evidoria

← Back to browse

Good practice

MSP COVID-19 Chatbot — Uruguay's IBM Watson-Powered Public Health Assistant

Uruguay · Montevideo · See the Uruguay profile

From March 2020, Uruguay's health ministry ran an IBM Watson chatbot on WhatsApp, Messenger and 60+ health-provider sites, processing over a million COVID-19 queries in its first three weeks without collecting any personal data.

more than one million
Messages processed in first three weeks (March 2020)
more than 60
Web portals embedding the chatbot
16-plus
Topics understood after December 2020 upgrade
160
Further topics identified by IBM for MSP validation
MSP COVID-19 Chatbot — Uruguay's IBM Watson-Powered Public Health Assistant

Details

Maturity
Scaling
Promoter
Ministerio de Salud Pública (MSP), with Agesic and IBM (developed by Isbel)
Period
March 2020 – December 2020 (major update); WhatsApp/Messenger/60+ portals
Keywords
public health, natural-language chatbot, pandemic response

Context

In March 2020, Uruguay's Ministry of Public Health (MSP), working with e-government agency Agesic and IBM — which donated the underlying Watson natural-language technology, built by local developer Isbel — launched a COVID-19 information chatbot to absorb the surge in citizen questions about the new virus.

Objectives

To free the MSP's phone hotline (0800 1919) to focus on urgent cases by answering routine COVID-19 questions via chat channels.

Activities

The chatbot ran on WhatsApp (098 999 999), Facebook Messenger (@MSPUruguay), and was embedded on more than 60 web portals belonging to different health-care providers around the country. An upgraded version launched in December 2020 moved from a rigid, menu-based design of four response categories to a natural-language model understanding free-text questions across 16-plus topics, with IBM identifying 160 further topics ready for MSP validation. It stored no personal data, redirecting testing-logistics questions (its most requested but least-answered topic) to official sources.

Results

Within its first three weeks, it processed more than one million messages, according to reporting at the time.

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful