Team MarVel's "Mary"

Mary - A patient listener and your trusted companion

Comment

Inspiration

While none of our team members had a personal story, there have been some scares. We imagined a few personas and tried to empathize with them. Talking with better halves also helped to gain some different perspectives. Our primary persona was Trinity - a wife and working mother with a family history of loss.

We realized that for a person like Trinity this can be very difficult time. Talking to family is not necessarily easy. Surfing and research is not the first thing that comes naturally to such a person but talking to someone close. Hence we thought that having a conversational ai / chatbot can help. We were able to co-related our analysis with some industry stats like Conversational AIs are most like to be used in Health Care for Information.

We also observe that multi-channel is a critical need today. Everyone interacts with the internet using minimum 2-3 devices daily. Also interacting via things like Google Mini, Alexa, smart watches, etc. is increasing. Hence we thought that if we are going to create a Conversational AI, it necessarily needs to be omni channel.

What it does

Presently, "Mary" can do 3 key items:

  1. Anwser your questions across areas like What is Breast Cancer? Its types, symptoms, treatment options, etc.
  2. It helps finding a Clinic near your specified location (US as of now)
  3. It helps to book an appointment at the clinc/doctor

It also prompts for a Survey while leaving to get continuous feedback.

How we built it

Capabilities - We brainstormed on what capabilities will be needed by different personas through the journey from diagnosis to recovery. Key personas were:

  1. The Patient - Trinity now and in future i.e. at the time of diagnosis and then post recovery
  2. The Relative - Anderson

We then identified the MVP based on what is most important for the primary persona and what we can do in our time-frame

Technology - We had to make 3 primary choices:

  1. Bot Framework
  2. Back-end APIs
  3. User Experience

Bot Framework Based on our past readings and tinkerings, we had 3 options for BOTS - Azure Bots, Google Dialogflow and Bot Press. We could manage to get credits for Google DialogFlow (so Azure was dropped - have used up credits for something else) and Bot Press was free. So we did 2 small POCs to see how easy / disfficult it was to work with these technologies given that we don't use either in our day job and have limited time for the Hackathon (All of us were trying to balance between Projects, other initiatives, and kids homework and home chores). Based on our POCs we went with Google DialogFlow.

Back-end This was easy. We all are doing lot of Microservices development with Spring Boot and hence the obvious choice

User Experience We had 2 key criterias:

  1. Hybrid - Develop once for Mobile and Web
  2. Speed - Faster development. Again, none of us do high amount of front-end development and hence key for hackathon

Based on our past experience on the options available and our criteria, we went with Ionic (with Angular).

Challenges we ran into

  • Communication - We are 3 members and operate out of 3 locations - India, UK and Netherlands. Coming together after completing day job was challenging.
  • Lack of Business Domain Experience - Healthcare is a new domain for us. Also, none of us had any encounters with this in our extended families, knowing what is relevant has been a big challenge.

Slack, Trello, Teams, GitHub

Accomplishments that we're proud of

We are very proud of the fact that we got very diverse pieces of software all working and integrated to realize a dot zero version of "Mark" in a very short time. And the fact that we were able to work like a single unit even in the face of distance, time-zones and day job.

What we learned

Soft aspects:

  • The power of Empathy, Shared Goal and purpose
  • The need to and significance of communication and collaboration
  • The taste of making your own product

Technical aspects:

  • Bots is a new and exciting area for us and we always wanted to get here. This hackathon gave us an opportunity to get started in this space

What's next for MarVel

  1. Sort out Data Privary and Copyright issues on the content and make "Mary" accessible to a limited audience and gather feedback
  2. A firm product road map for future use cases aligned to Patient, Relative/Partner and their life cycle of this painful journey
  3. Collaborate with experts from domain to provide the most relevant capabilities

Built With

  • api
  • dialogflow
  • ionic
  • java
  • locationservices
  • microservices
  • springboot
Share this project:

Updates