Google’s HeAR Model Now Available via Cloud API for Health Research

Google’s HeAR Model Now Available via Cloud API for Health Research

Google has announced that its HeAR model (Health Acoustic Representations), a pioneering AI system for analyzing health-related sounds, is now available to researchers through the Google Cloud API. This is a significant step forward in using AI to potentially revolutionize how various health conditions are screened, diagnosed, and monitored.

HeAR, initially introduced in a paper earlier this year, is designed to extract valuable health insights from acoustic data such as coughs, breaths, and other bioacoustic sounds that may indicate the presence of diseases such as tuberculosis (TB) or chronic obstructive pulmonary disease (COPD). The model was trained on a massive dataset of 313 million two-second audio clips, allowing it to discern intricate patterns within health-related sounds.

Researchers have found that HeAR outperforms other models across a wide range of tasks, demonstrating superior ability to capture meaningful patterns in health-related acoustic data. Importantly, models trained using HeAR achieved high performance with less training data, a crucial advantage in healthcare research where data scarcity is often a challenge.

The potential applications of HeAR are vast. For instance, Salcit Technologies, an India-based respiratory healthcare company, is exploring how HeAR can enhance their existing AI model, Swaasa, for early detection of tuberculosis based on cough sounds. This could be particularly impactful in regions where access to healthcare services is limited.

Sujay Kakarmath, a product manager at Google Research working on HeAR, emphasized the model's potential: "Every missed case of tuberculosis is a tragedy; every late diagnosis, a heartbreak. Acoustic biomarkers offer the potential to rewrite this narrative."

The StopTB Partnership, a United Nations-hosted organization, has also expressed support for this approach. Zhi Zhen Qin, a digital health specialist with the organization, stated that solutions like HeAR could "break new ground in tuberculosis screening and detection, offering a potentially low-impact, accessible tool to those who need it most."

HeAR’s potential extends beyond TB. With its ability to generalize across various microphones and environments, this model could enable low-cost, accessible screening for a range of respiratory conditions, marking a significant step forward in acoustic health research. Google’s goal is to make this technology widely available, supporting the global healthcare community in developing innovative solutions that break down barriers to early diagnosis and care.

It's important to note that HeAR itself is not a diagnostic tool. Rather, it's a neural network that outputs low-dimensional embeddings optimized for capturing the most salient parts of health-related sounds. Researchers can use these embeddings to build and improve their own models for specific health conditions.

Researchers interested in exploring HeAR can follow these instructions to request access to the API through Google Cloud.

Chris McKay is the founder and chief editor of Maginative. His thought leadership in AI literacy and strategic AI adoption has been recognized by top academic institutions, media, and global brands.

Let’s stay in touch. Get the latest AI news from Maginative in your inbox.

Subscribe