Detecting Respiratory Disease from Breathing Sounds

AI-powered, smartphone-based diagnostics transforming respiratory health screening worldwide

256K+ PCR-Tested Patient Recordings
84% Sensitivity/Specificity
0.93 AUC Score

About Us

Co-founded by University of Washington audio AI signal processing Professor Les Atlas, Audio AI Systems Inc. is pioneering breakthrough AI algorithms that identify various respiratory diseases through smartphone-recorded forced coughing sounds. We're building the future of diagnostics and disease surveillance with patented technology jointly owned with the University of Washington.

Our mission is to diagnose diseases like COVID-19, TB, flu, RSV, COPD, asthma, and lung cancer from the sound of coughs and breathing patterns. We're also developing a respiratory disease spread forecast system using cough surveillance testing to rapidly track and predict the emergence of both known and unknown pathogens.

MIT Sandbox Innovation Fund Program AI Technology

Technology

Patented AI Innovation

Our Complex Clipping technique (US 2024/0281639 A1), co-developed with University of Washington, enables high-accuracy disease detection even with noisy real-world data. This breakthrough allows us to rapidly train models for multiple diseases from a single data capture point.

Instant Results

Transform diagnostic timelines from days to seconds. Our smartphone-based solution delivers immediate, preliminary assessments before invasive and costly lab tests, making respiratory health screening accessible to anyone, anywhere.

Scalable Detection

Our AI platform can identify indicators for multiple respiratory diseases including COVID-19, tuberculosis, influenza, RSV, COPD, asthma, and early-stage lung cancer — all from simple audio recordings of coughs and breathing.

Disease Surveillance

Beyond individual diagnosis, we're building population-level respiratory disease forecasting systems. By analyzing cough patterns across communities, we can detect and predict outbreaks of both known and emerging pathogens in real-time.

Edge Computing

Leverage advanced edge computing and smartphone processing to deliver AI-powered diagnostics without requiring cloud connectivity. Privacy-first architecture ensures patient data security while maintaining high performance.

Massive Dataset

Our algorithms are trained on 256,600+ PCR-tested patient recordings, providing robust performance across diverse populations and acoustic environments. Continuous learning from real-world data ensures ongoing improvement.

Team

Prof. Les Atlas

Co-Founder / Chief AI Innovation Scientist

PhD in Electrical Engineering. Professor of Audio Signal Processing and AI at University of Washington. Author of pioneering work in CNN AI. Leading R&D of breakthrough audio analysis algorithms for medical diagnostics.

Amil Khanzada

Co-Founder / CEO

UC Berkeley Computer Science (top 5%). Former MBA/AI student at Berkeley. One Young World delegate. Leads partnerships, product strategy, market research, regulatory pathways, and funding initiatives.