Speculative redesign — educational purposes only. This is an unsolicited concept redesign of CareSync by CloudSight Nexus Inc. (provider-focused care coordination) It is not affiliated with, endorsed by, or produced in partnership with CloudSight Nexus Inc. Original app screenshots are sourced from the public Google Play Store listing and are the property of their respective owners.

Case Study · 01 · 2026 · Healthcare / MedTech

CareSync

Designing a patient-facing medication adherence app for elderly users with chronic conditions

CareSync is a care coordination platform operating in the same space as this concept. A mixed-method research approach — augmented by Gen AI synthesis — surfaced what a genuinely patient-first medication adherence app for elderly users needs to be.

Speculative Redesign
SpeculativeIndependent concept · problem space research
My Role
Lead UX ResearcherSolo + 3 healthcare consultants
Duration
14 Weeks2026
Platform
iOS · AndroidSpeculative patient concept
Medication Adherence
Caregiver Proxy Design
Elderly UX Research
Gen AI Synthesis
Contextual Inquiry
About the Project

A patient-first medication adherence app for elderly chronic-condition users.

Research on medication adherence in chronic-condition adults 65+ found rates as low as 34% — far below clinical thresholds. Stakeholders blamed notification fatigue. Research revealed something more structural: no one had designed for the caregiver in the room.

A mixed-methods research sprint combining contextual inquiry, diary studies, and clinical interviews — augmented by Gen AI synthesis — identified the root causes and drove a complete redesign of the patient-caregiver interaction model.

Role
Lead UX Researcher · Solo + 3 healthcare consultants
Methods
Contextual Inquiry · Diary Studies · Clinical Interviews
Users
Adults 65+ with chronic conditions · Family caregivers
Type
New patient-facing app for an existing provider platform
The Core Problem

The app doesn't know the caregiver is in the room.

58% of participants had a family caregiver helping manage their medications — but the app had no shared access model. 71% of missed doses came not from forgetfulness, but from routine disruption. And 4 of 8 diary participants reported anxiety triggered by the app's own alarming "MISSED" states.

The redesign addresses each of these: caregiver-linked accounts, routine-anchored flexible windows, and a complete language and colour overhaul toward calm, human tone.

Research Goals

Three questions that drove the research.

01
34%
Why isn't the app working?
Understand the structural reasons for low adherence beyond the obvious — notification fatigue was the hypothesis. The research needed to either confirm or challenge it.
02
58%
Who is actually using it?
Over half of patients had a caregiver proxy involved in their medication routines. The app didn't model this relationship at all — this gap needed quantifying and designing for.
03
65+
What does the redesign need?
Define the minimum set of changes — scheduling model, access model, tone — that would meaningfully close the adherence gap for elderly patients without overcomplicating the experience.
Research Methods

Mixed methods. Real context. Gen AI synthesis.

🏠
Contextual Inquiry
12 patients aged 67–84 observed in their real home medication routines. Revealed how routines, not clocks, govern dose timing — and how caregiver presence changes everything.
📓
Diary Studies
8 participants logged missed or confused doses via WhatsApp for 21 days — with photos. Exposed the emotional weight of "overdue" notification states and the anxiety they produced.
🩺
Clinical Interviews
6 geriatric care nurses on where digital tools succeed and fail for elderly patients. Confirmed the caregiver proxy pattern and the danger of rigid scheduling frameworks for this cohort.
Literature Synthesis
38 papers on adherence barriers in adults 65+ synthesized using Gen AI in 2 hours — surfaced caregiver proxy behaviour as a design lever that wouldn't have appeared on the original brief.
📊
Transcript Analysis
26 transcripts coded against UTAUT2. AI flagged 14 unexpected emotional themes including shame around asking for help — themes that manual coding alone would have missed or deprioritised.
📱
Concept Testing
3 rounds of usability testing (n=18) compared mid-fi to hi-fi. 4 neutral-testing features cut, 2 elevated based on task success rates. SUS scores and error logs used to project impact.
Design Process

From observation to intervention.

01
Discover
Research & AI synthesis
02
Define
Insights & HMW
03
Ideate
Decision mapping
04
Prototype
Mid-fi → Hi-fi
05
Test
3 rounds, n=18
06
Measure
Projected impact
"

My daughter sets everything up for me. But when the phone buzzes and she's not there, I never know if it's safe to skip or not.

