“AI has been instrumental in accelerating drug discovery, reducing the time it takes to bring new treatments from the lab to the patient. By analyzing vast datasets, AI can identify potential drug candidates that would have taken years to find using traditional methods.” – Dr. Eric Topol, Author of Deep Medicine
“AI has the potential to cure diseases that have long been deemed incurable by uncovering hidden patterns in data that were previously invisible to human researchers.” – Dr. Andrew Ng, Co-founder of Google Brain
A persistent cough often signals that something might be wrong with your health, but just how much can be determined from the sound of a cough? A recent report highlighted by Google Research reveals that a team of researchers in India is harnessing an AI model developed by Google to “listen” for specific respiratory diseases, including tuberculosis (TB) and chronic obstructive pulmonary disease (COPD).
Seeing the unseen with AI-powered health monitoring
A short video on how powerful acoustic AI models transform simple bodily sounds, such as cough, into a symphony of rich and actionable data for health and wellness breakthroughs. pic.twitter.com/EWXqtcifyh
— Mindaugas Galvosas, MD (@MGalvosas) January 19, 2024
Google introduced its bioacoustic foundation model, Health Acoustic Representations (HeAR), in March, aiming to advance medical audio analysis.
The Google Research team trained HeAR using an extensive dataset comprising 300 million audio samples. Specifically, for detecting cough-related diseases, the model was trained with approximately 100 million cough sounds. This model is designed to analyze how people speak—including elements like tone, pitch, and pace—to identify potential health issues, such as dementia, through AI and machine learning.
“With the help of AI, we are now able to predict and prevent chronic conditions such as diabetes and heart disease before they manifest, paving the way for a new era of personalized medicine.” – Dr. Regina Barzilay, Professor of AI and Health at MIT
“HeAR is adept at recognizing patterns within health-related sounds, making it a powerful tool for medical audio analysis,” explained Shravya Shetty, Google Research Director of Engineering. “Our findings show that, on average, HeAR outperforms other models across various tasks and demonstrates an impressive ability to generalize across different microphones, underscoring its capacity to capture significant patterns in health-related acoustic data.”
In practice, Salcit Technologies, a respiratory healthcare company based in India, employs HeAR as part of its Swaasa app, which analyzes cough sounds to detect tuberculosis.
“TB is a treatable disease, yet millions of cases go undiagnosed each year, often due to a lack of convenient access to healthcare services,” Shetty noted. “Enhancing diagnostic methods is crucial to eradicating TB, and AI can play a vital role in improving detection, making healthcare more accessible and affordable worldwide.”
Google AI Can Hear Disease in the Sound of Your Cough
Google’s Health Acoustic Representations (HeAR) model helps researchers in India detect life-threatening diseases through audio analysis. pic.twitter.com/iFeVG5VJb4
— ᴄʀʏᴩᴛᴏ ɴᴇᴡꜱ (@RUPESHK64053040) August 20, 2024
Zhi Zhen Qin, a digital health specialist with the Stop TB Partnership, emphasized the potential impact of HeAR, stating, “Solutions like HeAR could revolutionize tuberculosis screening and detection, providing an accessible, low-impact tool for those who need it most.”
While much of the conversation around artificial intelligence has been dominated by concerns over AI-generated deepfakes, the potential of AI to enable early detection of life-threatening diseases continues to drive research in this field.
For example, in June, researchers at the University of Cambridge in the U.K. developed an AI model called EMethylNET. This model shows promise in early cancer detection, diagnosis, and treatment planning, particularly during critical early stages when intervention can be most effective.
Dr. Gabriel Zada, a neurosurgery specialist at Keck Medicine of USC, previously highlighted the importance of knowing the tumor type and subtype during surgery to make informed decisions, such as whether to fully remove a tumor or just biopsy it.
Another company leveraging AI for early cancer detection is New York-based medtech firm Ezra, which uses advanced imaging and AI to scan critical areas of the body, including the brain, lungs, liver, and prostate.
Here is a demonstration of how https://t.co/FkBVrAIzM4 works.
You cough into the phone and you will get an analysis of what kind of respiratory condition you might have. These are some of the health parameters that, I think, patients will meet first on their #AI journey. pic.twitter.com/CE02GcqZWY
— Berci Meskó, MD, PhD (@Berci) August 16, 2024
Dr. Daniel Sodickson, a medical advisor for Ezra, emphasized the importance of early detection, noting, “It makes a huge difference for every patient and also for the medical system. Five-year survival rates for cancer can jump from 20% to 80% if caught early, potentially saving millions of lives.”
Quotes
- “In the near future, AI could unlock cures for complex diseases like Alzheimer’s by modeling the interactions of proteins and other biological molecules at an unprecedented scale.” – Demis Hassabis, Co-founder and CEO of DeepMind
- “AI’s ability to simulate millions of potential drug interactions in silico is bringing us closer to finding cures for rare and complex diseases that have eluded us for decades.” – Daphne Koller, Co-founder of Coursera and CEO of Insitro
- “We are entering an era where AI-powered precision medicine could potentially eradicate cancer by tailoring treatments to the unique genetic makeup of each patient.” – Dr. Siddhartha Mukherjee, Oncologist and Author of The Emperor of All Maladies
- “The future of AI in healthcare is not just about curing diseases but about fundamentally changing how we approach health, shifting from treatment to prevention through early detection and personalized care.” – Dr. Atul Butte, Director of the Institute for Computational Health Sciences at UCSF
Major Points:
- Researchers in India are using Google’s AI model, HeAR, to analyze cough sounds and detect respiratory diseases like tuberculosis (TB) and COPD.
- Launched in March, Google’s Health Acoustic Representations (HeAR) was trained on 300 million audio samples, including 100 million cough sounds, to identify health issues through sound analysis.
- The HeAR model is used by Salcit Technologies’ Swaasa app to improve TB detection, potentially making healthcare more accessible and affordable globally.
- AI models like EMethylNET, developed by the University of Cambridge, show promise in early cancer detection and treatment planning, emphasizing the importance of identifying tumors during surgery.
- Early detection, aided by AI, significantly increases survival rates for diseases like cancer, with potential to save millions of lives.
Conner T – Reprinted with permission of Whatfinger News
