Wearable sports monitoring equipment and method for characterizing sports performances or sportspersons

Abstract

The invention concerns wearable electronic devices, systems and methods for sports performance monitoring. In one embodiment, the invention provides a device comprising a heartbeat sensor for providing a heartbeat signal, a motion sensor for providing a motion signal and processing means adapted to calculate at least one performance parameter depicting said sports performance and/or the person using temporal characteristics of periodic features of the heartbeat signal compared with temporal characteristics of periodic features in the motion signal. The invention allows for utilization of an existing relation between cadence and heart rate for characterizing the performance or the person in a novel way.

Claims

1. Wearable electronic equipment for monitoring a sports performance of a person, comprising: a heartbeat sensor for providing a heartbeat signal in response to heartbeat of the person, a motion sensor for providing a motion signal in response to movement of the person, and a processing means adapted to calculate at least one performance parameter depicting said sports performance of the person based on temporal characteristics of periodic features of the heartbeat signal compared with temporal characteristics of periodic features of the motion signal, wherein the processing means is adapted to: compute, using the heartbeat signal and motion signal, a time delay parameter depicting the delay between a response in the heartbeat signal to intensity changes in the performance, and a response in the motion signal to intensity changes in the performance, and compute said at least one performance parameter as a fatigue index, such as EPOC, and/or as a fitness index, using said time delay parameter.

2. The wearable electronic equipment according to claim 1, wherein the processing means is further adapted to: associate first time stamps with individual heartbeats detected from said heartbeat signal, and associate second time stamps with said periodic features detected from said motion signal, using said first and second time stamps to detect and/or estimate a heartbeat frequency at which the temporal difference in the periodicities of the heartbeat signal and said motion signal remain essentially constant over a plurality of periods of the signals, and calculating a performance parameter based on said detected and/or estimated heartbeat frequency.

3. The wearable electronic equipment according to claim 2, wherein the processing means is further adapted to detect and/or estimate, using said first and second time stamps, a heartbeat frequency at which the temporal difference between the individual heart beats and temporally related individual movements remains essentially constant over a plurality of periods of the signals.

4. The wearable electronic equipment according to claim 1, wherein the processing means is further adapted to calculate a Fourier transformation of the heartbeat and motion signals, and to compare characteristics of the signals in a frequency domain in order to calculate said performance parameter.

5. The wearable electronic equipment according claim 1, wherein the processing means are adapted to compute said at least one performance parameter as a fatigue index using the magnitude of said time delay parameter and/or by observing changes in the time delay parameter during the performance.

6. The wearable electronic equipment according to claim 5, wherein the processing means is further adapted to determine a first value of said time delay parameter when the person is at rest and a second value of said time delay parameter during a performance, and to compute said at least one performance parameter using a mathematical function dependent on said first value and said second value.

7. The wearable electronic equipment according to claim 1, wherein the processing means is further adapted to read a predetermined time delay parameter and/or a behavior model of such a time delay parameter, and to compute said at least one performance parameter as a fitness index, using the heartbeat signal, the motion signal and the time delay parameter.

8. The wearable electronic equipment according to claim 1, further comprising an electric heart rate module integral with or functionally connectable with a heart rate belt or a smart garment having integral heart rate measurement electrodes so as to form said heartbeat sensor, said electronic heart rate module including said motion sensor.

9. The wearable electronic equipment according to claim 1, wherein the electric heart rate module further comprises said processing means.

10. The wearable electronic equipment according to claim 1, further comprising a wristop computer or mobile phone capable of wireless communication with an electric heart rate module for receiving at least said heartbeat signal, motion signal or data processed in such a heart rate module.

11. The wearable electronic equipment according to claim 1, further comprising: at least one displayless electric heart rate module integral with or functionally connectable with a heart rate belt or a smart garment having integral heart rate measurement electrodes so as to form said heartbeat sensor, and the module comprising said motion sensor and at least part of said processing means, and at least one wristop computer or mobile handheld device capable of wireless communication with the electric heart rate module and being provided with a display for visualizing said performance parameter.

12. The wearable electronic equipment according to claim 1, wherein said heartbeat sensor comprises an electrical ECG sensor, an optical sensor, a pressure sensor or an acceleration sensor.

13. A method for monitoring a sports performance of a person, comprising the steps of: measuring the person's heartbeat using a wearable heartbeat sensor for providing a heartbeat signal, measuring movement of the person using a wearable motion sensor for providing an motion signal, and in one or more computing units functionally connected to the wearable heartbeat sensor and the wearable motion sensor: detecting periodic features in the heartbeat signal and in the motion signal, determining temporal correlation of the periodic features of the heartbeat signal and the motion signal, and calculating at least one performance parameter based at least partly on said temporal correlation, wherein the at least one performance parameter comprises a fatigue and/or a fitness index determined by: computing, using the heartbeat signal and motion signal, a time delay parameter depicting the delay between a response in the heartbeat signal to intensity changes during the performance, and a response in the motion signal to intensity changes during the performance, and computing said fatigue and/or a fitness index using the time delay parameter.

