Research & Projects

Project 1

Multimodal Influenza Forecasting

The Multimodal Influenza Forecasting project integrates genomic, environmental, and healthcare data to enhance the accuracy of influenza outbreak predictions. Traditional models often rely on singular data sources, limiting their scope. By leveraging machine learning to combine diverse data streams, this project addresses the complex factors driving flu dynamics, enabling more effective public health interventions. This research is being conducted at the University of Michigan under the advisement of Professor Alexander Rodriguez and in collaboration with Professor Liyeu, this multidisciplinary effort brings together expertise in machine learning, genomics, and epidemiology.

Objectives

The primary objectives of the Multimodal Influenza Forecasting project is to integrating diverse data modalities such as genomic data, environmental factors, and healthcare trends to create a comprehensive forecasting model.

Results

The project is currently in the preliminary stages, with efforts focused on problem formulation and initial data integration. While final results are pending, early work has highlighted the challenges of aligning multimodal datasets and the importance of balancing data diversity with model interpretability. This phase is helping to refine the methodological framework and identify key variables influencing influenza forecasting.

Project 2

Learning Selection Bias

This project focuses on learning and mitigating selection bias in agent-based models (ABMs) for epidemiology. Selection bias can significantly skew predictions and resource allocation, especially when models disproportionately emphasize severely infected individuals requiring hospitalization. Motivated by the need for equitable and accurate public health interventions, this project aims to quantify and correct these biases to improve decision-making processes. The research was conducted in collaboration with Professor Sindhu Kutty at the University of Michigan under the advisement of Alexander Rodriguez. The project combines expertise in machine learning, epidemiology, and public health modeling to address this critical challenge

Objectives

The primary objective of this project is to identify and address selection bias in infectious disease datasets, particularly focusing on the disproportion of severely infected individuals that are hospitalized.

Results

The formulation of this project is currently ongoing. Early efforts are focused on exploring potential correction strategies. Future results are anticipated to provide actionable insights into mitigating selection bias, with broader implications for enhancing the equity and reliability of epidemiological modeling.

Page updated

Report abuse