AfyaSasa

Track the Silent. Treat the Sudden. "AI-powered ovarian cyst care — for every woman, everywhere."

Comment

Inspiration

At AfyaSasa, we’re inspired by a single truth:

Every woman deserves a timely, confident, and data-driven diagnosis — regardless of her location.

Ovarian cysts are common; however, in many low-resource settings, such as Kenya, treatment decisions are often delayed or uncertain due to a lack of diagnostic tools, high costs, or data silos. We were moved by the silent suffering of countless women and the pressure faced by gynecologists, especially those working in under-equipped clinics. AfyaSasa was born to bridge this gap with AI, empathy, and actionable insights.

What it does

AfyaSasa is an AI-powered health platform that helps clinicians accurately predict, explain, and manage ovarian cyst cases while also empowering patients and streamlining administrative operations.

Key features by role: Clinician:

Input patient features to get real-time AI predictions and growth forecasts.

Receives cost breakdown (NHIF, insurance, out-of-pocket)

Views available medication/tools per facility

Sends printable or email-friendly reports to patients

Access dashboards: inventory, analytics, cost comparison

Uses the AfyaSasa LLM chatbot for ovarian cyst guidance

Gets automatic follow-up reminders

Patient:

Uses a chatbot to ask questions and understand symptoms

Book appointments based on location, clinic, and time

Pays through available channels and receives confirmation + reminders

Admin:

Manages all modules and monitors appointments, inventory, and cost analytics

Ensures time slot matching between patients and clinicians

Oversees reminders, reports, and facility performance

How we built it

AfyaSasa is a full-stack application using modern tools to integrate machine learning with real-time user interactions.

Tech Stack: Frontend: React.js, Tailwind CSS

Backend: Node.js, Firebase (Authentication + Firestore)

Machine Learning Models: Trained using scikit-learn

LLM Chatbot: Deployed using Modelbit

Deployment: Vercel (Frontend), Modelbit (Models + LLM)

Security: HIPAA-aligned design with RBAC, encrypted data flow

Other Integrations: Mobile payments (e.g., M-Pesa), Insurance logic, Notifications

Challenges we ran into

Creating accurate models with limited clinical datasets

Integrating real-time clinician-patient slot matching

Building explainable AI that clinicians trust

Balancing speed and usability across mobile and desktop

Managing secure, scalable, role-based access for sensitive medical data

Integrating both cost transparency and inventory data in a seamless way

Accomplishments that we're proud of

Achieved a moderate model accuracy in internal validation using supervised learning models

Built a working end-to-end multi-role system from scratch

Deployed both our prediction model and chatbot using Modelbit

Enabled real-time dashboards and appointment logic that’s production-ready

Created a system capable of transforming women's health access across clinics

What we learned

Building healthcare tech means designing for trust, safety, and empathy

Explainability is just as important as prediction accuracy

Patients need simplicity; clinicians need depth

We gained real-world experience in deploying ML in production

Collaboration across data science, design, medicine, and development is key

What's next for AfyaSasa

Integrate Cyan Systems for a smarter health data infrastructure and scaling model management

Expand to cover other gynecological conditions

Collaborate with hospitals and county governments in Kenya

Enable SMS support for low-connectivity areas

Add multilingual chatbot options (Swahili & local dialects)

Collect feedback from clinicians & patients to improve UX

Push for real-world clinical pilots & NHIF integration

Built With

Share this project:

Updates