CardioPulmo — AI heart and lung sound screening on a phone

Screening PWA Live 3 Models

CardioPulmo is a phone-based screening aid that listens to heart and lung sounds and returns a cloud-analysed screening read. It is built for the reality of Indian primary care and outreach camps: a health worker with a phone, a queue of people waiting to be seen, and no cardiologist or pulmonologist within reach that day. In the words used on the Clinoble homepage, it is AI heart-and-lung sound screening from the phone already in your pocket, with recordings analysed by three cloud-hosted deep-learning models with ambient-noise quality gating.

What phone-based auscultation screening changes

Conventional auscultation assumes years of trained ears behind the stethoscope. In a screening camp or a primary health centre, the person doing the listening is often an ASHA worker, a nurse, or a general duty medical officer who has neither the time nor the specialist training to separate a benign murmur from one that warrants a cardiologist's attention, or a coarse crackle from a wheeze that would change the referral pathway. CardioPulmo does not ask that expertise to sit at the point of recording. A phone already in the health worker's pocket becomes the capture device; the interpretation work moves to models running in the cloud; and what comes back is a structured screening signal the health worker can act on — refer onward, observe, or move to the next person in the queue — without having had to make that call unaided from a raw sound.

That shift matters most exactly where specialist listening is scarcest: rural screening camps, primary health centres with a single medical officer covering a wide catchment, and medical college outreach where student and intern hands are doing first-pass screening under supervision. It does not remove the clinician from the loop. It gives the person holding the phone a second, consistent input to weigh before deciding who is sent onward for clinical assessment.

Three cloud-hosted models, and why quality gating matters

Recordings captured on the phone are analysed by three cloud-hosted deep-learning models rather than a single classifier, so a heart sound and a lung sound are not forced through the same interpretive logic, and the screening read reflects more than one model's view of the recording. Because the whole approach depends on the audio itself being usable, CardioPulmo applies ambient-noise quality gating before a recording is treated as analysable. A camp hall, a crowded PHC corridor, a ceiling fan, or a rustling shirt can produce audio that no model should be asked to interpret with confidence. Gating exists to catch that at the point of capture, so the health worker is prompted to re-record in a quieter moment rather than receiving a screening read built on unusable input.

This is a quality control step, not a clinical one. It decides whether a recording is fit to be scored at all; it says nothing about what the resulting score means for the person being screened. Treating those two questions separately is deliberate: in field conditions, the most common failure mode is not a subtle sound being misread but a recording that never carried the sound clearly in the first place, and a screening workflow that cannot tell those apart quietly loses trust.

A screening aid, not a clinical conclusion

CardioPulmo is built and described throughout as a screening aid — decision support that flags recordings worth a closer listen, not a system that names disease. The output is triage support: it helps a health worker decide who should be sent for clinical evaluation and who can reasonably be reassured or reassessed later. The clinical assessment itself, and every conclusion that follows from it, stays with the treating clinician.

Nothing in the CardioPulmo output is intended to substitute for a clinician's examination. Clinoble makes no claim of regulatory clearance, third-party assessment, or independent performance evaluation for the screening read, and no performance figures are published here. In practice the workflow is designed so that a flagged or uncertain screening result routes the person to a doctor for a proper listen and, where warranted, further investigation. The phone shortens the queue to that doctor; it does not stand in for one.

Where it fits in a camp or PHC workflow

In a screening camp, CardioPulmo sits at the same station where blood pressure and basic vitals are already being taken. The health worker records a short chest sound with the phone, the recording passes through quality gating, and the screening read comes back in time to inform whether that person is directed to the camp's on-site doctor or advised to follow up locally. Because the capture step is short and needs no specialist listening skill, it can be folded into an existing camp station rather than requiring a new one.

In a primary health centre, the same flow slots into a routine consult — a screening capture ahead of, or alongside, the medical officer's own examination, giving a second input rather than displacing the officer's clinical judgment. For a medical college department, the same tool can sit inside outreach and community medicine postings, giving students and interns a structured way to practise auscultation screening under supervision, with every flagged case still reviewed by a qualified clinician before any action is taken.

What a deployment looks like

Because CardioPulmo runs as a progressive web app, a deployment does not require installing anything from an app store or provisioning dedicated hardware. A phone with a browser and a working connection back to the cloud-hosted models is the starting point, which keeps the barrier to trying it low for a district team or a camp organiser who cannot commit to new equipment.

For a district screening programme or a PHC network, the practical planning work is therefore not procurement but workflow: who records, where recordings happen, how consent and record-keeping are handled locally, and how flagged cases are routed onward to a treating clinician who can act on them. Those are the questions Clinoble expects to work through directly with district health teams, camp organisers, and medical college departments before any rollout, and they are also where a screening tool succeeds or fails in the field. MedTech teams looking at a phone-first, cloud-analysed approach for other health systems can raise the same questions through the contact route below.

The product itself, with its own more detailed technical description, lives at cardiopulmo.com.

Who this is for

District health officers planning phone-based heart-and-lung screening alongside existing camp and PHC screening activity.
Primary-care physicians who want a structured second input at the point of auscultation, without changing who makes the referral call.
Medical college departments running community medicine or outreach postings where students need a supervised, structured screening workflow.
Screening-camp organisers who need a fast, low-training capture step that does not depend on specialist auscultation skill being present on site.
MedTech teams abroad evaluating a phone-first, cloud-analysed heart-and-lung screening approach for other health systems.

Who builds it

CardioPulmo is built by Clinoble Innovations Private Limited, a company based in Hyderabad, Telangana, India. Its founder and director is Dr. Jaideep Rao M., MBBS, MD Community Medicine, and that background shapes the product: the design questions CardioPulmo answers are field-screening questions — who gets recorded, under what noise conditions, and how a flag becomes a referral — rather than laboratory ones. Clinoble also has two Indian patent applications pending in quantitative clinical imaging, IND 202641088451 and IND 202641090786.

CardioPulmo sits alongside Clinoble's other screening and care-delivery work, including TB F.I.R.S.T, a bilingual progressive web app covering the Find-Interpret-Refer-Support-Track tuberculosis cascade, and TeleVaidyam, a teleconsultation product for rural Telangana primary care. The common thread is a phone in a health worker's hand at the point of contact, with the heavier interpretation and record-keeping happening elsewhere.

Current status

CardioPulmo is live as a progressive web app at cardiopulmo.com, with recordings analysed by its three cloud-hosted models under ambient-noise quality gating. It is offered, and should be evaluated, strictly as a screening aid that supports — and never replaces — clinical judgment and the referral decisions clinicians make.