Reproducible Paper - Diagnosis of patients with blood cell count for COVID-19: An explainable artificial intelligence approach

An Open Source code released to increase reproducibility in academic and professional research.

Paper info

• Title: Diagnosis of patients with blood cell count for COVID-19: An explainable artificial intelligence approach
• Access Link: here
• Journal: Journal of Health Informatics (JHI)
• ISSN: 2175-4411
• Journal Impact Factor (QUALIS): B5 for Computer Science and Engineering

Authors info

• Kaike Wesley Reis (corresponding author): LinkedIn and Lattes
• Karla Patricia Oliveira-Esquerre: LinkedIn and Lattes

Overall AI framework

Disclaimer: The supplementary material provided in this repository contains extra analysis and discussions compared to the paper discussion. This decision was made to make the results presented more focused and objective for the paper.

Best selected model info

• AI model: Random Forest

• Overall parameters (including hyperparameters):

  {'bootstrap': True,
  'ccp_alpha': 0.0,
  'class_weight': 'balanced_subsample',
  'criterion': 'gini',
  'max_depth': 21,
  'max_features': 'sqrt',
  'max_leaf_nodes': None,
  'max_samples': None,
  'min_impurity_decrease': 0.0,
  'min_impurity_split': None,
  'min_samples_leaf': 1,
  'min_samples_split': 2,
  'min_weight_fraction_leaf': 0.0,
  'n_estimators': 1000,
  'n_jobs': None,
  'oob_score': False,
  'random_state': 1206,
  'verbose': 0,
  'warm_start': True}
  

Repository info

• The .ipynb was used to developed the source material related to this paper.
• The numbers at the beginning of each notebook represent the pipeline order:
  • 0 and 1 for Pre-processing
  • 2 for Model development
  • 3 for Results evaluation (model selection)
  • 4 for Qualitative analysis
• The original dataset can be found here at Kaggle's platform.

Requirements

The requirements.txt was generated outside a virtual environment and then presents all packages installed on the machine without exception. Given this fact, I separated the main packages, followed by their versions, used for this paper:

Package	Version
numpy	1.18.1
pandas	1.0.5
missingno	0.4.2
matplotlib	3.1.3
seaborn	0.10.0
scikit-learn	0.23.1
skopt	0.8.dev0
xgboost	1.1.1
scipy	1.4.1
joblib	0.14.1
shap	0.38.0
umap	0.4.6
imbalanced-learn	0.7.0