Let me start with a personal story. My mother has asthma. When I speak to her on the phone from abroad, I can immediately tell from her voice whether she’s having an asthma attack or not. I notice the difference between her regular tone and that of when she’s unwell. No matter how much mothers try to hide their sickness, their children can often detect it just by hearing their mother’s voice over the phone. Similarly, if you’re coughing or talking to someone over the phone, your loved ones can often sense that you’re not well simply from your voice. Not just your loved ones — researchers are now investigating whether artificial intelligence can analyze your voice to determine your health status.

The development of Artificial Intelligence (AI) technology is progressing rapidly, and now AI is capable not only of diagnosing diseases but also of detecting early symptoms through listening. Recently, Google and other tech companies have been developing new AI models based on bioacoustics, bringing revolutionary changes in disease detection.

Bioacoustics: Detecting Disease Through Sound

Bioacoustics is an interdisciplinary field where biology and acoustics come together. The main goal in this area is to analyze sounds produced by humans and animals to extract important health-related information. For example, common bodily sounds such as breathing, coughing, and sniffles often contain subtle cues that can help in identifying various diseases.

Google’s Health Acoustic Representation (HeAR) model is an example of this, capable of detecting early signs of illness from bodily sounds. This AI model will help at-risk populations in remote or rural areas easily diagnose diseases through mobile phones, where expensive traditional equipment and specialists are often unavailable.

AI in Tuberculosis Diagnosis

According to the World Health Organization (WHO), around 4,500 people die from tuberculosis (TB) every day, and about 30,000 new cases are reported daily. AI technologies are helping in the early detection of this disease, which is crucial for prevention. In Bangladesh alone, nearly 40,000 people die each year from TB. The HeAR model has been trained on a dataset of nearly 100 million cough sounds to identify early signs of tuberculosis.

Hyderabad-based startup Salcit Technologies in India has combined Google’s model with its own machine learning model called “Swaasa”. The “Swaasa” model analyzes a 10-second cough recording via a mobile application, enabling disease detection with about 94% accuracy. Anyone can take the test with their mobile phone for 200 rupees, while the traditional spirometry test costs around 3,000 rupees.

What is a spirometry test?
Spirometry is a basic test that can identify lung problems. It measures how much air you can expel in a single forceful breath. The test is carried out with a small machine called a spirometer, which has a mouthpiece attached by a tube. Other names for the spirometry test are pulmonary function test (PFT) and lung function test.

The Promise of AI in Healthcare

In addition to identifying deadly diseases like tuberculosis, AI technologies based on bioacoustics play a vital role in early detection of other complex diseases, including cancer. Google is currently researching an ultrasound-based model at Chang Gung Memorial Hospital in Taiwan for early breast cancer detection. If this model is successfully developed, it could create affordable and easily accessible breast cancer screening, especially for developing countries.

This technology isn’t yet fully commercialized, but with proper implementation, it could make disease detection more democratic and accessible in the future. Canadian company Ubenwa is developing an AI model to analyze infants’ cries for health-related information.

Below are some examples of startups using this technology:

• Swaasa, India: Swaasa is an AI-powered mobile application that analyzes a 10-second cough recording to help diagnose respiratory diseases such as TB. Developed by Salcit Technologies, it is the first medically-approved mobile device of its kind in India. With this app, you can simply diagnose respiratory diseases by recording your cough on your phone.
• Google HeAR, USA: Google’s HeAR (Health Acoustic Representations) model uses bioacoustics technology to analyze cough sounds for early signs of illnesses like TB. Although it’s not yet widely available, it is expected to become easier to access soon, especially for remote areas.
• ResApp, Australia: ResApp is an Australian app that analyzes cough sounds to detect respiratory diseases such as pneumonia, asthma, and COPD. Available for both iOS and Android, it is making AI-driven healthcare simpler and more effective.
• Ubenwa, Canada: The Ubenwa app analyzes babies’ cries to help identify health issues like respiratory distress at birth. While primarily designed for newborns, it demonstrates how bioacoustics can revolutionize healthcare.

Future Prospects

Bioacoustics and AI technologies like these offer many potential applications that could revolutionize the healthcare sector in the future. Here are some possible areas:

1. Diagnosis of Other Respiratory Diseases

By analyzing bodily sounds such as coughing, breathing, and sniffles, it is possible to diagnose respiratory diseases like pneumonia, chronic obstructive pulmonary disease (COPD), and asthma. This will greatly help assess patients’ physical conditions, especially where there is a shortage of healthcare professionals.

2. Sound Analysis for Children’s Health

Analyzing children’s cries and vocalizations can help identify early health issues such as neurological disorders, respiratory distress at birth, and others. This could open new avenues for monitoring newborns’ health.

3. Mental Health Monitoring

Subtle changes in voice, such as tremors or pitch variations, can be used to identify mental health issues like depression, anxiety, and stress. AI and bioacoustics may play a significant role in mental health analysis.

4. Detection of Heart Diseases

By analyzing specific internal body sounds, such as heartbeat and breathing patterns, it may be possible to identify early signs of cardiovascular diseases. This could make it much easier to predict heart attack risks in advance.

5. Monitoring Elderly Health

By analyzing the voice and breathing patterns of elderly people, it will be possible to monitor issues such as physical weakness, breathlessness, and other health concerns. This can assist in home-based elder care.

6. Detection of Autism and Other Neurodevelopmental Disorders

Analyzing children’s speech patterns and other vocalizations, such as cooing, gurgling, and babbling, could make it possible to identify early signs of autism and other neurodevelopmental disorders.

7. Hearing Problem Diagnosis

Sound analysis technology can be used to diagnose hearing problems in children and the elderly, helping to detect hearing issues at an early stage.

8. Physical Performance and Fatigue Assessment

By analyzing breathing patterns and sounds based on physical condition, it is possible to assess the physical performance and fatigue of workers in various workplaces.

9. Coronary Artery Disease Detection

With bioacoustic technology, it may be possible to detect early signs of coronary artery disease by analyzing specific heart-related sounds.

10. Multifaceted Disease Detection via AI

As these technologies advance, they could also be used for general fitness and health monitoring, providing daily health information by analyzing coughs, breathing, heartbeats, and other physiological activities.

The use of this technology is not limited to disease diagnosis — it can also be considered a vital tool for expanding the overall reach of healthcare.

Opportunities and Challenges in Technology

While this technology has opened new horizons for disease diagnosis, challenges remain. Many in rural areas are not accustomed to using technology, making it difficult for them to accurately record coughs. Additionally, relying too heavily on AI without adequate training can sometimes lead to misleading results. Nevertheless, doctors and technologists remain optimistic about the revolution AI is bringing to healthcare.

Conclusion

AI technologies based on bioacoustics are opening new frontiers in disease diagnosis. These sound and audio-based technologies not only detect early symptoms of diseases, but also bring affordable and accessible testing solutions.