Sehat Sathi
A wearable ECG prototype for rural Kutch, held on by magnets meant to double as its contacts, lit through a motif from local embroidery.
Seven named versions in one studio, from a printed case to a petal shell lit through a Kutch motif. Never clinically tested.
- my part
- Designer and maker, working alone, from research to electronics
- when
- June to August 2024
- context
- UNSW Master of Design, Design Studio 2 (DDES9017)
- team
- Solo studio project. The working code for the v3 board came with help from AI.Gerhard Bachfischer was my tutor
- status
- Prototype
- size
- 65 × 30 × 27 mm, the printed shellsourceMy own measurement of the printed shell: 65 mm long, 30 mm wide and 27 mm high.

Sehat Sathi, in 3D

outcomes
7
named versions, v1 to v4measured · sourceCounted in my UNSW Design Studio 2 report (Aug 2024), p.14 to 19: v1, v2, v2.2, v3.1, v3.2, v3.3 and v4.
The brief
I was born and raised in Gujarat, so for Design Studio 2 at UNSW I gave myself a brief close to home: a heart monitor for the villages of Kutch. A hospital is a long way off: on the map in my report, the drive from Tera to Leva Patel Hospital Road in Bhuj is 81 km and 1 h 47 minreported · sourceGoogle Maps route from Tera to Leva Patel Hospital Road (81 km), on p.6 of my UNSW report, Aug 2024. The same page's label says the nearest hospital is 70 minutes away; I use the map's number.. It had to be something people would wear every day, and a medical device on the body invites questions, so I made it look like it belonged to the region once the size and the interactions were settled.

motif from Kutch embroiderysourceUNSW report p.17 and p.19.
one buttonsourceUNSW report p.15.
motif from Kutch embroiderysourceUNSW report p.17 and p.19.
one buttonsourceUNSW report p.15.
Four directions
Over the studio I made seven named versions, in four directions: a printed case, an e-ink face, light and touch, then the motif shell. Here are the main ones. The decisions below are where each one turned.

v1, a printed case

v2.2, the e-ink face

v3.1, how it attaches

v3.3, the first motif shell

v4, the petal shell
What I mapped first
I looked at two portable ECGs, D-Heart and VivaLNK. My report lists possible weaknesses for each, with no prices or measurements behind them: D-Heart’s 12-lead setup may need training, VivaLNK’s battery may drain quickly, and both may cost too much for low-income villages. So the brief became a monitor someone could wear every day with no training.
decision 1
It had to look like it came from Kutch

After a digital mind map and some paper notes, I drew this map by hand. It puts direct factors like diet, physical activity and medical access next to indirect ones like transport, social norms and socio-economic status, and next to the life of the region, from festivals to physical labour. At the top right, traditional aesthetics sits beside wearables. My premise was that a device that looks medical might put people off wearing it, so the form had to come from the place.
Making it smaller
First the connectors came out, then the screen.
decision 2
The e-ink screen showed a face

v1 was a printed case around the board and sensor. For v2 I tried an e-ink screen, because it doesn’t give off light, and I took inspiration from emotionally durable design and gave it animated faces instead of data. Here it’s on the bench, wired to three electrodes. After my first assessment I shrank it into v2.2 by taking out connectors and tidying the wiring, and that version ran real-time activity tracking, trained on a small sample of data.
decision 3
The screen came out because it wouldn't shrink

By v2.2 I couldn’t make it any smaller with the screen inside, so I changed direction. I moved to a newer processor, a XIAO ESP32-S3, soldered by hand onto a 6-axis IMU and an ECG sensor board; this is the test circuit on 17 July 2024. I got working code for it with help from AI. The screen’s job went to an RGB light, a button and a haptic motor, which is why Sehat Sathi talks in light: the screen was the part I couldn’t shrink.
21 × 17.8mmreported · sourceSeeed Studio, XIAO ESP32S3 getting-started wiki, specifications table: 'Dimensions 21 x 17.8mm' (read 22 Oct 2026).the XIAO ESP32-S3 board
On the body
Then it had to live on a garment: how it holds on, how it lights up, and where it sits.
decision 4
The magnets that hold it on are meant to be the ECG contacts

By v3.1 I wanted the device to come off the garment for washing and charging, so I tested how it would attach: a puck on a base, wired to three electrodes. In v4 the base plate holds three magnets, one in each pocket, and they are meant to do two jobs: hold the device to the top, and serve as the contact points for the ECG electrodes. The idea is that you lift it off, the top goes in the wash, the device goes on charge, and it can be shared. Sharing came from the inner loop of the Ellen MacArthur Foundation’s butterfly diagram, which is share.
3magnetsmeasured · sourceCounted: three round pockets in the base plate of the exploded drawing of the final design, UNSW report p.19 (Aug 2024), and one magnet in each pocket in the printed v4. A 2025 CAD render of the plate (file of 18 Aug 2025) shows the same three.in the base plate
decision 5
The light comes through an embroidery motif

I took the patterns from the region’s embroidery, redrew a petal motif in CAD and cut it through the shell, so the LEDs light the motif itself. Before the final shell I printed test tiles and held several of them up to a light; this white one is a lattice of petals. I also upgraded the LEDs so they could animate, for signals a wearer could learn: time to see a doctor, or a reward for meeting a goal.
My report shows two places it could sit, the neckline and the hem. Before my final submission I photographed my friend Maitri wearing it in both, and took one more of it on a dress form.


still on a USB cable
still on a USB cable
A year later
In 2025, a year after the studio, I came back to Sehat Sathi and designed a companion app for community health workers, built around a patient list and a step-by-step ECG recording. Its code was written mostly with AI, in Cursor, to my design. It has no link to the device: its connection and its recording are simulated in its own code. Its dashboard shows accuracy percentages with nothing behind them, so I’ve left those screens out.

1Pick a patient
Readings flagged as anomalies are tagged in red.

2Recorded
Thirty seconds, with the patient still. Then record another, or open the patient.

3Alerts
Urgent readings, with buttons to view the patient, mark it resolved or call.
Every name, reading and alert on these screens is demo data, and the recording is simulated. Two screens between the first two are left out: the patient's own page, which shows an AI confidence figure with nothing behind it, and the recording screen, whose copy promises high-precision monitoring from a simulated recording. Screenshots from August and September 2025.
