ElfBot

Example Conversation 1
Time taken for call to be answered by helpline worker

Inspiration

Even though mental healthcare therapists are needed, many of the healthcare institutions actually suffer and do not have enough professionals to fulfill the needs of the patients quick enough. Demand for mental healthcare far outstrips the supply of therapists available, leading to lengthy delays for treatment as well as longer wait times between sessions.

What it does

ElfBot helps to check whether patients are utilising tasks and strategies from each therapy session, and asks them for simple feedback. It also helps to stand in for helpline workers and keeps patients engaged in between therapy sessions.

How I built it

Build purely on Dialogflow with Facebook Integration.

Challenges I ran into

Intents were difficult to formulate even after talking over the ideas with therapists. Training data was also difficult to come by due to confidentiality clauses.

Accomplishments that I'm proud of

Quick and simple integration with voice enable sentiment analysis.

What I learned

FB Integration is fast and easy with dialogflow

What's next for ElfBot

We plan to expand ElfBot’s repertoire of intents to include various use cases, like the ones often encountered on crisis helplines that one will call a helpline for support. (https://www.mentalhealth.org.uk/sites/default/files/life_lines.pdf) To use FB's NLG and pre-trained models in transfer learning for our specific use case to help in minimising risks during chats or helping with extension of the therapy sessions.