In this video, we analyze a cleaned dataset of COVID-19 patient data, known as the Mortality Risk Clinical Data, to build a model predicting mortality risk or severity. The dataset includes patient demographics, comorbidities, and clinical measurements collected during the initial phases of the pandemic.

We begin with data preprocessing, including encoding age categories, normalization, and splitting the dataset into training, validation, and testing sets. To improve model performance, we handle class imbalance and utilize a feedforward neural network.

Further, we explore feature selection techniques such as PCA and Genetic Algorithm, comparing their effectiveness in improving model accuracy. Our findings reveal that PCA achieved the highest accuracy (82.93%), while the Genetic Algorithm offered valuable insights into domain-specific features.

Tune in to learn how these methods can optimize healthcare resource allocation and enhance early predictions for better patient management.

Documentation: [ Ссылка ]

свернуть