A more productive primary care physician (PCP): auscultate with confidence and screen more patients in less time, with rapid, accurate, non-invasive respiratory screening at the point of care.
Our innovative fusion of MedTech, AI, and clinical insights amplifies these often-overlooked sounds, enabling rapid and accurate identification of a wide range of respiratory conditions, in the hands of any PCP.
DxAssist is our product, built on acoustic Deep Learning models. A digital stethoscope and acoustic deep learning give the clinician the ability to screen for lower respiratory conditions in less than 4 minutes. The workflow is three steps: auscultate, analyze, act.
A clinical aide places the digital stethoscope at four standardized posterior sites. No auscultation training required.
Deep-learning models screen each recording for adventitious lung sounds, with a real-time capture-quality check on every take.
A calibrated result with a confidence score supports the clinician's decision at the point of care.
The DxAssist software guides a four-location auscultation and checks capture quality in real time, following the QSP protocol.
We pair the nuance of a seasoned pulmonologist's ear with acoustic deep-learning models, in the familiar form of a digital stethoscope, to surface what routine auscultation can miss.
Rooted in a decade of acoustic-AI experience, now focused on the sounds of the human body.
Built with practicing pulmonologists so the model reflects real auscultatory judgment.
Runs on a digital stethoscope and an edge device, no imaging or lab required.
We completed a multi-country prospective study across the Philippines, Vietnam, Uganda, and South Africa, evaluating our acoustic-AI models for screening tuberculosis from lung-sound recordings; the work is accepted for presentation at TBScience 2026 in Rio de Janeiro.
DxAssist is an investigational device. It has not been cleared or approved by the U.S. Food and Drug Administration and is not available for sale or for diagnostic use. Findings described here are from research evaluations, not a regulatory determination of safety or effectiveness.
Four countries, through the UCSF-led Rapid Research in Diagnostics Development (R2D2) TB Network.
Accepted for a moderated e-poster presentation at the Union World Conference on Lung Health.
20+ years leading Silicon Valley high-tech startups; international operations and business models. BS and MS in Electrical Engineering, Stanford.
LinkedIn →
Algorithm design and machine learning; medical systems at Roche Diagnostics; ex-McKinsey. MBA, UC Berkeley; MS Computer Engineering, EPFL Lausanne.
LinkedIn →
Entrepreneur in growth and strategy; led the international expansion of a diagnostic-lab business; ex-Kearney. MBA, Stanford; Industrial Engineering, PUCP.
LinkedIn →
Pulmonary disease specialist with 35+ years of practice, the "trained ear" behind the models. Albany Medical College, with honors.
Our collaboration with the UCSF-led R2D2 TB Network on digital lung-auscultation for tuberculosis screening has been accepted for a moderated e-poster presentation at the Union World Conference on Lung Health.
Read the release →Partner with us on clinical pilots, investment, or bringing acoustic AI to your patient population.
Audium Health, a company revolutionizing disease screening with acoustic AI, today announced that its collaborative research with the University of California, San Francisco (UCSF) and the Rapid Research in Diagnostics Development (R2D2) TB Network has been accepted for a moderated e-poster presentation at TBScience 2026, the scientific program of the Union World Conference on Lung Health, taking place in Rio de Janeiro, Brazil, from November 17 to 20, 2026.
The accepted abstract, "Digital lung auscultation for TB screening: Development and validation of deep learning models," presents deep-learning models that screen for tuberculosis from digital stethoscope lung-sound recordings. In a multi-country prospective study across the Philippines, Vietnam, Uganda, and South Africa through the R2D2 TB Network, the models achieved an area under the ROC curve of 0.76, with 75 percent sensitivity and 77 percent specificity, using only a low-cost digital stethoscope.
In acoustic AI, results improve as the volume and diversity of high-quality labeled recordings grow. Additional sites, and patients recorded under the same procedure, make the models more accurate and more broadly applicable.
We are fortunate to be working with UCSF and the R2D2 TB Network, and with their extensive TB knowledge and their efforts in recording and labeling lung auscultations across a variety of countries and demographics. With this data, we showed the feasibility of successful TB screening performance. Abel Villca, Co-founder, Audium Health, Inc.
The R2D2 TB Network exists to test promising diagnostics rigorously, against a microbiological reference standard, in the countries where TB actually occurs. Evaluating digital auscultation prospectively across four countries tells us the signal is real outside any single site, and we are continuing to expand this evaluation because that is how the field learns whether a tool holds up in routine care. PROPOSED DRAFT WORDING ONLY. Not written, reviewed, seen or approved by Dr Jaganath. Placeholder for a quote to be requested from Devan Jaganath, MD, MPH, Associate Professor of Pediatrics, University of California, San Francisco.
UCSF is extending the work across additional R2D2 sites, treating point-of-care TB screening as an ongoing program. With this additional data, Audium Health expects to realize a shared goal of bringing accessible TB screening to high-burden, resource-limited settings.
About Audium Health. Audium Health (https://audiumhealth.com) is a leader in the realization of acoustic deep learning models for in-seconds, at the point of care, clinical screening. Audium's DxAssist auscultation deep learning model is targeting cardiopulmonary screening for 100% of patients while improving a primary care physician's productivity by 5X and saving 0.5x annually for each primary care physician. Audium's Large Auscultation Model Partnership, LAMP is a collaborative effort in the screening for the many cardiopulmonary morbidities.
About the R2D2 TB Network. The Rapid Research in Diagnostics Development for TB Network (R2D2 TB Network) is an initiative supported by the National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health (NIH) that brings together experts in TB care, technology assessment, diagnostics development, laboratory medicine, epidemiology, health economics and mathematical modeling with highly experienced clinical study sites. Together, we seek to move the field forward by providing a transparent and partner-engaged process for the identification, evaluation, and advancement of the most promising TB diagnostics.
Media contact. Abel Villca, Co-founder, Audium Health. Media enquiries via the contact form.