Expert System TNM Classification of Malignant Tumors Lung Cancer

Course:

BME10. Artificial Intelligence and medical diagnosis decision support systems.

Supervisors:

Anastasios Delopoulos
Ioanna Chouvarda
Panagiotis Givissis

Contributors:

Grigorios Papapostolou
Ioannis Dervisis
Christos Karapapas

How to run an experiment

Experta needs Python 3.8 to work properly so a Python environment with 3.8 would be ideal.

Step1.

One easy way to create one is by Installing Anaconda and create an environment following these steps.

conda create -n dss-env-01 python=3.8  
conda activate dss-env-01  

Step2.

Once the environment is setup and activated, run the "synthesizer_nb" Jupyter notebook to generate some synthetic data.

Step3.

Run the "experiment" Jupyter notebook.

Feature Requests

1. Create rules that would also return exam orders apart from the Suspicion, T, N, M and Overal Stage estimates.
2. Decouple the value addition to the properties of synthetic data from the estimate function and have these values determined during the synthetic data generation.

Backlog

1. Find statistics about how many cases of each cancer stage and non-cancer typically exist in a cohort and adapt these percentages in the funtion that generates the synthetic data, to have more realistic data.
2. Create a web UI, using Streamlit or any other similar library.
3. Binarization of total financial and delay cost so that the results would be more useful in the exploratory data analysis.