Using AI to measure the quality and equity of private health services serving the poor in Mexico
Stefano M. Bertozzi, Professor
Public Health
Applications for Spring 2025 are closed for this project.
Mexico’s public healthcare system has faced underfunding, leading to saturation of primary services and reducing the quality and accessibility for the population. Moreover, there is still a large population that remains uninsured. As a response, people have sought private services to cover their demands. In the last 20 years, Pharmacy-adjacent-doctor-offices (PADOs) have expanded to cover such demand. Popularized by the largest pharmacy chain in the country, Farmacias Similares, PADOs are small offices with physicians sharing the same space and owned by pharmacies. The expansion of private healthcare services has helped to increase access, but there is scarce research on the overall population health effects resulting from this expansion. To start studying the effect of PADOs on the population’s health, we first need to find all PADOs in the country. There are no data sources with the location and information of all PADOs, especially small ones belonging to independent or small-size pharmacy chains. In this project, we will leverage Google Maps, Street View information, and AI to build this dataset. Nevertheless, some information needs to be retrieved manually from these sites to validate the algorithms, obtain start of operations dates from each PADO, and use econometric models to measure the causal effect.
We will need 3-5 students who can read Spanish at a basic level and who are interested in automated identification and classification of pharmacies and PADOs – as well as in analysis of the data. Students should have some quantitative skills (e.g. at least Data 8) but could be majoring in a variety of areas, from public health to economics, data science, business, etc. The specific tasks assigned will depend on each student’s skill set and interests.
Role: This project uses web data from Google Maps, Street View, and web searches of pharmacies in Mexico.
Students working on this project will receive introductory training in data cleaning and good practices when creating new data sources using massive online information and AI. Additionally, basic econometric/statistical analyses will be introduced to understand the objective of the project and the use of the data.
Tasks:
● Label PADOs using Google Maps and Street View information
● Looking for unregistered PADOs in selected cities of the country
● Identify start of operation dates through web searches and phone calls
● Review survey and administrative data using AI to create new variables
● Review global health policy and health economics literature related to
the effect of expanding pharmacy and primary care services in LMIC
Qualifications: Sufficient Spanish to understand basic written Spanish is essential; fluency is desirable, as is knowledge of the Mexican context. Candidates with an interest in international health policy, Latin American policy, and health economics will greatly benefit from this project.
Day-to-day supervisor for this project: Jorge Morales Alfaro, Ph.D. candidate
Hours: to be negotiated
Off-Campus Research Site: The work will be predominantly online (Zoom) with some in-person training sessions to be held at the beginning of the semester.
Biological & Health Sciences Digital Humanities and Data Science