Mibot is a depression and anxiety information and symptom severity diagnostic chatbot.
I built Mibot (Mental information robot) as part of my final year project at Brunel University. According to Adult Psychiatric Morbidity Survey (APMS, 2014), one in six adults in England had a common mental disorder (CMD) with only a third of them receiving professional treatment. Some of the main barriers to help-seeking are; lack of insight and mental health literacy, self-reliance, lack of accessibility, confidentiality and trust, and fear and shame.
Mibot’s purpose is to help people overcome some of these barriers by using directed dialogue to help people check their depression and anxiety symptoms, give them valuable information on these mental problems, and provide help-seeking options. There is plenty of room for improvement, so please feel free to engage in a conversation with Mibot and give your feedback by commenting or filling up the survey.
This is an amazing project and I wish you all the best in your pursuit to reduce the social stigma around mental health literacy.
I just checked out the Bot and the conversation is extremely detailed and yet very natural and friendly. Great job at building this bot
We would love to show case your bot in our Bot Examples pages for more visibility, and will try to include it one of our newsletter in the future. Let us know if this is something you would be interested in doing.
From what i understand, with this bot, you are looking to increase the awareness around mental health issues. I think if we work together, we can get some reputed news sites and other stakeholders to cover this Bot in their publications, which can give this bot a much greater and well deserved visibility.
Would love to know your thoughts around this.
Please drop me a mail at: vinit@hellotars.com to discuss this further. @ish Please take a look at this.
Hi George,
Excellent work on this! I think you have found a great use-case for chatbots!
Initial feedback: there are moments that I would like longer message delays between speech bubbles. My general rule is two seconds of delay for each line of text (as displayed in the gambit not the actual conversation). This gives a better illusion of a chat response. I think for your purpose you can speed it up and go 1.5 seconds per line.
I think this type of project has a lot of potential. I’d like to see experiments with conditional jump logic and allowing the user to type responses from time to time. Especially at the end, asking the user to type how they are feeling in their own words. Then you could send them to different responses depending on keywords. Like if “suicide” or “suicidal” or another common phrase associated with this is used in their response, direct them to the suicide hotline that they can click and call on the spot.
Appreciate the time you spent to try Mibot and your feedback.
I agree that the message delays could be longer in some larger message bubbles, especially when two or three chat bubbles are filling almost the whole interface. However, the idea behind the delay not exceeding the 2.5 second frame is to keep the users engaged in parts where the text displayed might not be as interesting for all users (e.g., information about the questionnaire name and its creators).
Regarding the conditional jumps based on user input, is something I wanted to add to give the user some control over the conversation and an alternative input to fixed option buttons. Unfortunately, this was part of my final year project and only a part of my dissertation, so I couldn’t spend as much time as I wanted to. I really like the idea of conditional jumps for different types of help options and especially the hotlines, since people tend to postpone their problems, even if they have to do with their well-being.
