This project analyzed death certificates, some grand jury reports, and some news articles documenting deaths of inmates at the House of Reformation and Instruction for Colored Children at Cheltenham, the House of Correction at Jessup, and the House of Refuge for Juvenile Delinquents, now the Charles H. Hickey Jr. School in Baltimore County. There were 177 recorded deaths that occurred between 1877 and 1941.
Scanned images of the death certificates, which contained handwritten text, were processed with a Python script and the Google Gemini 2.5-flash large language model, producing structured data for each entry. The data was then analyzed and verified by Capital News Service journalists working in Jour 772, Data Journalism, under the supervision of Associate Professor Rob Wells, Ph.D.
The CNS team ran multiple queries and verification checks using the R programming language and the tidyverse libraries. This process included checking the AI output against each source record. The CNS team standardized dates and created second-level categories to understand broader patterns in causes of death over this time period.
The CNS team encountered numerous gaps in the death certificates, including birthplace, hometown, and age. These gaps were marked as missing details in the data analysis. The CNS team used the resulting data to produce a variety of graphics, interactive tables, and a map.
The CNS team: Akira Kyles, Zaka Hossain, Nicole Ramos, Brandyn Fragosa, Sophia Hernandez-Pina, Bex Heimbrock, David Landerman, Raphael Romero Ruiz
You must be logged in to post a comment.