An Scalable and Interoperable Real-time Software Platform for Forecasting the Onset-time of Sepsis Open Access

Amrollahi, Fatemeh (Summer 2019)

Permanent URL:


Sepsis, a dysregulated inflammatory response to infection, is difficult to diagnose in advance of life-threatening physiological decompensations. Multiple studies have demonstrated improved outcomes when this condition is recognized and treated early. Nemati et al. have developed a real-time, high-dimensional machine learning algorithm capable of detecting sepsis four to six hours prior to clinical recognition, capable of substantially reducing the untoward effects associated with the condition. In this work, a software platform was developed that consumes live patient data, securely transports it into a cloud environment, and interprets it in real-time. Our approach leverages the benefits of cloud-based managed services that are scalable and fault tolerant. Though there are several pathways for extracting live data from electronic health records (EHR), the AIDEx platform proposed in this work is an EHR vendor-agnostic open-source solution that can be easily deployed in any clinical environments. 

Table of Contents


Sepsis: A Health Crisis

Sepsis Prediction Algorithm

Real-time Software Platform for Forecasting the Onset of Sepsis

AIDEx: A FHIR Based Real-time Software Platform for Forecasting

the Onset of Sepsis


The AIDEx 1.0 Services

Sepsis App on Tele-ICU: A SQL Based Real-time Software Platform

at Emory Hospital for Forecasting the Onset of Sepsis

Overview of SepsisApp on Tele-ICU

Tests and Statistical methods

Conclusion and Future directions


About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files