and Sequential Organ Failure Assessment (SOFA). However, these methods have low specificity and sensitivity. There is also no universally recognized clinical biomarker for sepsis diagnosis. As an alternative, Machine Learning algorithms have demonstrated promising outcomes in prediction of sepsis. However, clinically scalable machine learning models have not been developed, yet. In this project, comparative machine learning models along with an ensemble method of Random Forest, Gradient Boosting Decision Trees and Neural Network algorithms will be developed with tailored feature engineering to predict sepsis 6 hours before its onset.

Keywords: Sepsis Prediction, Machine Learning, Artificial Intelligence (AI)

 Conference Details

Session: Presentation Stream 10 at Presentation Slot 2

Location: CoFo 002 at Tuesday 7th 13:30 – 17:00

Markers: Randell Gaya, George Brooks (GTA)

Course: MSc Computer Science, Masters PG

Future Plans: I’m looking for work