Abstract

Our project is to create a chatbot to gather data from cancer patients, following the holistic health assessment which was previously given to them prior to checkups. The problem with the current form is it is tedious, time-consuming and often needs to be completed with the consultation.

We developed a proof of concept chatbot system that aims to make the data gathering process easier and more accessible for patients, and allows for a more tailored experience. This chatbot system includes APIs for the actual chatbot and data retrieval, as well as plans for future development of interfaces to allow doctors to edit chatbot dialogues.

What problem does this solve?

The current form of collecting the holistic health data is slow and inconvenient for both the patients and their doctors. It doesn't allow for the checkup or doctors appointment to be productive, and results in the appointment being taken up filling out the current EHNA survey.

UCLH wants us to create a solution to this inconvenient and time-consuming current method of data collection. The way we can do this is through developing a chatbot which can make the survey more fun to fill out, and more efficient. This will mean we can fill out the form, which usually occupies appointment time, prior to the meeting.

Key Features of the Project

• Adaptable dialogue model


• Chatbot API with 'Messaging' and 'Summary' endpoints


• Command Line Interface for chatbot testing


• React UI component which displays visual aspects of the chatbot


• Mockups for doctor interface and other plans for future development


• Documentation outlining use and setup of our API

Impact and Future Development

We have demonstrated the possibility of using a database model to solve the problem of gathering domain-specific information from a user.

We have designed and developed a system architecture for using this database model. We have made sure that we have created an extendable solution to allow for future/continued development. To make sure the project can be continued easily by future teams, we have provided API documentation, commented and well-tested code, as well as continuous integration.

Our project is hosted on GitLab, where our client has administrative rights. Our client has also been briefed on the possibilities of future work on this project, and we have laid out various possible forms of such work within this website.

Through writing medium posts, participating in both the National Academy of Engineering's Grand Challenges competition and Microsoft's Imagine Cup, we wanted to spread the word externally. We are also working on the publication of a paper which our client wanted, peer-reviewed by a medical student.