Ausculo

India's public telemedicine platform, eSanjeevani, connects rural health centres to city doctors by video. But a video call cannot auscultate — a doctor cannot listen to a chest through it, and auscultation is first-line for respiratory and cardiac screening. Ausculo set out to close that gap with a device cheap and simple enough to actually reach a rural health centre.

I joined an existing hardware effort at CARE, IIT Delhi as its only designer, and owned research, the physical form, and the interface. Heart-health diagnostics is a subject I came to for personal reasons, which is a large part of why I stayed with it.

The work reached validated concept stage and was published at HFES 2025. It did not ship — the electronics were not ready to support the interface I had designed. I have kept that framing here rather than implying a launch.

Most of India lives in villages; most specialists work in cities. eSanjeevani's hub-and-spoke model bridges that by putting a Community Health Officer at the "spoke" — a Health & Wellness Centre — and a doctor at the hub. It works for consultation and triage. It fails the moment the doctor needs to hear something.

Cardiovascular disease accounts for roughly 28% of all deaths in India (WHO), and it is precisely the category where a chest sound changes the decision. Digital stethoscopes that could transmit audio already existed — they just had not reached rural India, and the reason was mostly price.

That last figure was not a guess. Interviews produced a clear willingness-to-pay ladder: ₹500–2,000 for medical students, ₹2,000–5,000 for MBBS doctors, ₹15,000+ only for established specialists. The brief I inherited assumed a ₹15,000 device. For the users it claimed to serve, that was already out of reach.

The project I inherited assumed ASHA workers would carry the stethoscope into villages. It was a reasonable-sounding assumption, and it was wrong.

I re-targeted the primary user to CHOs, with medical students as a secondary group (recording and replaying sounds has obvious teaching value). This one change cascaded through everything downstream — the interface, the training aids, the price ceiling, and what "simple enough" had to mean.

Doctors were genuinely hard to reach — rarely free during working hours, and unresponsive without a personal referral. That constraint shaped the method: I could not shadow or run long interviews, so I built approaches that fit the access I actually had.

• Secondary sweep — papers, plus Reddit, Twitter, LinkedIn groups and YouTube reviews, because Amazon and Google Trends only yielded generic feedback.
• Stakeholder interviews — doctors, medical students, ASHA workers, CHOs and telemedicine startups.
• Stanford BioDesign — disease-state analysis across respiratory, cardiovascular, gastrointestinal, obstetric and vascular auscultation.
• Competitive teardown — I bought and dismantled a Littmann CORE.
• Expert sessions — Dr. Desh Deepak (Pulmonology, RML Hospital) and Dr. Sanjay Sood (Director, eSanjeevani).

What each group actually said

The teardown was the most useful hour

Taking apart a ₹27,543 Littmann CORE surfaced problems no spec sheet lists. It took over 2.5 hours to charge and lasted 8 hours — not enough for an Indian hospital shift. Sound amplification amplified the rubbing of the chestpiece into something shrill. Worst of all was the workflow: you wear the stethoscope, then pick up your phone, then open the app, then re-pair it, and only then can you listen. Both hands are already occupied.

Context, not features, was the only real differentiator available. Eko and Littmann already had noise cancellation and AI murmur detection. What they did not have was a device that survives a rural Health & Wellness Centre.

The two rejected hypotheses below are here deliberately. Both were things I believed, built, tested, and abandoned — which is the part of the process that actually moved the design.

Doctors could not articulate a form preference in the abstract, and asking them to rate designs on a scale produces answers that shift with whatever they saw last. So I ran a forced-choice pairwise study instead — and had to solve two problems to make it fit inside the access I had.

Problem one: too many pairs. A full round-robin across 33 designs is 528 comparisons — nobody sits through that. I grouped the comparisons using a merge-sort–style divide-and-conquer scheme, which brought each session down to 28 while still producing a global ordering.

Problem two: scoring. I used the Elo rating system from chess, so each design's score updates against the strength of what it beat. Elo also sidesteps the independence-of-irrelevant-alternatives problem that rating scales suffer from, and it keeps the respondent's task down to one question: this or that.

No off-the-shelf tool did this, so I built the ranking site myself in 10 days, capturing each respondent's specialty, city, age and institution type alongside their choices.

Getting there took a dead end first. I kept refining techno-aesthetic detailing and every result still looked like every other stethoscope on the market. The break came from chess piece geometry — clean, gripped, visually distinct — which I then printed and hand-tested for ergonomics rather than trusting the render.

Constraint mapping, form variants, and printed candidates hand-tested for grip — the enclosure had to hold two PCBs 3cm apart while leaving the sensor face in direct contact.

Delhi — moderated think-aloud sessions with doctors. I asked participants to narrate what they were thinking, and set a completion time deliberately to add pressure, on the theory that mild stress exaggerates the friction you are trying to see.

Dharamshala — the round that mattered. Dharamshala itself is not remote, but the areas around it are, and it has a real density of PHCs and sub-centres. Access came through a contact who knew the CMO, Dr. Rajesh Guleri. The trip was not funded; I paid for it because testing a rural device anywhere else felt indefensible.

Because I had no fully working device, sessions ran as role-play: a CHO would place the stethoscope, take instruction from a remote doctor and work through recording, while doctors imagined reviewing the audio against a patient record. I gave participants a simplified heuristic checklist — visibility of system status, match to the real world, user control — and asked them to flag anything missing or unclear.

The device did not ship. The research did — as a published paper, a validated form direction, and a mapped integration path for whoever picks it up next.