Authors

Abstract

Meddy is a virtual assistant that answers popular questions about COVID-19 and provides feedback to the medical community about information needs. The chatbot will be modeled upon the intent-entity structure proposed by Sánchez-Díaz [1] and the workflow outlined by Reyes [2]. We use Dialogflow’s NLP capabilities to implement this model and workflow. Meddy will be released to the public through Facebook Messenger and Google Assistant which are two very accessible platforms. Feedback and data analysis are just as important for us, that is why unknown questions will be reviewed by experts in order to generate a response and then update the database systematically. This project is being supported by Tec Salud and Meddy will be embedded to the official website on COVID-19 maintained by Tec de Monterrey.

Expected outcome

On the one hand, we expect to generate a massive database containing high quality information about this disease and people’s most important concerns. Questions are expected to follow a Pareto distribution, in which frequent questions constitute less than 20% of the total, and this phenomenon will allow medical experts to specifically focus in those information necessities. On the other hand, we foresee that user experience will improve proportionally to the number of interactions, as the database grows with new questions and answers. However, this can be limited by the capacity of response of human experts answering the questions.

On top of all, developing an agent with AI, that gives support to medics and other health service providers obtaining relevant information from the clinical questioning such as, clinical and epidemiological history, graduation of symptoms with clinometical scales and clinical data related to the clinical state of this group of patients; would improve the efficiency of the medic and provide better quality for each individual in need.

Finally, through these two goals we expect to diminish the anxiety that this type of event causes for the general public and physicians. We envision a world in which medical providers have a virtual assistant that aids in the retrieval of critical information about patients, and health authorities have enough information to focus in what matters most to the people.

References

[1] A Knowledge-Based Methodology for Building a Conversational Chatbot as an Intelligent Tutor (Modelo para utilizar intenciones y entidades en el contexto de reconocimiento de lenguaje natural) [2] Methodology for the Implementation of Virtual Assistants for Education Using Google Dialogflow (Modelo para implementar un chatbot usando Dialogflow)

Hackathons

Technologies

dialogflow, firebase, javascript

Devpost Software Identifier

257425