Iron Will

Iron Will, our project mascot :)

Inspiration

Anemia remains a pressing global health crisis, particularly affecting vulnerable populations. According to the World Health Organization, 42% of children under six and 40% of pregnant women worldwide are anemic. This alarming statistic highlights the urgent need for effective screening and education about iron deficiency, which can have devastating effects on growth, cognitive development, and overall health.

By providing a simple and effective way to assess iron deficiency, we hope to raise awareness and encourage proactive health management among women and children. This project could lead to better health outcomes, reduce the prevalence of anemia, and ultimately contribute to healthier communities around the globe.

What it does

Iron Will is a website that aims to provide women in underrepresented communities access healthcare. Through AI models that detect signs of anemia and chatbots that provides information about treatment. Moreover, with speech-to-text powered technology, registration for medical care will become more accessible to those with disabilities.

How we built it

We used mainly React.js to provide a seamless and interactive user experience. For the image analysis component, we developed Convolutional Neural Networks (CNNs) that detect signs of anemia through eye, nail, and palm images, utilizing technologies including Python, TensorFlow, Scikit-Learn, Matplotlib, dlib, and NumPy. Our AI-powered chatbot was implemented using Streamlit and powered by OpenAI's GPT-3.5 model, providing real-time health insights and personalized recommendations. Additionally, we integrated a speech-to-text feature within our forms utilizing the Web Speech API, allowing users to convert speech into text in multiple languages.

Challenges we ran into

Propelauth Integration: Encountered difficulties setting up authentication and managing user sessions
Streamlit: Faced issues with image upload and retrieval during data processing
LLM Costs: Had to manage expenses for accessing the OpenAI API to power our application
Encryption Issues: Struggled with ensuring secure encryption for form submissions
Learning Curve: Overcame the steep learning curve of TypeScript and React under time pressure

Accomplishments that we're proud of

CNNs for Eyes, Nails, and Palms: Developed convolutional neural networks (CNNs) to analyze images of eyes, nails, and palms for medical diagnostics. Achieved high accuracy in detecting relevant conditions through image analysis.
Implementing Chatbot: Successfully launched a chatbot to assist users with medical inquiries and support. Integrated natural language processing (NLP) capabilities to improve understanding of user queries.
Implementing Speech-to-Text Form with Multilingual Feature: Developed a speech-to-text form that allows users to submit information verbally. Incorporated multilingual support to accommodate diverse user populations.

What we learned

Our team learned a lot about combining speech recognition, natural language processing, and computer vision to solve accessibility issues in healthcare while developing Iron Will. We gained experience navigating the challenges of developing a complete solution that meets the many needs of marginalized communities. Through the process, we also learned how crucial it is to have efficient project management, cross-functional teamwork, and ongoing learning in order to get over technical obstacles and produce significant results.

What's next for Iron Will

Data Encryption: Consider using private key encryption to enhance security. Evaluate current encryption methods and identify potential vulnerabilities.
Chatbot Enhancement: Fine-tune the chatbot model specifically on medical data to improve accuracy. Incorporate relevant medical terminology and context for better understanding. Ensure the chatbot can access and utilize doctor-approved data and resources. Implement a feedback loop to continuously improve the chatbot's responses based on user interactions.
Disease Detection Expansion: Broaden the scope of detection capabilities to include diseases beyond anemia. Conduct research to identify common diseases that can be detected through similar methods. Collaborate with medical professionals to validate detection algorithms and ensure reliability.
User Data Protection: Integrate Propelauth to provide an additional layer of security for user data. Implement authentication and authorization protocols to protect sensitive information.
Application Development: Develop mobile applications in addition to existing web applications to enhance accessibility. Design user-friendly interfaces for mobile users to improve engagement and reach a broader audience.