The primary goal of this project is to create a chatbot system which will allow clinicians to build out their own conversations on an open-source and free-to-use platform and deploying these conversations onto Minecraft. The use of open-source tools is an integral part of our project as clearly stated by our client.
For our project, we had interviewed various stakeholders, including clinicians from the NHS and a university student. From these findings, we have been able to produce a MoSCoW, which will be a guideline for what is required and what functions can be optional for the system.
As stated above, we had interviewed various stakeholders for our system. Below, we will be providing samples of the interview questions and the answers they had given.
How do you feel about the support that you've received in the past?
“I didn’t really enjoy being face-to-face with someone like that. It almost felt like I was being interrogated. Face-to-face therapy can be pretty scary especially when you have social anxiety problems.”
Would you prefer having an alternate source of support? If so, what else?
"I'd feel a lot more comfortable if I didn't have to face someone directly. I just don't want to feel like I'm sick or anything."
How do you feel about the patients responding to your questions in face-to-face sessions?
Sometimes it can be pretty hard to get patients to actually open up to a clinician or counsellor face-to-face. You definitely need to have a lot of patience and help them through this barrier. It can definitely be part for those who have some form of anxiety for example.
Do you think a non-face-to-face solution would help?
I think it definitely would. Rather than them having to talk to someone in-person, we can totally have some other form of support. Especially with the COVID-19 pandemic over us, it's definitely going to be hard to get support out to young people.
Once these interviews had been completed, we created two personas to resemble the stakeholders we had interviewed. They will be presented below.
Daniel is a 19-year-old university student who is an avid gaming enthusiast. Due to the stresses of his course, his personal life, and the COVID-19 pandemic, he has been struggling with his mental health. He is unable to receive in-person help nor does he have the confidence to do so if it was available. Daniel has used our service to receive acceptance and commitment therapy. Since Minecraft is an interactive game, he has been able to continuously seek help and has seen improvements in his mood and general well-being. With the tasks created by the NHS clinicians, he has been able to ask the MineACT bot for help in low moments.
Lisa is an NHS clinician who rarely played video games, only when her children were younger. Using video games as a means of helping young people receive acceptance and commitment therapy is a new experience for her. However, Lisa is has been able to get used to the mechanics of Minecraft very quickly due to the simplicity of the game. Though, Lisa's role in the overall system doesn't require her to be present within the game. She has to be able to come up with tasks, including intents expected from the user. She will create these conversations in Voiceflow, a service which allows you to create "Alexa Skills".
Below is the use-case diagram that we have produced for the overall project. A list has been created to acccompany the diagram too.
| Use-case | Use-case flow | Primary actors |
|---|---|---|
| Creating a conversation flow using Voiceflow. |
|
NHS Clinician |
| Creating and maintaining Google Sheet. |
|
NHS Clinician |
| Using the plugin within Minecraft. |
|
Therapy recipient |
| Playing a treasure hunt game. |
|
Thereapy recipient |
After discussing further with our client, we have come up with a MoSCoW table. This table shows what our overall requirements will be and what we should focus on implementing into the overall system. The MoSCoW has been split into two sections; functional and non-functional requirements.
| Requirements | Priority |
|---|---|
| The game must be able to detect trigger words based on inputs from the user – the user should be able to enter trigger words into the chat which will then provide an appropriate response based on what they have written. These trigger words/intents must be stored in a suitable place such as a spreadsheet. | Must have |
| The game must provide an activity for the user to do when a trigger word has been detected – this activity will be created by the NHS clinician using Voiceflow or some conversation design system that they wish to use. | Must have |
| Clinicians must have a way to edit the system to tailor towards whatever requirements they may have – they will be able to create conversations within Voiceflow and edit the intents in the spreadsheet provided. This means that they can add more intents and accommodate for change in language over the years. | Must have |
| The system should have a way to reward the player when completing a task – they can be given in-game items or points which will aid them throughout the game such as better equipment or experience points. | Should have |
| We won't have customisation settings for now (e.g: allowing players to change the colour of text etc.) – our goal is to create the main system at hand and deploy it before having anything that will allow the user to customise aesthetics. | Won't have |
| We won't have a system where the clinician may want to keep track of the requests made and the subject matter – perhaps this can be done another time where they will have a spreadsheet containing this information and who from. | Won't have |
| Requirements | Priority |
|---|---|
| The system must be able to work fast; there should be relatively little delay between a user's interaction and the bot's ability to respond to them. | Must have |
| Information must be kept secure and in a readable format (i.e: intents, API keys etc.). | Must have |
System developed by Arifuzzaman Choudhury, Nishan Sen, and Ruochen Sun
With the support of our industrial partners Dr Joseph Connor (NHS) and Professor John McNamara (IBM)
Supervised by Dr Dean Mohamedally and Hongxiang Chen
University College London
