/iAtros_challenge

hypertension_fhir_challenge_iatros

Primary LanguageJupyter NotebookMIT LicenseMIT

Hypertension Alert System

The repository contains the solution for the iAtros AI & Backend Engineer challenge.

The solution contains two parts.

  1. First a prototyping notebook for function testing and explaining reasoning process.
  2. A cli tool that queries for a patient and predicts it mesurments returning an alert summary.

Prototyping

The notebook (iatros_challenge.ipynb) contains the step by step implementation of the solution incluiding:

  1. Implementation of requests between HAPI-fhir server and app

  2. Pagination handeling

  3. Filling the Dataframe

  4. Running the main query

  5. Preprocess and clead the Dataframe

  6. Binary classification

    • Model setup
    • Model definition

  7. Running the model

  8. Prediction test on a row of dataframe

Installation

1. Install Conda environment

  1. conda env create -f environment.yml
  2. Activate the environment: conda activate iatros
  3. Verify the environment was installed correctly: conda env list

2. Install fhir-client

In main working directory from terminal run pip install fhirclient --no-index --find-links client-py-4.0.0.zip

3. Install PIP Requirements

  1. pip install -r requirements.txt

How to Run the prediction App

From terminal:

Example: Arterial Pressure Prediction for a subject on database

  • Args:
    • observ: Code for observation defaul to 85354-9 LOINC code
      'Blood pressure panel with all children optional'
    • subject: Subject code in HAPI-FHIR test server
    • path to pretained model, default to models/hypertension_model

python app.py --observ 85354-9 --subject 1598464 --model ./models/hypertension_model.pkl