Loading..
Building DVMA: Gemini AI for Dementia Care
Back to all articles
AI & Healthcare7 min read

Building DVMA: Gemini AI for Dementia Care

How I built DVMA (Dementia Virtual Memory Assistant) — a Flutter and Firebase mobile app with Gemini AI chat, face recognition, and role-based access for patients, caregivers, and admins.

Muhammed Ajmal U K
Muhammed Ajmal U KFull-Stack AI Developer • MCA at CET

DVMA — the Dementia Virtual Memory Assistant — is the project I am most proud of. It is a mobile application designed to support people living with dementia in navigating daily life, preserving memories, and staying connected with the people who care for them. Building AI features for a vulnerable audience changed how I think about technology.

#Designing for Three Very Different Users

Dementia care is a team effort, so DVMA ships with three distinct role-based experiences. Patients get task management with smart reminders, a memory album, a personal diary, and an AI companion. Caretakers get monitoring, communication, and live location tools. Admins get user management, broadcast notifications, and a reports dashboard. Each role sees only what it needs.

#Where Gemini AI Fits In

  • A compassionate chat assistant for memory support and gentle task guidance
  • Prompt design tuned for clarity, patience, and simple language
  • AI help scoped to support — never replacing human care or medical judgment
  • Graceful fallbacks so the app stays useful even when AI is unavailable

#Face Recognition With ML Kit

One of the most meaningful features is family face recognition. Using ML Kit, the app helps patients identify family members and helps caretakers verify identities on the go. It is a small feature technically, but for someone experiencing memory loss, recognizing a loved one is everything.

#The Stack

  • Flutter and Dart for a single cross-platform codebase
  • Firebase Authentication, Firestore, and Cloud Storage for real-time sync
  • Gemini AI API for the conversational assistant
  • OneSignal for reliable push notifications
  • ML Kit and Cloud Vision for face recognition

#What This Project Taught Me

Building for healthcare means building with empathy. Every design decision — font sizes, reminder tone, what the AI is allowed to say — had to answer one question: does this genuinely help someone having a hard day? That discipline made me a better engineer, because the hardest problems in software are rarely technical.

Share

Enjoyed this article?

Leave a reaction or share your appreciation

Muhammed Ajmal U K
Written by

Muhammed Ajmal U K

MCA student at College of Engineering Trivandrum (CET) and Full-Stack AI Developer from Kerala, India. Creator of ToolPix AI, KallanCop, and DVMA.