Real time sepsis prediction using tcn and tiny ml

Sepsis is a life-threatening medical condition caused by the body’s extreme response to infection, leading to tissue damage, organ failure, and death if not identified early. This project proposes an intelligent sepsis prediction system using deep learning and Tiny Machine Learning (TinyML) techniques to provide efficient and rapid detection in healthcare environments. The system integrates Temporal Convolutional Networks (TCN) for sequential data modeling and compares its performance with classical machine learning models such as Random Forest and XGBoost. The proposed model analyzes patient physiological parameters and vital signs to classify the condition as “Sepsis Detected” or “Sepsis Not Detected.” The TCN architecture effectively captures temporal dependencies in medical time-series data, enhancing early diagnosis accuracy. Meanwhile, TinyML deployment ensures low-power, resource-efficient prediction suitable for edge healthcare devices. The project employs Python with Flask for backend processing and HTML, CSS, and JavaScript for the user interface, offering an interactive platform for real-time prediction and visualization. By combining deep learning with TinyML, the system aims to support clinicians with fast, reliable, and accessible sepsis detection, improving patient survival rates and advancing intelligent healthcare monitoring systems.

Keywords: Sepsis Prediction, Temporal Convolutional Network (TCN), Random Forest, XGBoost, TinyML, Flask, Healthcare Analytics, Deep Learning, Early Diagnosis, Time-Series Data.