Review highlights precision gains alongside validation and real-world deployment challenges
AI-powered wearable technologies are enabling more detailed monitoring of strength training and performance, according to a recent review published in Applied Sciences. The study evaluates research conducted between 2015 and 2025, examining the integration of artificial intelligence with wearable systems. It identifies both advancements in performance tracking and limitations related to validation and real-world application.
A review titled “Convergence of Artificial Intelligence and Wearables in Strength Training and Performance Monitoring: A Scoping Review,” published in Applied Sciences, analyzes the role of AI-driven wearable technologies in strength training environments. The study compiles findings from 13 studies conducted between 2015 and 2025 to assess current developments, constraints, and future directions.
The review notes that traditional strength training analysis has relied on laboratory-based systems such as force plates and motion capture technologies, which offer accuracy but are not widely accessible for routine use. In contrast, wearable sensors—including inertial measurement units, pressure insoles, and smart textiles—are increasingly used to collect biomechanical data during real-world training.
When combined with artificial intelligence models, these systems can classify movements, assess technique, and estimate internal loads with improving accuracy. Deep learning models, including convolutional and recurrent neural networks, are used to identify complex movement patterns. Applications such as squat classification and functional exercise recognition have demonstrated high accuracy through analysis of time-series sensor data. Hybrid models incorporating temporal and attention mechanisms further improve detection of variations in movement quality.
The study also highlights the ability of AI-based systems to estimate internal biomechanical forces, such as bone loading and ground reaction forces, by integrating sensor data with machine learning and biomechanical modeling. These estimations contribute to improved injury risk assessment and performance analysis.
Advancements in sensor design, including strain sensors integrated into garments and pressure-sensitive textiles, allow distributed monitoring across the body. However, the study indicates that performance can be affected by sensor placement, calibration, and environmental conditions, limiting reliability in practical applications. It also notes that many biomechanical studies are conducted in controlled environments with small sample sizes, restricting broader applicability.
In physiological monitoring, AI-enabled wearables support continuous tracking of metrics such as fatigue, energy expenditure, and cardiovascular responses. The review identifies a shift toward integrated systems that analyze multiple data inputs, including heart rate, motion signals, and biochemical indicators, to provide a comprehensive view of performance and recovery.
Machine learning models are used to estimate metabolic cost and exercise intensity without reliance on laboratory equipment. These models can also approximate post-exercise oxygen consumption, supporting analysis of training load and recovery. Fatigue detection is another application, where AI models identify early indicators based on time-series data, enabling adjustments to training intensity.
The integration of biochemical sensing technologies, such as sweat-based sensors, enables monitoring of hydration, electrolyte balance, and metabolic conditions. When combined with AI, these systems interpret biochemical signals in relation to performance metrics. However, the study states that many physiological models rely on indirect indicators rather than clinically validated measures, which may affect accuracy.
The review also examines predictive analytics applications tailored to specific sports. AI models are used to analyze biomechanical and physiological data, classify movements, and assess training loads across different disciplines. Techniques such as ensemble learning and deep neural networks identify relationships between training variables and performance outcomes.
Real-time feedback systems are also being developed, allowing continuous updates based on incoming data. These systems support decision-making by providing ongoing performance insights. However, the study notes that predictive models are often trained on limited datasets, which may affect their applicability across broader populations.
Several challenges are identified, including the lack of standardized datasets and validation protocols. Many studies rely on small, homogeneous samples and internal validation methods, which may overestimate performance. Additional concerns include the limited interpretability of deep learning models, sensor-related issues such as signal variability, and considerations related to data privacy and user adoption.
The study concludes that AI-enabled wearable technologies are transitioning from experimental systems to decision-support tools. It highlights the need for larger datasets, standardized validation approaches, and the integration of explainable AI methods. While applications such as exercise classification and basic monitoring are expected to expand in the near term, more complex functions such as injury prediction and automated training adjustments require further validation.