Inspiration
The pandemic deeply affected our lives, highlighting the need for everyone to understand how regional health differences and travel contribute to the death rate of respiratory diseases. Our team set out to explore the trends of influenza, pneumonia, and COVID-19, looking to uncover how these illnesses unfolded across different regions of the U.S. and how the spread of COVID-19 triggered travel restrictions and disruptions.
What it does
Our project analyzes mortality trends related to pneumonia, influenza, and COVID-19 across U.S. states. Using K-Means Time Clustering, we identify regional similarities and differences in mortality trends over time. Additionally, we examined the correlation, on the hosted Streamlit application, between these trends and flight cancellations, demonstrating the relationship between travel restrictions and the spread of respiratory illnesses.
How we built it
We used K-Means Time Clustering–an extension of the traditional K-Means algorithm tailored for time series data. Unlike standard K-Means, which clusters based on spatial proximity, K-Means Time Clustering accounts for the temporal dimension, allowing us to group states based on the similarity of their death trends over time.
Challenges we ran into
One of our primary challenges was dealing with dependency conflicts while attempting to self-host our Streamlit app. It was extremely difficult to ensure compatibility between various libraries and dependencies; after several hours wasted, the solution we decided on was to remove the problematic packages. Additionally, we learned how to add a temporal component to the traditional K-Means clustering method which helped us to overcome the challenge of using clustering for time series data.
Accomplishments that we're proud of
We are proud of identifying United States patterns in health-related mortality between regions with relatively unlabeled data in this area. Our geospatial visualizations were also quite effective in communicating these patterns. Additionally, correlating flight cancellations with spikes in respiratory disease deaths allowed us to illustrate the impact of public health measures on mobility and health outcomes.
What we learned
Throughout this project, we learned the importance of pushing, pulling, and fetching with care. Our group had a couple scary instances of merge conflicts that were able to be resolved using git’s pull.rebase feature. Had we not learned to use this tool and done it correctly it could have results in hours of work being lost.
What's next for Analysis on Influenza, Pneumonia, and COVID Trends & Travel
Moving forward, we aim to expand our analysis to include additional health factors, such as vaccination rates and demographic information, to provide a more comprehensive understanding of regional health disparities. We also want to explore predictive modeling techniques to anticipate future trends and assist policymakers in preparing for similar health crises.
Log in or sign up for Devpost to join the conversation.