In recent years, smartwatches and other wearables have steadily expanded their role in fitness and health monitoring. Initially designed for basic activity tracking, these devices have evolved into advanced tools capable of capturing a wide range of physiological signals. As technology continues to develop, wearables like smartwatches, earbuds, belts, and trackers are beginning to offer deeper insights into our health. They can now monitor heart rate, sleep quality, and physical activity, while also providing more specialized measurements such as atrial fibrillation detection, blood pressure monitoring, and even emerging metrics like advanced glycation end products (AGEs), which are linked to metabolic health and biological aging [1-2] (Figure 1). Although we are just beginning to unlock the full potential of these devices, they are paving the way for a more personalized approach to health and wellness, empowering users with data that can guide their fitness journeys and overall health management.
Figure 1. Samsung Galaxy Watch with Enhanced Bio-Active Sensor and AGEs Measurements Capabilities [1-2]
Building on the growing importance of wearable technology in health monitoring, the 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC 2024) in Orlando, Florida, provided the perfect platform to explore how these advancements are shaping the future of personalized health analytics. The event brought together experts and innovators from around the world, all focused on leveraging technology to improve healthcare access and outcomes [3].
Traditionally, assessing cardiovascular fitness has relied on metrics like resting heart rate and VO2 max, often requiring clinical settings or specialized equipment. While these metrics offer valuable insights, they are limited by their static nature and the conditions under which they are measured. With the increasing availability of wearable technology, there is a growing interest in more dynamic and real-time assessments that can be integrated into everyday life.
Smartwatches and other wearables, combined with modern signal processing and deep learning techniques, are at the forefront of this shift. Moving beyond static measurements, these devices provide continuous monitoring of physiological signals. Among these advancements, the ability to track heart rate recovery stands out as a critical development. Heart rate recovery, which measures how quickly the heart rate returns to normal after exercise, offers a real-time reflection of cardiovascular health and autonomic nervous system function. Generally, a quicker recovery is indicative of better fitness and a reduced risk of cardiovascular issues. The continuous monitoring capabilities of smartwatches make them especially well-suited to track heart rate recovery, allowing users to closely monitor this vital sign, understand their fitness levels more deeply, and make more informed decisions about their health and exercise routines (Figure 2).
Figure 2. General Concept of a Heart Rate Forecasting and Recommendation System During Workouts
In this context, the research initiated by the SRUKR Health Innovation team and presented at EMBC 2024 focused on leveraging smartwatch data to predict heart rate recovery during high-intensity workouts (Figure 3). By applying advanced deep learning and signal processing algorithms to the data collected from smartwatches, the team aims to provide users with personalized insights into their fitness levels and recovery patterns. This approach not only enhances the utility of wearable devices but also represents a step towards more individualized health and fitness monitoring.
Figure 3. EMBC 2024 Highlight – A glimpse of the conference atmosphere (left figure) [3] and SRUKR participant (right figure)
The study presented by the SRUKR Health Innovation team at EMBC 2024 leverages data from the Galaxy Watch, including accelerometer and photoplethysmography (PPG) signals, to offer real-time, personalized insights into cardiovascular fitness. By integrating advanced signal processing and deep learning techniques, the research provides more accurate predictions of heart rate recovery, helping users optimize their exercise routines and better monitor their overall health.
The primary goal of this research is to address the existing challenges in real-time monitoring and prediction of heart rate recovery—a crucial factor in preventing overexertion and improving exercise safety. The team developed a robust model based on an LSTM (Long Short-Term Memory) architecture, which is capable of forecasting heart rate dynamics during both the exertion (sprint) and recovery phases of high-intensity interval training (HIIT) (Figure 4). This approach not only tracks heart rate in real-time but also anticipates future trends, providing users with valuable feedback throughout their workouts.
Figure 4. System Architecture of the relaxation heart rate forecasting model: DR – dropout; LN – layer normalization
One of the notable achievements of this study is the creation of a specialized method that enhances the accuracy of the model by considering the patterns and trends in heart rate changes during exercise. This comprehensive approach ensures that the model captures the subtle variations in heart rate, leading to more precise predictions. The model's architecture, featuring bidirectional LSTM networks and sophisticated connections, is specifically designed to manage the complex timing and sequence of heart rate data, significantly improving accuracy, particularly during the recovery phase.
The findings also emphasize the benefits of personalized modeling in exercise physiology. The research demonstrates that models trained on data tailored to individual users provide more accurate predictions than those based on generalized data. This highlights the potential of wearable devices, combined with AI, to offer highly personalized health and fitness monitoring, making them invaluable tools for not only fitness enthusiasts but also individuals managing chronic health conditions or undergoing rehabilitation.
The research presented is a glimpse into the future of wearable technology in fitness and healthcare. As smartwatches and other devices continue to advance, their ability to deliver real-time, personalized health insights will significantly improve, offering not just fitness tracking but comprehensive health management. The integration of AI with wearable sensors is paving the way for early detection of health issues and tailored interventions, making these devices indispensable in our daily lives.
Stay tuned for what’s next—these advancements are set to change the way we live and care for ourselves.
[1] https://news.samsung.com/global/unlocking-new-possibilities-for-preventative-wellness-with-new-galaxy-watch-and-bioactive-sensor
[2] https://www.samsung.com/ae/watches/galaxy-watch/galaxy-watch-ultra-titanium-gray-lte-sm-l705fdaaxsg/
[3] https://embc.embs.org/2024/