Fitness development and performance assessment of elite athletes requires an understanding of many physiological factors, many of these are direct and indirect measures of athlete energy expenditure. Many methods are physiological factor assessments and require the athlete to be constrained by laboratory equipment or periodic interruption of activity to take measurements such as blood samples are required to be taken. This paper presents a method that is entirely ambulatory and noninvasive, using microelectromechanical systems (MEMS) accelerometers. The commonly used output of commercial accelerometer-based devices (known as "counts") cannot discriminate activity intensity for the activities of interest. This, in conjunction with variability in output from different systems and lack of commonality across manufacturers, limits the usefulness of commercial devices. This paper identifies anthropometric and kinematic sources of inter-athlete variability in accelerometer output, leading to an alternate energy expenditure estimator based mainly on step frequency modified by anthropometric measures. This energy expenditure estimator is more robust and not influenced by many sources of variability that affect the currently used estimator. In this system, low-power signal processing was implemented to extract both the energy estimator and other information of physiological and statistical interest

This paper describes the development of a complete wearable swimming system for performance analysis, together with a sample application. The system comprises wearable nodes, data processing tools in MATLAB™ and integration with video. The swimming nodes are small in size and designed to be worn on body segments of interest, typically lower leg, lower arm, and the sacral or cervical regions. Each node contains inertial sensors, screen, data storage and RF communications for synchronisation and data download. The device is controlled using a microcontroller with a scheduler-based operating system to conserve power and is custom-packaged with a user interface and USB port that is fully waterproof. The cost of manufacture is a few hundred dollars in small-run quantities. The developed analysis software builds upon previously developed tools, can communicate with the nodes individually and can synchronise the recording of multiple units through a custom-developed protocol. Video is integrated into the developed tools as a method of presenting the sensor data alongside a more traditional analysis tool. A case study of the system analyses swim stroke phase with video and demonstrates the utility of the system as a tool for temporal stroke phase identification in the high-performance environment.

A sensor data acquisition platform has been developed for in-situ sporting applications encompassing stand alone high speed sampling and storage of multiple accelerometer data. Ambulatory monitoring of elite athlete in competi- tion or training environments was then undertaken. The platform itself is divided into functional blocks and con- trolled by a small microcontroller using a custom OS. Hence it is easily customizable for a variety of sporting applications. The platform is packaged so as to be robust, hermetically sealed and biomechanically neutral to the athlete. Results and derivative data from rowing and swimming are presented as sample applications. In swimming appli- cations, stroke characteristics for a variety of training strokes are analyzed and found to be better than hand timed data and comparable with results obtained from technology and labor intensive methods, including video analysis. The rowing application when coupled with other monitoring techniques such as impeller velocity, enables recovery of intra and inter stroke phases as a means to assess performance over the entire course and has been used by competition rowers to improve performance at national and international competitions.