DadaCare: Tech-Enabled Cervical Cancer Companion

Inspiration

Cervical cancer continues to disproportionately affect women in rural and underserved regions. We were inspired by the stories of women who suffer silently due to late diagnoses, lack of awareness, and inefficient health systems. The word "Dada" means "sister" in Swahili — this app is built as a digital sister, a companion to support, inform, and connect women and healthcare providers with the right resources at the right time.

What it does

DadaCare is a Flutter-based mobile app that empowers patients and doctors to better manage cervical cancer care. It provides:

OTP-based patient login (no email needed, to accommodate rural populations).
Book Appointments with hospitals based on availability and payment method (Cash, M-Pesa, Credit).
Nearby Clinics Page using Google Maps API to locate and display free or nearby screening centers.
AI Risk Prediction using a custom-trained ML model that suggests pap smear frequency based on patient history.
DadaCare Assistant, a multilingual chatbot powered by Dialogflow ES that supports voice input/output and answers questions about cervical cancer.
Doctor Dashboard with features like inventory management, patient referrals, AI recommendations, appointment viewer, and an integrated e-Registry.
Live Research Paper Explorer powered by EuropePMC API, with full-text reading inside the app.
Secure Referral System with file and note sharing between hospitals.

How we built it

Frontend: Built using Flutter (Dart), designed for cross-platform deployment with clean UI/UX.
Backend:
- Firebase Authentication (OTP and Email-based),
- Firebase Realtime Database for e-Registry, patient referrals, and appointments,
- Flask + PythonAnywhere backend for AI model hosting and logic.
ML Models: Custom trained models using patient risk factors (e.g., age, smoking, HPV).
Dialogflow ES: For chatbot intelligence with Kenyan Swahili and English support.
EuropePMC API: For live medical research paper integration.

We kept data security in mind by generating UUIDs for patients and never storing identifiable data in analytics or registries.

Challenges we ran into

Bad Datasets: Public cervical cancer datasets were inconsistent and noisy, requiring a lot of cleaning before training the model.
Tool Limitations: Firebase and Dialogflow ES are powerful, but full functionality often requires billing and credit cards — a challenge for grassroots teams.
Model Conversion: Finding a way to intergrate the models to flutter.
Voice Processing: Dialogflow ES had limited Swahili NLP support; we had to simplify intents and fallback logic.
Referential Logic: Creating a smooth, ethical referral workflow that protects privacy while maintaining usability took iteration.

Accomplishments that we're proud of

Built a full-stack application from scratch that brings real impact to cervical cancer care.
Integrated live AI recommendations in a low-resource setting with no server costs.
Created an interactive e-Registry that allows research without sacrificing patient privacy.
Enabled doctor-to-doctor referrals — a feature missing in many local systems.
Embedded live scientific paper access and in-app PDF viewer to empower practitioners with the latest evidence.

What we learned

Health tech needs to be culturally sensitive: even simple things like login methods affect adoption.
Voice assistants and AI must be tuned to local realities — language, internet speed, and literacy.
Data ethics and usability can go hand-in-hand with the right design thinking.
Real-world testing matters: feedback from actual clinicians helped refine features and flow.

What's next for DadaCare

Implement M-Pesa API for real appointment payments.
Partner with public health entities to collect verified e-Registry data for model retraining.
Add teleconsultation features using secure video APIs.
Expand chatbot to support more regional languages (e.g., Kikuyu, Luo).
Enable doctor credential verification through Kenya Medical Practitioners Board APIs.
Build a government dashboard for heatmapping cancer trends and bottlenecks. The ngo dashboard will perform network analysis and help allocate donated resorses