Graduation project: a chatbot for people with epilepsy

Stage 1. User Research

In the initial phase, extensive research on conversational flows was conducted, focusing on user habits, preferences, and frustrations with current chatbot tools. This led to the creation of detailed user personas representing the app's target audience and highlighted the need for more interactive chatbot solutions. Applying Nielsen's heuristic "Visibility of system status" ensured users were kept informed during interactions.

We also explored current health-related chatbots, like "May Health", a digital therapist for breast cancer care, and "Mila" from Albert Einstein Hospital, which assists patients with treatments in dementia, heart failure, and tracheostomy to map the user experience, identifying key points and risks.

Additionally, to develop the chatbot, I deepened my understanding of epilepsy, motivated by my own experience with the disease and assisted by medical research and my doctor. This thorough research was crucial for creating a chatbot that effectively met the needs of users with epilepsy.

Stage 2. Ideation & Conceptualization

After concluding the usability analysis and insights creation phase, I conducted an exploratory analysis of Facebook Groups about epilepsy, listing 74 frequently asked questions from users, companions, and family members. These questions were structured into a mind map and organized using the Card Sorting technique by myself and my advisors.

With mapped personas, we developed both the chatbot's persona, emphasizing empathy and trust, and user personas based on our research. Utilizing the Lean Canvas, we strategically planned the chatbot's development, addressing business objectives, user context, conversation requirements, user intentions, KPI goals, chatbot personality, conversational cycles, and marketing strategy.

Stage 3. Chatbot Design Execution

• Chatbot flow: The next step was to create the chatbot's internal flowchart, to then start its creation within the platform that was used in the final project. The flowchart was created to facilitate the creation of the chatbot's connections, subdivided into five areas with pre-defined questions.

  If the user does not find what they were searching for, they will be provided with the text box for suggestions, and then if these questions are selected to be answered, they will later enter the chatbot's question database.