SEWS: A Web-Based Sepsis Early Warning System Using Random Forest for Real-Time Risk Prediction

Abstract

Sepsis is a life-threatening condition requiring early detection to reduce mortality. This paper presents a web-based Sepsis Early Warning System (SEWS) using a Random Forest model to predict risk from clinical parameters such as age, vital signs, and laboratory values. A React and TypeScript interface enables healthcare professionals to input data and receive real-time predictions. The Flask backend performs data processing and model inference, while Firebase ensures secure authentication and storage of patient history. The system displays color-coded risk levels, supporting quick clinical decisions and improving usability, accuracy, and scalability in real-world healthcare environments.

Introduction

The text describes the development of a Sepsis Early Warning System (SEWS) that uses machine learning to predict sepsis risk in real time. Sepsis is a life-threatening condition caused by the body’s extreme response to infection, and early detection is crucial for reducing mortality. Traditional diagnostic methods often fail to detect complex patterns in patient data, leading to delayed diagnosis and treatment. To address this, the proposed system uses a Random Forest classifier to analyze patient demographics, vital signs, and laboratory values for early prediction.

The system is implemented as a web-based application, where a React-based frontend collects patient inputs and a Flask backend processes the data and generates predictions. Firebase is used for authentication and secure storage of patient records and prediction history, enabling continuous monitoring and analysis.

The architecture is divided into multiple layers, including data input, communication via REST APIs, machine learning processing, and application output. The workflow involves data collection from ICU datasets, preprocessing (cleaning, normalization, feature selection), model training, and real-time inference. Performance is evaluated using standard metrics such as precision, recall, F1-score, and AUC.

Conclusion

The Sepsis Early Warning System effectively applies machine learning and web technologies for early sepsis detection. By integrating a Random Forest classifier with a React frontend and Flask backend, the system analyzes patient clinical data and delivers real-time risk predictions. Firebase ensures secure data handling and historical tracking. The color-coded visualization (Low, Medium, High) enhances usability and allows healthcare professionals to interpret results quickly. The system is reliable, efficient, and scalable, making it suitable for practical healthcare and clinical decision support applications.

References

[1] A. Boussina, H. Khurana, A. D. Bedoya et al., \"Impact of a deep learning sepsis prediction model on clinical outcomes: a multi-site before-after implementation study,\" npj Digital Medicine, Vol. 7, 52, 2024. [2] M. Moor, B. Rieck, M. Horn et al., \"Predicting sepsis using deep learning across international, multi-centre data,\" Nature Medicine, Vol. 29, pp. 1991-2002, 2023. [3] A. Rafiei, A. Ebrahimi et al., \"SSP: Early prediction of sepsis using a fully connected LSTM-CNN model,\" Computers in Biology and Medicine, Vol. 137, 104795, 2021. [4] S. P. Shashikumar, C. S. Josef, A. Sharma, and S. Nemati, \"DeepAISE - An interpretable and recurrent neural survival model for early prediction of sepsis,\" Artificial Intelligence in Medicine, Vol. 113, 102036, 2021. [5] Z. Zhang, C. Yan, and H. Qiu, \"An interpretable deep-learning model for early prediction of sepsis,\" Patterns (Cell Press), Vol. 1, Issue 8, 100142, 2020. [6] S. Bomrah et al., \"A scoping review of machine learning for sepsis prediction,\" Critical Care, Vol. 28, Issue 1, pp. 68, 2024. [7] Y. Tang et al., \"A time-series driven model for early sepsis prediction based on CNN-Transformer and LSTM-Transformer,\" BMC Medical Research Methodology, Vol. 24, Issue 1, pp. 15, 2024.

Copyright

Copyright © 2026 Mrs. Yashaswini R, Meghana M D, Mohith M, Nisarga P, Sowjanya S R. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.