Data Visualizing Birth and Maternity Trends in New York
The skills we demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
The New York State Maternity Law requires hospitals to provide data and information about their childbirth practices and procedures. This information aims to help individuals understand what to expect, to learn more about their childbirth choices, and to plan for their baby’s birth.
The 'Hospital Maternity Information: Beginning 2008' dataset is available on the New York State Department of Health website, as part of the state's open data initiative. This dataset contains information reported by hospitals that have been compliant with New York State’s Maternity Information Law.
The state website allows individuals like me to view some visualizations of some of the metrics in the dataset by hospital and by year. This style of visualization made it hard to metrics as they changed over the years and thus, difficult to get insights on trends. Metrics cannot be compared side by side. A geospatial representation of the data would also be useful in looking at how trends vary across the state.
To address these issues, I created a shiny app that would allow users to view trends on a geospatial representation of New York State and look at trends as they changed over time. In addition, I wanted to be able to compare county data as opposed to the current available visualization on the New York State Department of Health website, where Hospitals can be compared to one another.
With a goal of creating a geospatial visualization of New York State Maternity Data, I used Leaflet and Google Vis to plot county shapes on New York State maps. Bar charts and Correlation plots allow the user to compare different metrics and measurements by county-vs-state averages and by county-vs-county averages. Hospital specific data can be visualized on a map as well.
Interesting Data findings
While not surprising, most births occurred in counties with or in proximity to big metropolitan areas like New York, Kings (Brooklyn), Queens, Nassau, and Westchester. There is a high density of hospitals in these areas as well. The high compliance of these hospitals account for most of the births recorded in the data set. Surprisingly, the highest incidence of cesarean sections are in Nassau and Suffolk counties. Areas considered to be Upstate and West New York tend to have numbers below the state average.
For a more robust and insightful app, I would like to include New York State infant outcomes data in addition to county and state demographics. I would like to investigate why certain counties have no reported data despite the State having laws and requirements for hospitals to provide data.
I would update the way metrics are measured as a proportion rather than actual birth counts. This would normalize data and comparisons between counties with big population differences would be possible.
Correlations and relationships between infant outcomes and delivery conditions can be made. It would also be interesting to find if there are any demographic trends for birth and delivery conditions.
Click here for app.
Click here for code.