Skip to content

Latest commit

 

History

History
47 lines (32 loc) · 3.35 KB

ABOUT.md

File metadata and controls

47 lines (32 loc) · 3.35 KB
Raw

CareBuddy Design

Business Objective: The team wanted to make an app for individuals experiencing difficulties following healthcare routines, improve healthcare accessibility to all communities, and provide personalized healthcare by leveraging AI.

UI/UX

  • Our design was inspired by the movie Big Hero 6, a story about a young boy befriending a healthcare robot.
  • Using a template from Figma, we customized the design by adding a mascot for our app. We made sure to use uniform button designs such as an arrow button for transitioning between screens and sending messages. Once the preliminary design was finalized, we used TeleportHQ’s Figma plugin to convert the design to HTML and CSS. Lastly, we altered the design to fit our CSS layout.

Front-End

  • React was our chosen front-end framework because of the team's familiarity with Javascript.
  • We chose various CSS styling components, such as Sassy Stylesheets, Bootstrap, and MaterialUI to have an abundant resource of front-end library components, such as forms, buttons, and auto-sizing text boxes.
  • We created a chat form for the user to help CareBuddy provide customized responses.
  • The majority of the LLM interaction occurs on the Conversation screen through REST GET and POST API requests facilitated by the Axios HTTP client
  • We incorporated accessibility recommendations by adding aria labels to HTML elements and utilized Semantic HTML with the help of Siteimprove Accessibility Checker Chrome plugin

Back-End

  • CareBuddy uses a View-Controller architecture to enable rapid development and simplicity.
    • Our time constraints and priority to develop the core features without being concerned with the technical aspects of integrating and mapping database models influenced our decision.
  • We chose Express.js and Node.js to integrate back-end services to create GET and POST Endpoints to save and store the user's Favorites.
  • We effectively used in-built React utilities to allow short-term data storage, allowing us to demonstrate the core functionality of CareBuddy.
  • We used OpenAI after comparing a sampling of popular AI APIs, such as Amazon Bedrock, Azure Cognitive Services, Meta's LLAMA2, BERT on HuggingFace.com, and Groc Cloud.
    • OpenAI had the easiest integration documentation, simplest API schema, accurate responses, and affordable pricing options.

Future Development

  • Scalability & Optimization:

    • We would like to deploy this website to AWS and implement a PostgreSQL database.
    • We would like to expand this into a patient monitoring system that will complete the feedback loop to the provider.
      • We believe sending the patient’s data back to the practitioner will better inform the physician about what care to recommend to the patient.
    • Create additional features, such as CareBuddy being able to schedule appointments.
    • Refactor the codebase to utilize React Native, Swift, or Flutter to ensure a modern, scalable mobile architecture.

  • Compliant Security:

    • Integrate User OAuthentication to be compliant with website security standards.

  • Enhanced AI Capabilities:

    • Integrate healthcare LLM models and utilize RAG to custom-tailor the responses based on the user's search history and favorites.

  • UI Enhancements and Branding:

    • Allow for a light and dark mode toggle in the screens
    • Swap out the CareBuddy mascot to avoid copyright infringement