HERHEALTHAI

• 

  The landing page of our solutionhttps://github.com/Linda-riziki/Cervicare-Web-App

Inspiration

Cervical cancer is the 2nd most common cancer among women in Kenya, with 3,000 annual deaths primarily due to late diagnosis (Kenya Cancer Registry). What shocked us was that 80% of cases are detected at advanced stages. Despite being preventable with early screening.

Our inspiration came from three key gaps:

1. Low screening rates: Only 16% of Kenyan women are screened regularly due to cost and accessibility barriers.
   
2. Fragmented care: No integrated system connects screening, diagnosis, and treatment.
   
3. Financial opacity: Patients often abandon care when faced with unexpected costs (e.g., a LEEP procedure costs ~KSh 15,000).
   

We asked: How might AI bridge these gaps to save lives?

What it does

HERHEALTHAI is an end-to-end AI-driven care platform that:

• Predicts risk: Uses a Random Forest model (accuracy: 92%) to classify women into Low/Medium/High risk based on:
  $$ \text{Risk Score} = f(\text{Age}, \text{HPV Status}, \text{Sexual History}, \text{Pap Smear History}) $$
  
• Coordinates care: Chatbot answers FAQs, while a patient-clinician portal tracks symptoms and appointments.
  
• Democratizes access: Real-time inventory tracking for Pap smear kits and cost transparency (e.g., HPV test: KSh 1,200).
  

Impact: Early detection for 30% more women, reducing late-stage diagnoses by half.

How we built it

Tech Stack:

• Frontend: HTML/CSS + JavaScript (mobile-first for low-resource settings).
  
• Backend: Express JS (REST APIs for clinic integrations).
  
• AI/ML: Scikit-learn Random Forest.
  
• Database: MongoDB Atlas (secure, cloud-based).
  
• Chatbot: LangChain + n8n + Streamlit (multilingual support).
  

Key Workflows:

1. Data Collection: For the chabot
   
2. Model Training: Optimized for precision (minimizing false negatives) using GridSearchCV.

Challenges we ran into

1. Cost Barriers: Hardware for GPU training was prohibitive.
   
   • Solution: Leveraged Google Colab’s free tier.

Accomplishments that we're proud of

• Local Impact: Designed for Kenya’s unique constraints (e.g., intermittent internet).
  
• Holistic Approach: First solution combining AI + supply chain + financing.
  

What we learned

• AI ≠ Magic: Models need localized data, a "global" cervical cancer tool would fail here.
  
• Stigma Matters: 40% of pilot users requested anonymity in sexual history logging.
  
• Scalability Tradeoffs: Random Forest was chosen over neural networks for interpretability in clinics.
  

What's next for HERHEALTHAI

1. Expand Partnerships: Onboard 10+ clinics in Western Kenya (USAID grant pending).
   
2. HPV Integration: Add vaccine tracking via Kenya’s National Immunization Program.
   
3. Offline Mode: Cache predictions for areas with no internet (using CRDTs for sync).
   

Long-Term Vision:
$$ \text{FEMGUARDIANS} \rightarrow \text{HERHEALTHAI}: \text{A generalized platform for breast cancer, maternal health, and beyond.} $$

Call to Action: "Join us in making cervical cancer history one prediction at a time."

Built With

• database:
• db
• frontend:-html/css
• javascript-(responsive-for-mobile).-backend:-javascript-(express-js).-ml-model:-scikit-learn-random-forests.-chatbot:-langchain
• mango
• n8n
• streamlit