The Challenge.

  • Emergency Medical Records (EMR) hold thousands of records including patient histories, physical exams, examination results and diagnostics.
  • However, this data is difficult to access and is highly unstructured, making it tricky to analyze and use for decision making.
  • Patient diagnosis is a complex process involving a lot of variables; having a diagnostic support tool can be useful for training and validation purposes.

The Solution.

  • Designed and built a model to predict and output a list of potential diagnoses based on a patient’s history, physical exams and other patient information.
  • Used NLP/NLU to extract and analyze text data to create relevant features for the model.
  • Live integration of the diagnostic tool with the EMR so that it is always working with the most updated patient information.

The Results.

  • Improved and standardized the diagnostic process with a decision-support tool.
  • Streamlined training of new clinicians.
  • Collected and pre-processed a massive amount of data for future analysis.
style="vector-effect: non-scaling-stroke;">