Conversational interfaces like chatbots and voice apps are in the frontline of the global covid response. Dashboards, maps and APIs have been complemented with conversational projects, and health organizations use the, to bring public health information to citizens, organize vaccination campaigns and support other initiatives in their response, alleviating the workload of emergency lines. Alexa and Google Assistant are banning voice apps related to coronavirus, but a few institutions are promoting their own medical backed, official solutions. Different project have covered different activities, such as:
- Giving information: definitions, prevention, current figures…
- Fake news or hoaxes identification and mitigation.
- Self-assessment, screening and triage for people with symptoms.
- Notifications of suspects for dispatch of ambulances or domestic medical care.
- Pharmacovigilance for confirmed cases treated at home with traditional treatments or in the context of medical trials.
- Quarantine checks, including symptoms assessment, geolocation or contact reports.
- Transmission and explanation of lab results.
- Family finder.
The World Health Organization has a WhatsApp chatbot that can provide advice and information about COVID19. It had conversations with over 100 million people during the first 3 days. It is available in English, Arabic, French, Spanish and over 20 other languages. There are different phone numbers for each language.
The UCSF Health and Northwell Health are using a chatbot created by Conversahealth that can help with screening and triage, quarantine checks or lab results followup.
The CDC chatbot is made based on the Microsoft Healthcare bot framework and can help with self-assessments, tele-health and consult scheduling. I tested this chatbot and noticed they ask a lot of personal information, including age, sex and location, before they ask any covid-related questions. Wanting to know more about our users is good, but too much curiosity can cause users to abandon before they find what they came looking for.
Covidbot is a mexican chatbot that fights misinformation about the disease by answering questions about myths and hoaxes. Fake news have been a trending topic during this pandemic, and this type of fact-checking services are useful to pick the signal in the noise.
Carinabot, created by the Spanish 1millionbots, answers questions about coronavirus from various Spanish and Latinamerican websites of public institutions or universities.
Most of these projects are conversational interfaces for a traditional service, consumed through an API. Instead of having users interact with different graphical user interfaces, chatbots and voice apps brings technology closer to humans, letting them ask their questions as they would to another human: using natural language in their mother tongue.
When looking at epidemiology data in a dashboard or a spreadsheet, one needs to know the exact word they are looking for: if the word used to name the number of people who died in a given period mortality, only looking for this word will answer the question. Conversational interfaces that rely in natural language processing with a well trained set of entities and intents will be able to answer questions such as “how many people died in June”, “what was the mortality last month”, or even a different version of the sentence with a gramatical or orthographical error. Follow up questions such as “and in Madrid?” could refine this query in a very fast and simple way.
Natural language processing and conversational interfaces make it possible to bring available data and information closer to humans. And this works in both directions: when asked conversationally, and able to respond in an open format, people are more likely to respond and to share more than they would in a traditional form.
These characteristics have made chatbots and voice apps helpful in the covid response. Thanks to this, more people have heard about chatbots and started using the, making it easier to chatbot creators to collect real user feedback from different segments in society, and some very interesting use cases to draw inspiration from. I hope the examples I collected for you give you ideas on how you can combine your data or APIs with conversational interfaces to help more real people in relevant, real world issues.
