Session ID: 

Real-Time Machine Learning Pipeline: A Clinical Early Warning Score (EWS) Use-Case

3:00pm - 4:00pm Tuesday, February 12
Orlando - Orange County Convention Center


Modern healthcare systems are generating clinical data at an ever-increasing rate and in a wide range of formats including structured, unstructured, image, and high-frequency waveform data. The volume of data is also growing at an exponential rate. Analyzing these silos of real-time clinical data and providing timely and personalized clinical interventions are challenging tasks. To address this problem, we developed and deployed a real-time machine learning pipeline, where any updates to the electronic health record (EHR) platform stream to a data science engine that generates real-time predictive notifications. To create the EHR agnostic pipeline, we used big-data open-source technology stacks, such as Mirth Connect, Apache Kafka, MongoDB, Apache Spark, Tensor Flow, and Keras. In this talk, we will share our experience of translating clinical challenges into the data science optimization problems, developing and deploying the real-time machine learning pipeline.

Learning Objectives: 

  • Explain the rationale behind the need for a real-time machine learning (ML) pipeline in personalized medicine
  • Explain the architecture of the healthcare ML pipeline
  • Identify the challenges and the lessons learned from developing and deploying process
  • Discuss how the ML pipeline can be generalized into various health care settings and cases of use


Senior Data Scientist,
Mount Sinai Health Systems
Senior Data Engineer,
Mount Sinai Health Systems


Chief Quality, Chief Clin Transformation Officer
Data Scientist