Apple wants to turn AirPods into heart rate monitors, will use AI for it

India Today

30-05-2025

Apple is exploring how its wearable devices like AirPods can double as heart rate monitors. A newly published research paper from the Apple Research team discusses the possibility of using AI-powered acoustic models to estimate heart rate from heart sound recordings. These heart rate recordings or sounds can be captured from the body using devices like AirPods. The study, titled 'Foundation Model Hidden Representations for Heart Rate Estimation from Auscultation', seeks to identify if foundation AI models trained on general audio and speech can accurately estimate heart rate from heart sounds. This non-invasive method, known as auscultation, typically involves listening to sounds produced by the heart. The idea is similar to how doctors use a stethoscope to listen to heart rate to diagnose and monitor various medical conditions. The researchers at Apple want to follow the same technique and use wearables like AirPods to capture heart sounds and use AI to measure heart rate by analysing it. Apple revealed that its researchers tested six major foundation models, including HuBERT, wav2vec2, and its own internally developed version of CLAP (Contrastive Language-Audio Pretraining), to see how well these models could pick up heartbeats from phonocardiograms — recordings of heart sounds. The research shows that even though these models weren't built for healthcare tasks, they managed to outperform traditional methods based on handcrafted audio features. 'In this work, using a publicly available phonocardiogram (PCG) dataset and a heart rate (HR) estimation model, we conduct a layer-wise investigation of six acoustic representation FMs: HuBERT, wav2vec2, wavLM, Whisper, Contrastive Language-Audio Pretraining (CLAP), and an in-house CLAP model,' Apple said in the paper published online. During the research, the Apple team used a publicly available dataset of over 20 hours of hospital-recorded heart sounds, annotated by medical experts. The team then split the audio clips into 5-second segments, with the AI analysing them to predict heart rate in beats per minute (BPM). The study revealed that mid-level layers in AI models worked best for detecting heart signals, while deeper layers—which are usually fine-tuned for speech recognition—were less effective at analysing biological sounds like heartbeats. This suggests that Apple would need to focus on specific parts of AI models rather than using them as-is for health tracking. Although the research did not reveal any plans for a commercial product, it does suggest Apple's intent to do more with its devices. Apple has already previewed the expanding possibilities of earbuds like the Beats Powerbeats Pro 2, which can offer heart rate tracking. But with AI, Apple wants its wearables to do even more. AirPods already feature high-quality microphones used for Active Noise Cancellation (ANC) and Transparency Mode, which could theoretically pick up subtle heart sounds. If Apple manages to integrate this AI-powered heart rate detection into AirPods, it could help users with passive heart rate monitoring without needing an Apple Watch and offer more advanced fitness tracking and early detection of irregularities with the heart.
advertisement