Nivedit Majumdar Nivedit Majumdar

The Future of Heart Rate Tracking with AI

Reinventing the wheel isn’t always easy. The sentiment applies more so in the space of sensor tech, wherein the only advancements that can be made are more often in the efficiency of measurement, footprint reduction and more sensitive recording parameters.

But then, there’s also the concept of reinventing the applications of the wheel. And that’s what advancements in sensor technology are all about. Heart rate sensors have moved on from their avatar of simply measuring heart rate, to more advanced applications such as pointing out exactly when and why there was a spike in the heart rate.

All this being said, what’s new in the world of heart rate sensors? And is Apple making some headway in this regard?


Before I actually discuss how heart rate sensors are getting better, it would be prudent to see the current state of affairs with the heart rate sensors most commonly found in wearables – such as smartwatches and fitness trackers.

As mentioned in this article on heart rate tracking, the Apple Watch makes use of green LEDs that are paired with light sensitive photodiodes, which detect the amount of blood flowing through the user’s wrist at any time. But more significantly, the process that is used to measure the heart rate parameters is the same for most wearables out there – and it’s called photoplethysmography (PPG).

Image Source: Cardiogram

According to an informative post on Cardiogram, the disadvantage of this technique is that PPG – unlike more advanced recording techniques such as electrocardiogram – measures only pulse waves. P waves or the intervals between heart contractions (PR, QS and QT) are not measured. And this omission can actually lead to a reduced efficiency in detection of diseases such as atrial fibrillation, or advanced forms of arrhythmia.


I’ll keep bringing up Cardiogram again and again, since they’re really showing great potential as far as gleaning information and insight from heart rate is concerned.

Cardiogram believes that the answer lies in involving a neural network that can be trained according to the heart rate data of millions of Apple Watch users. This has already been implemented by the company, and Cardiogram claims that as far as detecting specialised conditions such as atrial fibrillation was concerned, an accuracy of 97% was achieved.

This research was conducted with the involvement of UC San Francisco’s Health eHeart study, and is not dissimilar to similar programs by the Google Brain division, which is incorporating the usage of artificial intelligence to detect disorders.


Heart rate sensing has always been at the helm of affairs as far as the Apple Watch was concerned, and this statement has been given more credibility with the launch of watchOS 4, and the new Apple Watch as well.

The new version comes with inbuilt cellular connectivity, which would imply that the heart rate can be recorded and stored anywhere, any time, without the need of a separate phone. More importantly, the resting heart rate and recovery heart rate will be indicated right on the home display.

One thing that I personally liked about the new Apple Watch ecosystem was the feature of tracking the heart’s recovery after a workout, during the cool down period. Researchers have observed that a prolonged duration for the heart rate to recover back to normal might indicate that the heart rate conditions aren’t ideal, thereby implying the need for better heart conditioning.


With more and more research programs being roped in by Apple, the next improvements in the Watch’s heart rate sensor will come in the form of distinguishing between irregular rhythms that are genuinely concerning, from the patterns which won’t be a cause of worry.

According to CNBC, Apple is partnering with a group of clinicians at Stanford, as well as telemedicine vendor American Well, to test the Apple Watch’s heart rate sensor’s capabilities in detecting abnormal heart rhythms. As I’ve already mentioned earlier, the readings from the Apple Watch are also being used by Cardiogram for their studies.

The bottom line is this: Data, specifically that of heart rate, will now become more comprehensive and insightful.


So this is what I think will be the future of heart rate tracking: sensors might not diversify or evolve as much, but they will be paired with artificial intelligence to seamlessly detect which patterns are clear warning signs.

In addition to this, I think heart rate data can now be more than just ‘stand alone’ observations, and will incorporate other factors in as well. For instance, location, time, and perhaps even exercise data. All of this will lead to more holistic insight as far as actually using the heart rate is concerned.

Apple and Cardiogram are just getting started, and it will be interesting to follow their developments over the next few months.