BILKENT UNIVERSITY · DEPT. OF COMPUTER ENGINEERING · SENIOR PROJECT T2518 · 2025/26

CAREGIVER · 01A FIELD GUIDE TO CARING — № 01

ReMind

Care that thinks ahead.

A caregiver-first mobile companion for families living with Alzheimer’s and dementia. Medication, mood, vitals, and location — held together by a privacy-first design that runs on the device, not the cloud.

  • PROCESSINGOn-device, by default
  • FORAlzheimer’s & dementia
The product

See ReMind in two minutes.

A short tour of the caregiver experience — reminders, location safety, mood check-ins, and the calming activities patients do inside the app.

FIG. 00  ·  PRODUCT DEMO  ·  02:00  ·  CLICK TO PLAY WITH SOUND
Overview

Two people live the illness. The one who has it,
and the one who keeps watch.

ReMind is built for both, but it answers to the second. It is not a replacement for care. It is a steadier hand in the work of caring — keeping medication on time, drawing a soft line around the world, listening for the small signs that a day is turning, and sitting quietly when nothing needs to be said.

  • Live location & geofencing
  • Medication & routine reminders
  • Vitals from a smartwatch
  • Mood check-ins & MoodAI
  • Calming memory activities
  • Privacy-first by design
By the numbers

The scale of the quiet crisis.

Dementia is the work of millions of households. Most of that work happens out of sight, performed by family members who learned on the job.

55M

PEOPLE WITH DEMENTIA, GLOBALLY · 2024

SRC — WHO Global Dementia Observatory

83%

CARED FOR BY UNPAID FAMILY MEMBERS

SRC — Alzheimer’s Disease International

2.8×

PROJECTED GROWTH BY 2050

SRC — Lancet Public Health, 2022

FIG. 01

Who provides the care?

  • 83%Family caregivers, unpaid
  • 12%Professional, in-home
  • 5%Residential facility
FIG. 02

People living with dementia, projected (millions)

202055
203078
2040115
2050153

SRC — Lancet Public Health, 2022 · projected prevalence, 65+ population

A day with ReMind

A Tuesday in May.

Twenty-four hours of caring for someone with dementia, in two voices — the caregiver, and the app that quietly keeps watch.

  1. “First alarm. The day already feels long.”

    — Sarah, daughter
    06:30morning

    Soft light, gentle tone. “Good morning, Margaret.”

    routine wake
  2. She doesn’t want her pill. We’ve had this morning before.

    — Sarah
    08:15medication

    Donepezil 10 mg, on time. Logged: taken. ✓

    scheduled reminder
  3. Pharmacy run. I’ll be twenty minutes.

    — Sarah, on the watch
    10:42geofence

    Margaret stayed within the home zone. Heart rate 72. Calm.

    passive monitoring
  4. She seems quieter today. Hard to say.

    — Sarah
    13:20mood

    A gentle check-in, not a survey. She tapped: okay.

    mood log
  5. The watch buzzes on my wrist before I’ve finished my tea.

    — Sarah
    16:05attention

    Pacing pattern detected. Suggested: a familiar voice memo.

    MoodAI signal
  6. We watched the same show again. She laughed in the same place.

    — Sarah
    19:50memory

    Memory game finished: 4 of 5. Up from yesterday.

    cognitive activity
  7. She’s asleep. So am I, almost.

    — Sarah
    22:30good night

    All vitals normal. Geofence armed. Goodnight, Margaret.

    night mode

No fictional patient was used in the making of this page. Sarah and Margaret are stand-ins for the millions of dyads ReMind is built for — the moments are drawn from caregiver interviews conducted during analysis.

Capabilities

Three quiet capabilities, held together.

01

A soft line around the world.

Caregivers draw familiar zones — home, the garden, a daughter’s house. ReMind quietly notes when those lines are crossed, and only then. Live location isn’t surveillance; it’s the answer to a question already on the caregiver’s mind.

  • Caregiver-defined safe zones
  • Live location, family circle only
  • Alerts only when needed
FIG. 03 — GEOFENCE
02

A pattern is more than a moment.

A single check-in lies. MoodAI watches the shape of a week — engagement with games, response time to reminders, the rhythm of replies — and surfaces a small note when the curve starts to bend. Not a diagnosis. A nudge worth knowing.

  • One-tap mood check-ins
  • On-device pattern detection
  • Context for the doctor’s visit
FIG. 04 — MOODAI
03

Activities that calm, not test.

The activities inside ReMind aren’t about scores. They’re short, familiar exercises — matching pairs, gentle puzzles, a song to finish — designed to settle a restless afternoon and give the caregiver fifteen quiet minutes back.

  • Calming sessions, not assessments
  • Familiar imagery, generous timing
  • Progress shared only with caregiver
FIG. 05 — ACTIVITIES
Privacy & security

Care is intimate. The data of care is not a market.

ReMind’s privacy stance was the first design decision, not the last. Four commitments anchor the system.

  1. i

    Processed on the device, whenever it can be.

    MoodAI’s pattern detection, geofence calculations, and vitals interpretation run locally. Your phone is the brain, not a relay to ours.

  2. ii

    Protected in transit. Encrypted at rest.

    What travels between the app and Firebase is protected by TLS 1.3. What stays on the device is processed locally. Cloud-stored data is encrypted at rest with AES-256.

  3. iii

    No third-party advertising. None.

    There is no advertising surface in ReMind. There is no data brokerage. There is no “trusted partner” receiving anonymised exhaust.

  4. iv

    Minimal sharing. Clear controls.

    Family members see only what the caregiver chooses to share. Each toggle is a sentence in plain language, not a checkbox in a settings tree.

Documents

The work, in writing.

Five documents trace ReMind from a question to a working answer.

  • i

    Project Specification Document

    T2518 · problem framing, scope, stakeholders

    Open PDF
  • ii

    Analysis & Requirements Report

    T2518 · user research, requirements, use cases

    Open PDF
  • iii

    Detailed Design Report

    T2518 · system architecture, data model, UX

    Open PDF
  • iv

    Final Report

    T2518 · implementation, evaluation, results

    Open PDF
  • v

    CS491 Presentation

    slides · the project, told in an hour

    Open PDF
Team

Five students, one team T2518.

Bilkent University · Department of Computer Engineering · CS491 / CS492 Senior Project · Academic Year 2025/26

Want a walkthrough?

Read the reports for the full story, or reach the team for a guided tour of the design.