pepecura / ml-on-gcp-patient-readmission

Predict the likelihood of hospital readmission for diabetes patients using BQML and AutoML on GCP.

Healthcare application of Machine Learning on Google Cloud Platform

Predicting patient hospital readmission with BQML and AutoML

colab link for the notebook

What is the business problem:

Predict the likelihood of hospital readmission for diabetes patients.
Avoid early discharge of patients to home in order to decrease the rate of readmission.

Download data from:

The public data used in the exercise will be downloaded from the Kaggle website. This will require you to sign in to Kaggle. Find the descriptions of the variables here.

Machine learning steps:

The technology used:

• AI Platform on GCP
• BigQuery ML
• Looker
  

Follow this notebook to build your healthcare ML application on GCP using BigQuery and Python libraries for data preparation and feature engineering to enrich your data, BigQuery ML and AutoML for model building and evaluating the results and Looker for visualisation.
The details for the visualisations in Looker are not covered in this repository. Please check the Looker website and the documentation here.

Model results:

the end.