14. The method according to claim 13, wherein the at least one performance parameter comprises a fatigue index determined by computing said fatigue index using the magnitude of the time delay parameter and/or by observing changes in the time delay parameter during the performance.

15. The method according to claim 14, further comprising: determining a first value of said time delay parameter when the person is at rest and a second value of said time delay parameter during a performance, and computing said fatigue index using a mathematical function dependent on said first value and said second value.

16. The method according to claim 13, wherein the at least one performance parameter includes a fitness index determined by providing a predetermined time delay parameter and/or a behavior model of such time delay parameter, and by computing the fitness index using the temporal correlation of the heartbeat signal and motion signal, and the time delay parameter.

17. The method according to claim 16, wherein the user is at a fully rested state at the beginning of the performance.

18. A non-transitory computer readable medium comprising computer-executable instructions for carrying out a method for monitoring a sports performance of a person, the method comprising the steps of: measuring the person's heartbeat using a wearable heartbeat sensor for providing a heartbeat signal, measuring movement of the person using a wearable motion sensor for providing an motion signal, and in one or more computing units functionally connected to the wearable heartbeat sensor and the wearable motion sensor, detecting periodic features in the heartbeat signal and in the motion signal, determining temporal correlation of the periodic features of the heartbeat signal and the motion signal, and calculating at least one performance parameter based at least partly on said temporal correlation, wherein the at least one performance parameter comprises a fatigue and/or a fitness index determined by: computing, using the heartbeat signal and motion signal, a time delay parameter depicting the delay between a response in the heartbeat signal to intensity changes during the performance, and a response in the motion signal to intensity changes during the performance, and computing said fatigue and/or a fitness index using the time delay parameter.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0046] FIG. 1A depicts a runner and his cadence.

[0047] FIG. 1B shows a graph of a runner's heartbeat and cadence vs. time.

[0048] FIG. 2 shows parallel schematic graphs of ECG and acceleration signals to illustrate correlation-based temporal analysis of the signals.

[0049] FIG. 3 shows a schematic view of a system according to one embodiment of the invention and some optional components and variations thereof.

[0050] FIG. 4A shows a block diagram of a heart rate belt and a wristop device according to one embodiment of the invention.

[0051] FIG. 4B shows a block diagram of a heart rate belt and a wristop device according to an alternative embodiment of the invention.

[0052] FIG. 5 shows a flow chart of the method according to one embodiment of the invention.

[0053] FIG. 6 shows a flow chart of the method according to an alternative embodiment of the invention additionally comprising a sports type determination phase.

[0054] FIG. 7 shows schematic graph of formation of a measured ECG signal by heartbeat-induced signal and cadence-influenced noise.

[0055] FIG. 8 shows a flow chart of filtering the heartbeat signal based on motion signal according to one embodiment of the invention.

[0056] FIG. 9 illustrates as a flow chart how persons heart rate related parameters can be linked to physical movement parameters introducing a time delay element in between the two domains according to one embodiment.

[0057] FIG. 10 illustrates as a flow chart how the different power estimates from intensity and heart rate based sensor sources are connected together according to one embodiment.

DETAILED DESCRIPTION OF EMBODIMENTS

[0058] FIG. 1A a runner 100A taking rhythmic steps 110A, i.e. moving forward with a relatively stable cadence. FIG. 1B shows real heart rate data 120 and cadence data 110 measured from a running performance (cadence multiplied by two to get both feet frequency). It can be seen that as the performance proceeds and the cadence takes its maximum value (about 180 l/min), the heart rate 120 approaches the cadence 110, the two quantities having in the end approximately the same values. Thus, there is a relation between cadence and heart rate.

[0059] According to one embodiment, the present invention takes advantage of this relation by calculating at least one correlation factor dependent on temporal characteristics of periodic features of the heartbeat signal (heartbeats) compared with temporal characteristics of periodic features in the acceleration signal (e.g. steps, revolutions). Mutual temporal comparison of the characteristic features in this way is used to calculate at least one performance parameter depicting intensity of said sports performance and/or fitness level of the person.

[0060] FIG. 2 illustrates one method of making the mutual comparison using an artificial ECG signal (upper) and acceleration signal (lower) forms. From each signal, peaks are detected and their time points are recorded, i.e. the peaks are time stamped. Then the elapsed time ?t.sub.1, ?t.sub.2, ?t.sub.3 between each acceleration peak and the next heartbeat peak is calculated based on corresponding time stamps. Then, it is determined whether the durations ?t.sub.1, ?t.sub.2 and ?t.sub.3 are essentially equal or not, or potentially or reflect a systematic trend. At the anaerobic threshold, the time differences between the time stamped peaks, i.e. ?t.sub.1, ?t.sub.2 and ?t.sub.3, approach a constant value. Suitable statistical methods and figures can be used to evaluate the behavior of the time differences.

[0061] Instead of peaks, the detection can be based on some other characteristic features of the signals, depending on the intended use. In particular, for the acceleration signal, different detection algorithms may be necessary for different sports and/or different location of the acceleration sensor, in order to obtain the cadence in a reliable way.

