AI-Powered Apps Analyze Cough Sounds to Screen for Respiratory Diseases
AI-powered apps that analyze cough sounds are being developed as screening tools for respiratory diseases such as tuberculosis (TB). While not as accurate as traditional diagnostics, these apps can help identify active cases of TB and aid in early treatment, according to a story on technologyreview.com. The use of AI in analyzing cough sounds has gained traction in recent years, with the COVID-19 pandemic further driving interest in this field. Cough-tracking apps could also be valuable in clinical trials and the development of tools to diagnose various health conditions.
In a study conducted by researchers from the U.S. and Kenya, a smartphone-based diagnostic tool was trained and tested on recordings of coughs collected from a Kenyan healthcare center. The app’s performance was not sufficient to replace traditional diagnostics, but it could serve as an additional screening tool. Early identification and treatment of active TB cases can help prevent the spread of the disease.
The concept of using coughs and other body sounds as “acoustic biomarkers” to indicate changes in health has been around for several decades, but the field has seen significant growth in the past five years, fueled by the increasing use of AI. The COVID-19 pandemic has also led to the emergence of numerous startups focusing on the acoustics of cough. These startups are developing apps to diagnose COVID-19, influenza and TB based on the sound of a patient’s cough.
While some experts remain skeptical about the reliability of these apps as diagnostics, they acknowledge the potential value of apps that can detect coughs in general. Coughs are easily captured by smartphones and can serve as valuable health tools. Cough-tracking apps could also revolutionize clinical trials, where coughs are often difficult to track accurately. In addition to coughs, other sounds such as voice and respiratory sounds are being explored as clues for diagnosing various health conditions.
Researchers are developing a publicly available database of these sounds to aid in the development of diagnostic tools.
Although there are limitations to the current studies due to the use of private data sets, the increasing availability of data will likely lead to the development of more apps that can detect health problems based on cough or speech patterns. The impact of these apps on diagnosis and screening is yet to be determined, but ongoing developments in this field are worth monitoring.
Some of the apps, developed during the height of the Covid epidemic, include AudibleHealthAI (launched in 2020), and ResApp Health of Australia.
read more at technologyreview.com