A daily briefing, not a dashboard.

AMBL is two things designed together: a smart shoe and its companion app. Balance is one of the hardest Parkinson’s symptoms to treat: postural instability doesn’t respond well to medication. Symptoms change by the hour. Between visits, tracking the disease is theirs to do alone.

The apps that exist watch and report, so a clinician can review later. I wanted something “proactive rather than someone checking data and then reacting.” The hardware idea was stochastic resonance: a little random noise at the sole of the foot, which research suggests can help weak sensory signals register. The catch is that it only helps at the right dose, on the right foot. AMBL is a concept, not a cleared medical device.

I started with the science, because here a wrong call makes the product harmful, not just bad. I directed a research brief across 8 studies and preprints and turned it into five rules the design couldn’t break:

• Real random noise, never a regular vibration pattern.
• A dose held between 25 and 90% of each foot’s measured threshold, with no “turn it up” slider.
• Choose the foot from pressure data, never from hand tremor.
• Back off automatically if walking gets worse.
• Gate onboarding on a measurable deficit.

Then I tore down StrivePD, the leading patient-facing Parkinson’s app, across 38 screens. I kept its honest empty states and dropped its log-first home. I built the app prototype from version 1 to the final, and handed it off to the team.

We contacted 100+ clinics, hospitals and care facilities to reach people with Parkinson’s, and none landed. So anything clinical had to trace to the brief.

• Literature · Alsaqabi 2025, under review · 14 people with early-stage Parkinson’s

  Stimulation increased sway on the more-affected side (p = 0.027). That side had been picked from hand tremor.

  The app never asks which hand shakes more. The shoes pick the foot from pressure data and back off when walking gets worse.

• Concept test · early app

  There’s a lot of data, but I don’t know what to do with it.

  The home screen was showing raw inputs. People needed the conclusion.

• Shoe prototype test · 5 testers · observed

  4 of 5 felt distinct intensity levels across the insole zones, and testers independently described the motion response as “balance” or “imbalance” detection.

  The feedback already carried meaning, so the app’s job was to name what people felt, not explain it with numbers.

I flipped the home screen from a dashboard to a daily briefing. The old home led with “Avg symmetry 84%” and two charts. Version 5 led with one sentence; the final went further, to one orb that mirrors the shoes and one thing at a time. The engine won’t claim a pattern without enough data, and a quiet day says it’s quiet. Falls are one tap away but never loud, and sharing with a care partner is the patient’s choice.

8 → 5

Studies and preprints in the research brief, turned into five rules the design couldn’t break.

38

StrivePD screens torn down: its honest empty states stayed, its log-first home went.

4 of 5

Testers felt distinct intensity levels across the insole zones of the shoe prototype.