README.md

EARLY PREDICTION OF MORTALITY RISK AMONG COVID-19 PATIENTS

Btech Thesis project

PUBLISHED A RESEARCH PAPER: https://ieeexplore.ieee.org/abstract/document/9751945

• In this study, our primary aim is to Detect the Severe Covid-19 patient in the Early Stages by looking at the information on demographics, comorbidities, admission laboratory values, admission medications, admission supplemental oxygen orders, discharge and mortality.

• Motivation behind the Project is to help the overwhelmed hospitals by predicting the Severe Covid-19 Patients in Early Stages on the basis of their comorbidities, admission laboratory test values, admission medications, admission supplemental oxygen orders. While knowing the number of patients which may require a Intesive-Care-Unit(ICU) in the future, Hospitals can arrange the ICU beds accordingly, which can lead to Save the Patients life or knowing which patients don't need any ICU support or Not Severely affected by COVID-19 can go for home quarantine.

• 4711 patients dataset with conirmed SARS-CoV-2 infection were included in the study which contain 85 features.

Since, the dataset contain very large number of features(85 features) so we have used 7 features selection algorithm to Selct the Most Important features in the dataset.
We filtered out the feature which is most common in all 7 algorithms.

The Seven Features selection(FS) algo used are:-

1. FS with Pearson Correlation
2. FS by the SelectFromModel with LinearSVC
3. FS by the SelectFromModel with Lasso
4. FS by the SelectKBest with Chi-2
5. FS by the Recursive Feature Elimination (RFE) with Logistic Regression
6. FS by the Recursive Feature Elimination (RFE) with Random Forest
7. FS by the VarianceThreshold

After Selecting the best Features out of all features I have implemented 17 different models and check their accuracy, precision, recall, F1-Score and AUC_ROC and finally a voting classifier is made which includes the top best models out of all 17 models and vote is made by each model to predict the class 0/1 . The class which get most votes is our predicted class.

Models which are used in this project are

1. Linear Regression
2. Logistic Regression
3. Support Vector Machines
4. Linear SVC
5. MLP Classifier
6. Decision Tree Classifier
7. Random Forest Classifier
8. Ada Boost Classifier
9. Gradient Boost Classifier
10. XGBoost Classifier
11. LightGBM
12. Ridge Classifier
13. Bagging Classifier
14. ExtraTree Classifier
15. KNN
16. Naive Bayes
17. NN with Keras
18. Voting classifier

RESULT