-
-
BumpCare
Inspiration
BumpCare AI was inspired by the need for accessible, reliable, and AI-powered maternal healthcare guidance. Expecting mothers, especially in remote or underserved areas, often lack immediate access to healthcare professionals; even after giving birth, there is a lack of accessibility to knowledge on what to do further. My goal was to create a smart assistant that provides instant support, personalized insights, and guidance for maternal health and newborn care.
What it does
BumpCare is an AI-powered maternal health assistant designed to:
Answer maternal health-related questions using a chatbot interface.
Provide risk assessment based on entered vital signs.
Support voice input for hands-free interaction.
Offer text-to-speech responses for better accessibility.
Assist expecting mothers and new parents with evidence-based healthcare information.
How we built it
Backend: Developed using Flask, which serves as the API endpoint for processing user queries.
Frontend: Built with Streamlit to provide an interactive and user-friendly UI.
Natural Language Processing: Utilized a Retrieval-Augmented Generation (RAG) model for accurate maternal health-related responses.
Voice Input: Integrated SpeechRecognition and PyAudio for speech-to-text conversion.
Text-to-Speech: Used Pyttsx3 to read out AI-generated responses.
API Integration: The Streamlit app communicates with the Flask backend via HTTP requests.
Challenges we ran into
Speech Recognition Accuracy: Capturing and transcribing voice inputs accurately was a challenge, especially for longer sentences.
Model Deployment: Ensuring the chatbot model ran efficiently on local and cloud environments.
Port Conflicts: Flask server occasionally ran into port conflicts, requiring manual resolution.
Error Handling: Improving user experience by handling edge cases like empty inputs and failed API calls.
Accomplishments that we're proud of
Successfully built an interactive maternal health assistant with both text and voice-based interaction.
Implemented speech recognition and text-to-speech, making the assistant more accessible.
Developed a functional Flask-Streamlit integration for real-time responses.
Created a structured RAG-based AI assistant capable of answering healthcare-related questions.
What we learned
How to integrate Flask with Streamlit to build an AI-powered chatbot.
Implementing speech-to-text and text-to-speech functionalities in Python.
Overcoming API request failures and connection issues.
Handling user input validation to improve chatbot performance.
The importance of maternal healthcare accessibility and AI-driven solutions.
What's next for BumpCare- Smart Care for Expecting Moms & Newborns
Using a RAG (Retrieval Augmented Generation) model with semantic search to improve the accuracy of answers given by the chatbot and keep updating the medical database to attune it to current trends.
There is still some issue with the accuracy of answers generated, but I hope to improve it further by using a larger model, which requires more RAM than what I currently have.
The voice-to-text features recognise the text but fail to generate a response if not put into the text box. I will try to fix it too.
Deploy on the cloud (AWS/GCP/Azure) for wider accessibility.
Enhance the AI model with medical database integration for even more accurate responses.
Add multilingual support to reach a broader audience.
Integrate a symptom checker for more personalized maternal health assessments.
Expand to mobile platforms for easier access via an Android/iOS app.
Collaborate with healthcare professionals to ensure the AI assistant aligns with medical best practices.
Log in or sign up for Devpost to join the conversation.