The early detection of respiratory illnesses like COVID-19 remains a challenge, especially when relying on symptoms like coughing, which can stem from various causes. Without accessible screening tools outside clinical settings, individuals may delay testing or isolation, inadvertently increasing community transmission. A tool that provides preliminary cough-based screening could help bridge this gap.

How It Works: Layering Cough Analysis with User Data

One approach could involve an app that records and analyzes cough sounds using machine learning to estimate the likelihood of COVID-19. Users would record cough samples, and the app would combine this data with optional symptom inputs (e.g., fever, exposure history) for a more accurate risk assessment. Results could prompt further action—like seeking a lab test—while integrating telehealth services for convenience. Over time, anonymized data from users could refine the model, improving accuracy through real-world validation.

Stakeholders and Incentives

The tool could serve multiple groups:

• Individuals looking for quick, non-invasive screening before committing to testing.
• Employers and schools needing scalable health checks for groups.
• Public health researchers interested in tracking cough patterns to monitor outbreaks.

Possible revenue streams include premium features (e.g., telehealth links) or partnerships with health agencies seeking crowd-sourced data. Privacy would be maintained by processing audio locally and anonymizing shared datasets.

Validation and Iterative Improvement

Initial versions might rely on pre-trained models from research (with ~80–90% accuracy in controlled settings), but real-world performance would require validation against PCR tests. Partnering with clinics could help refine the model. To avoid over-reliance, the tool would clearly state it’s not a diagnostic, instead emphasizing follow-up testing when symptoms or risk factors are present.

Compared to existing tools like symptom checkers or cough-tracking apps, this approach focuses on acute screening rather than long-term monitoring, balancing accessibility with actionable insights. An MVP could start with basic cough analysis, then expand with features like exposure tracking or testing location guides.