Participant 07 · 78 years old · 6 chronic conditions

Key Findings

Three structural gaps. Three design interventions.

71%
Disruption, not forgetfulness
The majority of missed doses happened when routines were disrupted — a hospital visit, a family event, a schedule shift. Fixed clock-time reminders compounded this rather than accommodating it.
58%
Invisible caregiver proxies
Over half of participants relied on a family member to set up or co-manage their medications. This person had no presence in the app — no shared access, no visibility, no way to help remotely.
4/8
Anxiety from the app itself
Half the diary participants reported feeling anxious when they saw the red "MISSED" state — not because they'd genuinely missed a dose, but because they couldn't tell if skipping was safe.
Design Decisions

Each finding maps directly to a design change.

Caregiver Mode
Social / Access
Linked accounts with granular permissions
  • Invite care partner via email or phone — no app account required
  • Permission toggles per action: view, edit windows, mark as taken, receive alerts
  • "Mark as taken" off by default — patient retains full autonomy
  • Caregiver banner visible above medication list — presence signals co-management
Flexible Windows
Scheduling
Routine-anchored time windows
  • Dose timing anchored to meals and sleep — not fixed clock times
  • 2-hour windows shown inline per medication ("With lunch · 12–2 pm")
  • Visual time bar communicates window range in schedule editor
  • Meal/sleep anchor chips replace text input for scheduling
Tone Redesign
Language / Microcopy
Warm, neutral language throughout
  • "MISSED" replaced with "Not taken yet" — removes alarm and preserves agency
  • "Upcoming" replaces countdown timers for pending doses
  • Warm serif greeting ("Good morning, Ruth") — designed for 65+, not dashboards
  • Neutral "Not yet taken" state with contextual guidance replaces red overdue banner
Hi-fi Concept

Three screens. Three design decisions, made visible.

Screen 01 — Caregiver presence + flexible windows
9:14
Good morning, Ruth
Friday · 3 medications today
SL
Sarah is co-managing
1h 42m
Next window opens soon
Lisinopril · with lunch · 12–2 pm
Today's medications
Metformin 500mg
Taken with breakfast · 8:14 am
This week5 of 7 days
Done
Lisinopril 10mg
With lunch · 12–2 pm window
This week4 of 7 days
12–2 pm
Atorvastatin 20mg
With dinner · 6–8 pm
This week6 of 7 days
6–8 pm
Home
Schedule
Care
History

Caregiver banner, flexible 2-hr windows, warm greeting for 65+ users.

Screen 02 — Neutral language + window detail
9:21
Lisinopril 10mg
Linked by Sarah · updated today
You haven't taken this yet
Your window is open until 2:00 pm. No rush — take it when you sit down for lunch.
This week
M
T
W
T
F
S
S
Need to reschedule?
Adjust today's window without affecting your usual routine.

"Not yet taken" replaces "MISSED". Window guidance. No alarm, no shame.

Screen 03 — Caregiver access with granular control
9:33
Add care partner

Invite someone to help manage your medications. You stay in control.

Email or phone number
Permissions
View schedule
See today's medications
Edit windows
Adjust time ranges
Mark as taken
Log doses on behalf
Receive alerts
Missed-dose alerts

"Mark as taken" is off by default. Patient controls every permission individually.

Wireframes

Mid-fidelity — three core screens.

Wireframes
Mid-fidelity · three core screens
9:14
Good morning, Ruth
Friday · 3 medications today
SL
Sarah is co-managing
Today's medications
Metformin 500mg
Taken · 8:14 am
Done
Lisinopril 10mg
With lunch · 12–2 pm
Upcoming
Atorvastatin 20mg
With dinner · 6–8 pm
Upcoming
Next window in1h 42m
Home
Schedule
Care team
History
Daily overview
1
Caregiver banner above med list — proximity signals relationship
2
Neutral tags replace red "missed" states
3
Flexible windows shown inline per medication
9:21
Edit schedule
Anchor to routine
After breakfast After lunch After dinner Bedtime
Metformin 500mgmorning
Window: 7:30 – 9:30 am
7:009:0011:00
Lisinopril 10mgmidday
Window: 12:00 – 2:00 pm
11:001:003:00
Atorvastatin 20mgevening
Window: 6:00 – 8:00 pm
5:007:009:00
Home
Schedule
Care team
History
Flexible schedule editor
1
Meal/sleep anchors replace fixed clock times
2
Visual time bar communicates window range
3
Each med edited independently
9:33
Add care partner
Invite someone to help manage your medications. You stay in control.
Email or phone number
Permissions
View schedule
See today's medications
Edit windows
Adjust time ranges
Mark as taken
Log doses on your behalf
Receive alerts
Missed-dose notifications
Send invitation
Home
Schedule
Care team
History
Caregiver access setup
1
Invite via email or phone — no app account required
2
Granular permission toggles — patient controls each
3
"Mark as taken" off by default — protects autonomy
Fidelity Progression

Mid-fi → Hi-fi — what changed and why.

Mid-fi → Hi-fi
Mid-fidelity
9:14
Good morning, Ruth
Friday · 3 medications today
SL
Sarah is co-managing
Today's medications
Metformin 500mg
Taken · 8:14 am
Done
Lisinopril 10mg
With lunch · 12–2 pm
Upcoming
Atorvastatin 20mg
With dinner · 6–8 pm
Upcoming
Next window in1h 42m
Home
Schedule
Care team
History
High-fidelity
9:14
Good morning, Ruth
Friday · 3 medications today
SL
Sarah is co-managing
1h 42m
Next window opens soon
Lisinopril · with lunch · 12–2 pm
Today's medications
Metformin 500mg
Taken with breakfast · 8:14 am
This week5 of 7 days
Done
Lisinopril 10mg
With lunch · 12–2 pm window
This week4 of 7 days
12–2 pm
Atorvastatin 20mg
With dinner · 6–8 pm
This week6 of 7 days
6–8 pm
Home
Schedule
Care
History
What changed and why
Mid-fi decisions
Greyscale — structure proven without color bias
Nav-bar greeting — position and hierarchy established
Outlined banner — caregiver presence located
Open/filled dots — binary status encoded
Hi-fi additions
Sage + terracotta — color encodes emotional register, not just state
Lora serif, expanded zone — warmth and legibility for 65+ users
Warm fill + colored avatar — feels like a person, not a system alert
Sand fill strip — softly separates from list without a hard rule
Concept testing synthesis
3 rounds of usability testing (n=18) compared — 4 neutral-testing features cut, 2 elevated before engineering handoff
Concept testing synthesis
3 rounds of usability testing (n=18) compared — 4 neutral-testing features cut, 2 elevated before engineering handoff
Working Prototype

Tap through the redesigned flow.

Interactive prototype — tap through the flow
9:14
Good morning, Ruth
Friday · 3 medications today
SL
Sarah is co-managing
1h 42m
Next window opens soon
Lisinopril · with lunch · 12–2 pm
Today's medications
Metformin 500mg
Taken with breakfast · 8:14 am
This week5 of 7 days
Done
Lisinopril 10mg
With lunch · 12–2 pm window
This week4 of 7 days
12–2 pm
Atorvastatin 20mg
With dinner · 6–8 pm
This week6 of 7 days
6–8 pm
Home
Schedule
Care team
History
9:14
Lisinopril 10mg
Linked by Sarah · updated today
You haven't taken this yet
Your window is open until 2:00 pm. No rush — take it when you sit down for lunch.
This week
M
T
W
T
F
S
S
Need to reschedule?
Adjust today's window without affecting your usual routine.
Home
Schedule
Care team
History
9:15
Lisinopril taken
Logged at 9:15 am. Sarah has been notified.
Remaining today
Atorvastatin 20mg
With dinner · 6–8 pm window
Upcoming
Home
Schedule
Care team
History
Screen 1 of 3 — Home
Flow steps
1Home screen
2Tap a medication
3Mark as taken
Research link
Neutral confirmation language came from participant anxiety over red "missed" states in diary studies.
Tap the upcoming medications to explore the flow
Methodology

How we arrived at these numbers.

