← Back Mahika Shah
Digital Product Design · Zero-UI Innovation · Nov 2025 – Mar 2026

Lumen

More than 55 million people live with dementia, and Alzheimer's is the most common cause. I wanted to design something that helped without feeling like a medical device, and that never asked the patient to learn a single interface.

Remote Therapeutic Monitoring · Figma · Systems Thinking
Lumen

Role

Product Designer (end to end)

Focus

Zero-UI, Remote Therapeutic Monitoring (RTM), caregiver support

Status

Concept case study

Where the light had gone dim

The dignity gap in digital health

This started as a personal project because one gap kept bugging me. Most digital health products for Alzheimer's patients land in one of two buckets: complicated clinical tools that overwhelm families, or gamified "brain training" apps that feel patronizing. Neither one treats the patient with much dignity.

57M
People worldwide living with dementia; Alzheimer's causes about 60 to 70% of cases (WHO, 2021).
41%
Of nursing home residents have early-to-late stage Alzheimer's.
~4
Caregivers I interviewed to shape Lumen's design.

The clinical insight

Alzheimer's care happens at home, but medical decisions get made in the clinic. Lumen is designed around Remote Therapeutic Monitoring (RTM): tracking how a patient responds to a therapy at home and sharing it with their care team. It doesn't try to diagnose anyone, and it doesn't flood the family with medical alerts. As a concept, Lumen has not been built or assessed as a medical device, which RTM billing would require.

Reminiscence therapy uses familiar sensory triggers, like voices, music, and photos, to draw on long-term memory. Research suggests it can modestly improve quality of life, communication, and mood for some people with dementia, though the effects are small and inconsistent (Cochrane review). Patients gradually lose the ability to form new memories, but emotional and long-term memories often stay reachable longer.

What if a digital canvas could deliver reminiscence therapy on its own, with no navigation required from the patient, while quietly giving their care team data they can actually read?

The caregiver reality

18.4 billion hours of invisible labor

83% of care is unpaid family labor.
83% unpaid care
31 hrs/week average caregiver time
59% report high emotional stress
70% find coordinating care stressful

If a solution adds to the caregiver's workload, it won't survive real life. That idea shaped every decision in Lumen's software architecture.

Alzheimer's Association · 2024 Alzheimer's Disease Facts & Figures

Listening before designing

Memory loss shaped the product

I interviewed about four caregivers about their day-to-day experience caring for people with Alzheimer's: how they handle each day, what they had noticed, and what had started to get better. Four conversations is a small sample, so I treated what I heard as direction, not proof.

Several caregivers described major memory decline in the person they care for. That led to two product decisions.

Take reminders off the patient's memory

Smart speakers manage daily reminders and timers, and a familiar playlist plays from the tablet, so the patient doesn't have to remember them or operate anything.

Use reminiscence therapy

Photos, mementos, and familiar music prompt conversation about past events. Research suggests this can modestly support communication and wellbeing for some people with dementia, though the effects are small and inconsistent (Cochrane review).

Margaret and Sarah, below, are personas built from these conversations.

Two people, sharing one system

Two people. Two completely different interfaces.

Based on my caregiver interviews, I built two personas: someone in the early to moderate stage of Alzheimer's, and the family member caring for them. They need completely different things from the same system.

M
Primary User (persona)
Margaret, 76
Early-to-moderate stage · Lives at home

Gets anxious when she's left without anything to engage with. Typical digital interfaces (buttons, menus, home screens) ask too much of her and quickly turn into frustration.

Need: Zero-UI interaction
S
Secondary User (persona)
Sarah, 48
Margaret's daughter · Managing care remotely

Wants to know how her mom is doing day to day without hovering. She needs something she can bring to her mom's doctor, but the medical jargon overwhelms her.

Need: Make invisible caregiving visible
A closed loop of care

The Zero-UI bedside canvas and caregiver app

Lumen works as a closed loop of hardware and software. A locked-down bedside tablet acts as an ambient canvas for Margaret, and a mobile app serves as Sarah's caregiver dashboard.

