Case study · 2019 to 2025 · empatica.com ↗
EmpaticaA medical device should feel as calm to the person wearing it as it is rigorous to the clinician reading it. That’s the concept. I spent six years on both sides of that wristband.
Empatica makes medical-grade wearables and the platform around them, so clinicians and researchers can read a body from a distance. An MIT spin-off between Boston and Milan. The same technology alerts a caregiver when a person with epilepsy has a seizure, with Embrace and then EpiMonitor, flagged likely COVID-19 infections before symptoms during the pandemic, and in November 2022 received FDA clearance as a Health Monitoring Platform.
I was its Senior Service Designer, and for a stretch its interim Product Manager. I owned the end-to-end service and product design, from the wearable in a participant’s hand to the data on a physician’s screen, and drove five departments through full product cycles, including the one that ended in FDA clearance.
The platform is a chain. I owned the design of every link, and the handoffs between them.
01 · Wearable
Collects skin conductance, temperature, pulse, oxygen and movement, day and night.
MineThe first hour with the device: pairing, wearing, charging, knowing it works. If this hour fails, no study happens.
02 · App
Collects from the device and uploads to the cloud. Shows the person what is happening.
MineThe honesty of a status. Is my data arriving? What do I do if it isn’t? Written for the person wearing it, never for the clinician.
03 · Cloud
Turns raw signals into clinically validated measures, the digital biomarkers the FDA cleared.
MineWhat the interface may claim. A number is only shown if a study says it is true. That rule was mine to hold across teams.
04 · Portal
Where a study or a clinic sees, exports and trusts the data.
MineThe protocol as the real navigation: participants, sessions, gaps, exports. Trust is a feature, so every value shows where it came from.
05 · Alert
A seizure alert reaching a caregiver by call and text, with a location. A possible infection flagged before symptoms.
MineThe sentence. What it says, what it never claims, what happens next. The two stories below.
In order. Public anchors, and my part in each.
July. EmbracePlus, the next-generation watch, was announced that November. A platform used mostly by researchers: a wearable, an app, a portal, and a lot of hands in between. I started by drawing the whole service, who does what in which order, before touching a screen.
An algorithm on wearable data could flag a likely COVID-19 infection about two days before symptoms. Validated with BARDA, CE mark in 2021. I designed the moment the alert reaches a person, and the days after it.
November. EmbracePlus, the Care app, the cloud and four clinically validated digital biomarkers, cleared as one platform. I had driven Product, Clinical, Legal, Sales and Marketing through the cycle that got there.
For a stretch I ran the room: prioritisation, scoping, alignment across five departments. Design did not stop; it got a seat where the trade-offs were made.
February. The epilepsy monitor moves to EmbracePlus and launches in the US, FDA-cleared for adults and children from six. A seizure is detected, a caregiver gets a call and a text with the person’s location, and the battery lasts a week. I designed the experience around that alert: the person wearing it, the caregiver receiving it, and a first setup made for a family instead of a research team.
With one lesson I have relied on since. In a regulated product the workflow is the product. The screens are downstream of who signs what, in which order, with what proof attached.
Before AI was in my toolbox, it was in the product. An algorithm decides something about a body. A person has to live with the sentence.
The hard part was never the model. It was the sentence. How do you tell a person with no symptoms that their body looks like it is fighting something, without panic and with one clear next step? I designed that moment and the days after it: what the alert says, what it never claims, and what the person and their organisation do next.
A wristband detects a convulsive seizure and alerts a caregiver’s phone: with EpiMonitor, a call and a text with the person’s location. Every false alarm costs trust. Every missed event costs more. I worked on what the person wearing it sees, what the caregiver receives, and how the two learn to rely on each other without anyone having to understand the algorithm.