DVL-Sejong/COVID_Project

COVID Project

This project aims to characterize how diseases are spreading through data. We have worked on a project to collect, preprocess, and model data related to the spread of COVID-19 from multiple organizations.

Disease-related Repositories

DVL-Sejong's disease-related repositories are organized as follows:

Repositories	Summary	Etc
COVID_DataProcessor	COVID-19-related data collection and preprocessing (USA, Italy, India, China)
AutoCOVID19	Predict the number of people infected with COVID-19 using a ConvLSTM-based model, and optimize the model using the HPO libraries	Private
COVIDConvLSTM	Predicting the number of people infected with COVID-19 using a ConvLSTM-based model	Private
DeepNIPA	Expanded SEIR model using LSTM module and Neural ODE module	Private
SIRD	SIRD (Susceptible-Infected-Recovered-Deceased)	Classical compartmental model in epidemiology)
R0_Estimation	R0	Estimating basic reproduction number in epidemiology
NIPA	NIPA (Network-Inference-Based Prediction Algorithm), network inferenced COVID-19 prediction model in China
COVID_Evaluation	Comparing model performance among COVID-19 prediction models

Papers

The papers presented by DVL-Sejong are as follows:

Papers	Summary	Related Repositories
[K-Conference, 2021] COVID-19 Infected Case Prediction Model using Neural ODE	Created LSTM and NeuralODE‑based deep learning model which is predicting the number of infected people with COVID‑19 Using the NeuralODE module, designed the model for solving ordinary differential equations which is used to estimate the spread of disease	DeepNIPA(private)
[K-Journal, 2021] ConvLSTM-Based COVID-19 Outbreak Prediction using Feature Combination of Multivariate Dataset	Created ConvLSTM‑based deep learning model for analyzing which feature combinations are affecting the spread of COVID‑19 Using patient and location information, generated multivariate 3d array dataset, and adopted KDE kernel for maximizing the spatial feature Constructed 120 feature combinations by contextual & statistical methods, and optimized the model for each learning dataset using HPO libraries	AutoCOVID19(private) COVIDConvLSTM(private)