Can a Smartphone Hear Tuberculosis? The Promise and Pitfalls of Digital Cough Monitoring

By Wynter Cheng

---

Introduction

It’s 2025, an age of AI doctors, smartwatches that read your pulse, and cars that drive themselves. Yet tuberculosis (TB), an ancient disease, still kills over a million people each year. Caused by the bacterium Mycobacterium tuberculosis, TB primarily attacks the lungs, spreading through the air when an infected person coughs or sneezes. Once inside the body, the bacteria can slowly destroy lung tissue, leading to persistent coughing, chest pain, fever, and weight loss. Without timely treatment, TB can be deadly. And despite the technological advances, TB often  goes unnoticed until it’s too late. Why? Traditional screening still relies on sputum tests and X-rays, tools far out of reach for communities in rural areas. For example, in England, 45% of TB cases living in rural areas did not complete treatment compared with 26% of cases in urban areas.

But what if, your smartphone  could listen to your cough and warn you of TB?

From Lab Tests to Pocket Diagnostics

It might be possible. Researchers are developing smartphone-based cough monitoring systems that analyze audio data to detect cough patterns linked to TB. These systems rely on digital biomarkers (quantifiable signals captured from digital devices) and acoustic datasets trained on thousands of recorded coughs. Ultimately, they are aimed to detect “TB coughs”.  The result could be a simple, affordable, and widely accessible way to screen for TB outside traditional clinics.

Your Phone’s New Job Title! Cough Detective?

A 2025 study in medRxiv, “Evaluating Smartphone-Based Cough Monitoring for TB Screening and Triage in Uganda,” tested this concept in real-world clinics. Using the Hyfe Cough Tracker app, researchers combined smartphone microphones and machine learning to continuously record and classify coughs. They found that while the tool increased accessibility and patient engagement, its diagnostic accuracy in sensitivity and specificity remains limited, highlighting the need for larger datasets and regional calibration. Complementary work by Carter et al. (2023) in PLOS Digital Health, “Continuous Cough Monitoring: A Novel Digital Biomarker for TB Diagnosis and Treatment Response Monitoring”, demonstrated that tracking cough frequency over time could predict treatment response and disease progression. Rather than replacing lab tests, digital cough monitoring might serve as a 24/7 health sidekick, quietly keeping tabs on recovery and helping clinicians stay one cough ahead.

To build stronger models, researchers have released large-scale acoustic datasets. The Nature Scientific Data paper, “A Dataset of Solicited Cough Sounds for Tuberculosis Triage Testing” (2024), compiled thousands of labeled coughs across populations, providing the foundation for training and validating AI-driven cough classifiers. Unfortunately, when cross-compared with Hyfe’s performance evaluation on F1000Research (2022), results suggest promise but uneven generalizability: apps may perform well in one setting but falter elsewhere due to regional cough variability or background noise.

Coughing Up the Future: Promise, Pitfalls, and What’s Next

Smartphone cough monitoring offers a low-cost, non-invasive, and scalable approach to TB detection. Just a phone and internet connection, and you could democratize health diagnostics in under-resourced regions. Yet, limitations persist. For instance, AI models often struggle with low specificity: they can mistake coughs from other respiratory diseases like pneumonia or COVID-19 for TB, leading to false positives. Additionally, fragmented datasets where data are collected in different countries using varying recording methods and standards, making it difficult to train consistent, reliable algorithms that work across populations.

Still, the research community is steadily bridging these gaps by building shared databases and refining models to better distinguish TB cough patterns. Though this novel sidekick has yet to match the credibility of the lab, it can vastly extend its reach—turning every pocket into a potential public health sentinel.

---

References 

1. WORLD. TERBERCULOSIS. Who.int. Published March 14, 2025. Accessed November 13, 2025. https://www.who.int/news-room/fact-sheets/detail/tuberculosis

2. ABUBAKAR I, CROFTS J P, GELB D, STORY A, ANDREWS N, WATSON J M. Investigating urban–rural disparities in tuberculosis treatment outcome in England and Wales. Epidemiology and Infection. 2007;136(1):122–127. doi:https://doi.org/10.1017/s0950268807008333

3. HUDDART S, ASEGE L, JAGANATH D, et al. Continuous cough monitoring: a novel digital biomarker for TB diagnosis and treatment response monitoring. The International Journal of Tuberculosis and Lung Disease. 2023;27(3):221–222. doi:https://doi.org/10.5588/ijtld.22.0511

4. CHETRY B, GOSWAMI S, DUTTA C, GOGOI A, SAIKIA J P, NATH P. Ultra-sensitive detection of Mycobacterium cells on a smartphone through enhanced emission of autofluorescence signals. Biosensors and Bioelectronics. 2025;288:117821. doi:https://doi.org/10.1016/j.bios.2025.117821

5. HUDDART S, YADAV V, SIEBERTS S K, et al. A dataset of solicited cough sound for tuberculosis triage testing. Scientific Data. 2024;11(1). doi:https://doi.org/10.1038/s41597-024-03972-z 

---

This post is not a substitute for professional advice. If you believe that you may be experiencing a medical emergency, please contact your primary care physician, or go to the nearest Emergency Room. Results from ongoing research is constantly evolving. This post contains information that was last updated on December 2025.

---