MIMIC-III EHR Mortality Rate Predictions
- 0 Collaborators
The goal of this project is to find new ways of predicting a patient’s mortality rate based on the MIMIC-III dataset of ICU patients. I plan on accomplishing this by using deep learning algorithms as well as Super Learner Models and by drawing conclusions from severity scores. ...learn more
Project status: Under Development
Overview / Usage
Doctors will be able to easily read electronic health records because of the reduced amount of redundant information. Patients will also be able to easily read their records because of the simpler alternate terminology meanings and definitions. A patient's data can be used to form various predictions on what future steps the patient should take in terms of health, diet, exercise, etc.
Methodology / Approach
This project aims to make medical data, specifically data from EHRs (electronic health records) more user-friendly. I plan to do this by using various machine learning methods including deduplication and record linkage to help simplify the data that is being entered into these records. I will also be using a mixture of machine learning and natural language processing algorithms to understand the data that is being entered better and to make future predictions based on the data that has already been entered.
Technologies Used
Intel Optimized TensorFlow, Intel AI DevCloud
Repository
https://github.com/PallabPaul/mimic-mortality-predictions