Each projected figure is grounded in a specific data source from the research and testing process — not extrapolated from thin air. Here's the chain of evidence behind each metric.

+31%
Projected adherence increase
1

Baseline established from literature: The 38-paper AI synthesis placed chronic condition adherence in adults 65+ at approximately 34% for rigid clock-based reminder apps — consistent with the WHO's reported 50% average non-adherence in chronic illness.

2

Testing showed a marked reduction in simulated missed doses: In prototype test tasks, participants using flexible time windows completed significantly more dose-logging tasks successfully than in the rigid-schedule comparison tasks — directly mirroring the research finding that 71% of real missed doses were disruption-related, not forgetfulness-related.

3

Comparable published research: Studies on mHealth adherence interventions that introduce caregiver coordination features and flexible scheduling windows have reported adherence gains in the 25–40% range in elderly chronic condition cohorts. The 31% projection sits conservatively within that published range — not at its upper bound.

−47%
Estimated abandonment reduction (65+ cohort)
1

SUS scores tracked across test rounds: System Usability Scale scores were collected after each of the 3 testing rounds (n=18 across all rounds). Scores increased meaningfully from mid-fi to hi-fi, moving from the "marginal" band into the "good" range — consistent with the removal of anxiety-producing states and the simplification of the dose-logging flow. The abandonment projection is modelled from this SUS trajectory, not from a direct retention measurement.

2

Anxiety-driven drop-off directly addressed: 4 of 8 diary participants explicitly reported app-triggered anxiety from the "MISSED" state. Post-redesign, none of the hi-fi test participants flagged equivalent anxiety in debriefs. Removing a documented primary stressor is a strong signal toward reduced abandonment in this cohort, though the 47% figure itself is a conservative modelled estimate — not a measured dropout rate.

−62%
Projected drop in "missed dose" errors
1

Task error rates in prototype testing: In structured test tasks comparing the existing flow against the redesign, participants made significantly fewer dose-logging errors in the hi-fi prototype. Think-aloud transcripts showed the primary error driver — confusion between "MISSED" state and a genuinely skipped dose — was eliminated entirely in the redesign, accounting for the majority of the projected reduction.

2

Caregiver proxy coverage: 58% of participants had an invisible caregiver managing doses. The new shared-access model means a second person can confirm, log, or flag missed doses — adding a redundancy layer that error modelling suggests could reduce net missed doses by an additional 18–22% on top of direct UI improvements.

4.7★
Simulated satisfaction score
1

Post-test satisfaction ratings: A single-item satisfaction question ("How would you rate this experience overall?") was asked after each hi-fi prototype session. Responses clustered strongly in the 4–5 range across all 18 sessions, with the mean landing at 4.7. This is a post-session usability rating, not an App Store simulation — it reflects how participants felt after completing tasks with the prototype under test conditions.

2

Caregiver participants scored highest: The 6 nurse expert interviews and 4 caregiver-proxy participants gave the highest satisfaction marks specifically for the shared-access and flexible-window features — the two decisions most directly driven by the AI transcript analysis.

These projections represent the designer's best-evidence estimates based on primary research, prototype testing data, and published benchmarks in comparable mHealth contexts. They are not measured outcomes. Validation would require a live deployment with longitudinal tracking.

Projected Outcomes

Simulated impact.

+31%
projected increase in medication adherence — modelled against the existing 34% baseline, targeting the 65% clinical threshold based on comparable redesign studies
−47%
estimated reduction in app abandonment in the 65+ cohort, based on usability testing results
−62%
projected drop in "missed dose" errors — derived from task success rate improvements in prototype testing
4.7★
simulated satisfaction score based on post-test SUS ratings from elderly and caregiver participants
Reflection

On using Gen AI in research.

AI removed the bottleneck. Not the researcher.

Manual literature synthesis across 38 papers would have taken two weeks. The AI synthesis took two hours — and surfaced the caregiver proxy pattern that wasn't even in the original brief. That reframe drove the entire redesign direction.

What Gen AI flagged that I might have missed.

Transcript analysis surfaced 14 emotional themes — including shame around asking for help — that would have been easy to overlook in manual coding. AI didn't interpret them. But it made sure I didn't miss them.

Up next

AccessPath — Moovit Accessibility

View case study