The Canvas (patient-facing)

A locked tablet sitting in a charging stand. For most of the day it shows a soft sage-green glow that slowly breathes. There's nothing to navigate. The whole screen is one big touch target. Smart speakers in the room handle daily reminders and timers by voice.

The Trigger

Margaret just rests her hand anywhere on the screen. That simple, whole-hand touch starts a session, playing voice notes from family alongside familiar photos and music.

The Caregiver App

A mobile app where Sarah uploads memories, manages the audio library, and keeps a quiet eye on how Margaret engages over time.

The RTM Data Loop

In an RTM-style model, engagement logs would sync to the cloud so clinicians can review behavioral trends, and the family never has to decode complex medical data.

Trimming the wick, one wrong turn at a time

Accessibility constraints as design drivers

Designing for dementia meant throwing out a lot of standard UI patterns. A typical 44x44 pixel touch target just doesn't work for someone with reduced motor control and slower visual processing.

Visual and audio cues over buttons

My first version had a simple play/pause icon. I replaced it because abstract symbols lose their meaning as cognitive decline progresses. The fix was gross motor touch: the entire screen became the trigger.

Instead of buttons, I leaned on color (sage green for calm) and slow, soft transitions that fade in gently so nothing causes a startle. The UI steps out of the way completely and the content is all that's left.

Tending it without hovering over it

Designing the RTM dashboard

Lumen is designed to pass data along rather than diagnose, so the interface is intentionally restrained. It turns the patient's daily engagement into quiet, readable trends that caregivers can understand and doctors can act on.

Why simple is better (what I left out)

Health-tech is full of diagnostic dashboards. I deliberately stripped out the medical complexity to protect the caregiver's mental load.

✕ No line graphs or jagged charts. Clinical charts spike anxiety. I used soft area charts in a single color with varying opacity, so trends show up without looking like a stock market crash.
✕ No red/green success metrics. Alzheimer's doesn't have a "success" state. Sage green only ever means presence (the canvas was used), never a medical judgment.
✕ No cognitive scoring. Alzheimer's doesn't get better, and a "score" either creates false hope or unnecessary dread. That data belongs with the doctor, not on the family dashboard.
✕ Plain-language framing. Instead of a chart labeled "Weekly Engagement," the app says it the way a person would: "Mom found comfort in 4 memories this week."
← Swipe to explore →
  1. 01

    Home

    Opens with one plain sentence ("Mom engaged with 3 memories today") instead of a chart. The presence ring shows minutes of gentle engagement, and today's memories are listed by time of day, followed by a single observation like "Afternoon sessions seem to reduce evening pacing."

  2. 02

    Memories

    The library is grouped by rhythm (Morning Calm, Mid-day Stimulation, Evening Wind-down), so caregivers choose what plays when, not only what plays. Each card shows the photo, clip length and when it was added.

  3. 03

    Upload

    Two steps: a familiar photo, then a story in the family member's own voice. A prompt encourages saying your name and one sensory detail, like the smell of pine trees, because voice and senses anchor recognition better than images alone.

  4. 04

    Health

    RTM trends shown as a soft, single-color engagement rhythm over 7 days, with average sessions, average duration and the most engaging memory. One tap exports a 30-day summary for the doctor, framed as records to share, not a diagnosis.

What the light showed me, looking back

What this project taught me

The best assistive technology disappears

The patient should never feel like they're using a computer. Taking the UI away completely forced me to rely on ambient cues and physical affordances instead.

Restraint is harder than adding features

Deciding what not to build (no live surveillance, no cognitive scoring, no push notifications) was just as hard as deciding what to build. The app is simple because trust matters more than dashboards.

Understanding the clinical ecosystem

Studying how RTM works taught me how to think about fitting a really emotional experience into existing clinical workflows, so doctors get what they need and families don't carry extra weight.

Next Project

Rally: an AI food ordering app

View Case Study →