• M├ędicos por la Salud

What it does

We built a Telegram bot as a more reporter-friendly interface to collect data for the National Survey of Hospitals' daily reports.

Currently, daily status reports on electricity, running water and fatalities are harvested and processed manually, and there's no easy way to spot trends or prepare for contingencies.

With our new architecture, trusted reporters can be reminded automatically to fill in their daily status updates. All data is instantly recorded and ready to analyze.

How we built it

In order to be able to handle spikes in traffic that we've seen in similar bots in the wild in Venezuela, we built the chatbot in a serverless fashion by using AWS Lambda along with a managed SQL storage. We're storing all of the raw chat data, which means we're able to query and analyze it, including using NLP in the future to create better-performing versions of the bot.

Challenges we ran into

  • AWS network configuration took us more than we expected

What's next for Hospital Daily Reports 2.0

  • We could automatically produce the PDF reports available on the site, freeing up volunteers to perform data analysis or other tasks.
  • The architecture deployed here could be used to collect and manage all sorts of country-wide daily data outside of the hospitals: gasoline availability, repression status, etc. We could start to keep track of public respondents' answers and contrast them with trustworthy answers to assign confidence scores to each person and recruit potential new reporters.
Share this project: