Tracking Outbreaks With Social Media Chatbots


By Shwetha Srinivasan, M&E Associate, Hoa Nguyen, Research Coordinator, PSI Vietnam and Bram Piot, Sr. Surveillance & Monitoring Advisor

The COVID-19 pandemic exposed the need for innovative surveillance solutions to monitor and respond to disease outbreaks. An effective disease surveillance system is an essential tool in detecting and tracking early warning signs that can signal increased disease within a population. This system can then facilitate rapid response to mitigate the impact of that illness—and other, simultaneous public health threats. 

To track COVID-19 outbreaks in Vietnam, PSI through investments from Unilever and the UK Government-funded Hygiene & Behaviour Change Coalition leveraged its existing digital solutions for malaria surveillance, by developing a social media chatbot to strengthen COVID-19 surveillance through the private sector.  

Event-Based Surveillance for COVID-19 (and Beyond) 

“Event-based surveillance (EBS) is defined as the organized collection, monitoring, assessment and interpretation of mainly unstructured ad hoc information regarding health events or risks, which may represent an acute risk to health,” — Africa CDC. 

Event-based surveillance (EBS) for early warning and response to disease outbreaks was introduced in Vietnam by the Ministry of Health and partners in the early 2010s, and was gradually expanded to all 64 provinces in 2019. EBS is considered a potentially effective tool in early detection of unusual events that may signal an outbreak. Traditionally, data for EBS comes from a variety of sources, including official point-of-entry quarantine units, media monitoring, and the formal individual health declaration. Private sector contributions to EBS however, have largely been limited due to the lack of efficient reporting mechanisms combined with health providers’ busy workloads; a lack of understanding of the benefits of disease surveillance; and clients’ unwillingness to share information.  

We know that engaging the private sector in disease surveillance is especially important in the event of public health emergencies such as COVID-19, given that private facilities, particularly pharmacies, are often the first points of contact for people seeking care.  

In order to strengthen disease surveillance systems, PSI Vietnam developed a social media chatbot on Zalo, a popular social media messaging platform in the country, connected it to a dedicated M&E online, then engaged the private sector.  

Surveillance Chatbots 

PSI has experience developing social media chatbots in the Greater Mekong Subregion for malaria case reporting, so it was easy to rapidly adapt and scale the existing surveillance “bots” to report fever cases by private outlets in the context of COVID-19.  

The chatbot aims to:  

  1. Encourage private health outlets to report possible COVID-19 cases 
  1. Build a network for reporting 
  1. Collect and analyze data for tracking trends 
  1. Support local health authorities with COVID-19 and general disease surveillance 

How does it work? 

Using automated dialogue flows, the chatbot prompts providers to answer a series of questions that cover a series of symptoms in clients seeking care for fever treatment. Providers also report necessary personal information such as gender, age and location of client in the last 14 days. With the chatbot connected to a DHIS2 instance, it’s possible for case reports to be automatically uploaded into the system and presented in dashboards for easier analysis.  

Following a brief development and piloting phase, the chatbot was introduced in August 2020, and as of April 2021, nearly 3,000 private providers (340 clinics, 2,586 pharmacies) were enrolled and provided with basic training to report fever cases. Current coverage focuses on areas with high population density and/or industrial zones, in five key provinces: Hanoi, Ha Nam, Quang Binh, Bac Ninh, and Thai Nguyen. During the same period, 81,908 client visits were reported by 1,768, or 61% of outlets. Heatmaps show 10 districts with highest number of fever cases. Most commonly reported symptoms were cough (39%), high fever (28%), and sore throat (28%), while under 5% showed a combination of symptoms suggesting respiratory infections such as COVID-19 (fever, cough, breathing difficulty). Currently, the surveillance data are shared with provincial Centers for Disease Control and Prevention and local health authorities in Thai Nguyen and Bac Ninh for assessment and verification, potentially contributing to their response actions. They have access to a DHIS2 dashboard demonstrating outlet distribution, daily number of reported events by location, gender, age and symptoms. 

The deployment of the chatbot addressed some of the main challenges with private sector disease surveillance by building capacity and creating an enabling environment for engaging the private sector. Some key insights have emerged from an accessibility and readiness survey conducted with providers: more than half (58%) prefer using this chatbot over the traditional health declaration thanks to its ease of use, speed, and flexibility; 97% report via Zalo within 48 hours (65% report within 24 hours). Ease of registration, quick reporting processes and the ability to support emergency response to control the pandemic were some of the key conveniences and motivations outlined by providers overall. 

What else did we learn?  

Lessons learned during the chatbot’s 10-month implementation period include: 

  • Local health providers need support. Despite the Zalo chatbot being more convenient than any existing paper-based systems, health providers are not experienced in using digital reporting tools and as a result they require active follow-up through routine monitoring and supervision visits. Incentives (certificate, reminder calls etc.) may be used to motivate providers to report cases.  
  • Some clients have expressed concerns about sharing their personal information. Obtaining informed consent and ensuring providers are clarifying why and how client data is used are contributors to alleviating these concerns. 
  • Participation and commitment from local partners are necessary to sustain and maintain EBS in the private sector. Within EBS, PSI intervenes at the stage where surveillance data is collected and analyzed. However, the CDC at the provincial level, Department of Health and district healthcare centers are responsible for information verification, investigation, response and control measures. 

What’s Next? 

  • DHIS2 account management was handed over to the local CDC, who were trained on this critical health information system. We expect local health authorities to be increasingly involved in supporting private sector to report surveillance data.  
  • The reporting system has started building capacity and is driving a reporting habit among providers.  
  • The system can be easily modified to monitor additional diseases based on needs.  
  • Collaboration between local health authorities and the social media provider is necessary to maintain the system, expand into other areas and other diseases.  
  • We’re keen to further scale up our chatbot solution in the private sector, integrating private sector engagement into the existing EBS system. 

With the use of chatbots, PSI expects to streamline reporting for private sector and strengthen overall surveillance systems with timely and integrated reporting on outbreaks. Building a foundation for private sector engagement through capacity building and sustained partnerships with local governments will enable early detection and efficient response to public health emergencies in future. 

To discuss the implementation of chatbots in this capacity, please contact our digital health team at [email protected]

Banner photo courtesy of Flickr/Lawrence Sinclair.

Sign up to
Receive Updates

Donate to
Support Our Work