This project aims to model an end-to-end workflow of implementing Artificial Intelligence (AI) for the clinical environment. A possible use-case such as the selection of patients for a novel treatment or drug will be conducted by estimating the hospitalization time with a Neural Network. The diabetes readmission dataset from the University of California, Irvine (UCI) Diabetes was used for this project. The trial population is selected by predicting the expected days for a person being hospitalized. Then and arbitrary boundary is set for chosing whether or not this patient is shall be included or not. If so, a clear explanation of the how the prediction was calculated and additional possible risk factors will be given in order to make the workflow explainable. This project shows that given a proper explanatory approach, AI can be a useful tool for the modern clinical environment. The workflow finally reveals that AI can be a beneficial support tool for doctors, e.g. by effectively choose possibly suitable patients in the patient selection process.
+-- Code
| +-- Notebooks
| | +-- Clinical_EDA.ipynb
| | +-- Machine_Learning.ipynb
| | +-- Explainable_AI.ipynb
| +-- Scripts
| | +-- model_preprocessing_utils.py
| | +-- tensorflow_modeling.py
| | +-- utils.py
| +-- Source
| | +-- __init__.py
| | +-- main.py
| +-- Tests
| +-- test_main.py
+-- Paper
| +-- Final Paper
| +-- Related Work Paper
| +-- Bibliography.bib
|
+-- Presentation
| +-- Mid-Term Presentation
| +-- Final Presentation
|
+-- imgs
+-- requirements.txt
+-- README.md
+-- .gitignore
- Clinical_EDA as iPython
- Machine_Learning as iPython
- Explainable_AI as iPython
- Final Presentation as PDF
- Final Paper as PDF
- Coding Example: https://towardsdatascience.com/machine-learning-for-diabetes-562dd7df4d42
- Mapping Data: https://www.accessdata.fda.gov/scripts/cder/ndc/index.cfm
- Model building: https://www.tensorflow.org/tutorials
- Model building with Layers: https://blog.tensorflow.org/2019/03/regression-with-probabilistic-layers-in.html
- Evaluation: https://www.sciencedirect.com/science/article/pii/S1877050916323870
- Dataset UCI Machine Learning Repository: https://archive.ics.uci.edu/ml/datasets/diabetes+130-us+hospitals+for+years+1999-2008
The dependencies to this project are stored in the file:
- requirements.txt
I use python version 3.7.4
- Tim Löhr - If you have questions you can contact me under timloehr@icloud.com
This project was done during my Seminar Machine Learning in the Industry 4.0 from the Machine Learning and Data Analytics Lab at the Friedrich Alexander University in Erlangen-Nürnberg. Some parts of the code are under the licence of www.udacity.com. Those parts can mostly be found in the scripts section.
- Thanks a lot to Philipp Schlieper from the Machine Learning and Data Analytics Lab for a really good supervising through all my project. I can totally recommend this seminar!