[0062] Taking the deduction made above with reference to FIG. 2 even farther, if it is know that the acceleration signal is measured at a heart rate belt (and therefore gives peak at every step) and the durations ?t.sub.1, ?t.sub.2 and ?t.sub.3 are statistically equal, it can be concluded that the step and heart rate are equal and that the person is working at or close to his/her anaerobic threshold. On the other hand, the durations are systematically changing in one direction or the other, it can be estimated how far below or above the anaerobic threshold the person is working at. The knowledge on the deviation from the anaerobic threshold can be used together with heart rate and/or cadence and/or speed data in further estimations on the intensity of the performance or the fitness level of the person.

[0063] The performance parameter can in particular comprise anaerobic heart rate threshold level of the person determined by detecting or estimating the heart rate level at which the heart rate and stepping frequency are equal or the heart rate is a multiple of stepping frequency. In running, this means the condition that cadence multiplied by two equals the heart rate is satisfied. The parameter may also comprise deviation from anaerobic heart rate threshold level determined by detecting or estimating the heart rate level at which the periodicities of the heart rate and the acceleration are equal and determining the difference between the present heart rate level and the anaerobic heart rate threshold level determined. The outcome may also be a derivative of threshold level or deviation therefrom, i.e. another quantity calculated at least partly using the estimated threshold level or deviation.

[0064] The performance parameter can also comprise maximum step length or maximum speed in the aerobic range. In this embodiment, the step length of the person is determined using suitable means (e.g. acceleration-based foot pod measurement, combined acceleration and GPS measurements or user-given parameter). Then, the maximum speed v.sub.anaer that the person is able to maintain for a long period equals 2*cadence*step_length. This can further be used to calculate back the maximum step length or to estimate maximum oxygen intake and further fitness level of the person. For example maximum oxygen intake VO.sub.2,max is obtained using the formula A*v.sub.anaer+B, where A and B are predetermined constants. Then using VO.sub.2,max and the age and sex of the person, one can robustly estimate the fitness level of the person among a population with methods known per se.

[0065] According to one embodiment, the maximum step length or maximum speed in the aerobic range and/or anaerobic threshold heart rate are used as scaling factors in heart rate based energy consumption determination. Thus, the equipment can self-calibrate its energy consumption determination based on the fitness-related data obtained by means of the invention. These scaling factors can be used not only in characterizing running, but also other sports, after being determined on the basis of a running performance.

[0066] FIG. 2 is given to exemplify one potential method only in a simplified way using artificial signal forms. In practice, a longer inspection period would be needed to make reasonable conclusions on the relation between the cadence and heart rate and deviation from the anaerobic threshold. It is appreciated that the same information can be obtained using other correlation-based methods or through frequency analysis.

[0067] Heart rate of a person typically follows changes in the intensity of physical performance with a delay. Therefore, performances that contain a lot of physical intensity changes are challenging as concerns heart rate and physical parameter comparisons. FIG. 9 illustrates how heart rate related parameters 92 can be linked to physical movement parameters 94 introducing a time delay element ?.sub.?? in between. This enables real time calculation of performance parameters using either or both the heart rate and movement signals at all times. The time delay element ?.sub.?? describes the difference of response times of physical measurement and heart rate measurement to the intensity of the performance.

[0068] As a more detailed example, FIG. 10 shows how power estimations using heart rate and moment data are linked together with this method. Mathematically this link can be formulated in the form of differential equation

[00001]$\left[\begin{array}{c}{P}_{\mathrm{Movement}}^{\&}\\ P_{\mathrm{hr}}^{\&}\end{array}\right]=\left[\begin{array}{cc}0& 0\\ 1/{?}_{\mathrm{hr}}& -1/{?}_{\mathrm{hr}}\end{array}\right]?\left[\begin{array}{c}{P}_{\mathrm{movement}}\\ P_{\mathrm{hr}}\end{array}\right]$

where P.sub.movement is the power estimate based on physical movement data, P.sub.hr is power estimate based on heart rate signal, and ?.sub.hr is the parameter describing the time delay between heart rate response and physical movement based intensity change. Changes in ?.sub.hr, relative to resting condition ?.sub.hr.sub._.sub.rest, correlate with real time EPOC. Therefore ?.sub.hr describes person's fatigue level in real time, without any need for prior exercise history data. Function

EPOC=f(?.sub.hr, ?.sub.hr.sub._.sub.rest)

that is used for EPOC calculation, can be for instance partial linear model that is formed based on test database data. The process of FIG. 10 thus enables estimation of EPOC value and fatigue of person from intensity changes or in the reverse case to calculate fitness index at all times even during the intensity changes.

[0069] In more detail, FIG. 10 shows how one can determine energy consumption 1024 and/or metabolic equivalent of task (MET) 1020 starting from data provided by a motion sensor 1002 and heart rate sensor 1034, an optional power sensor 1014. In the process the motion sensor data from different sensor sources, for instance footpod speed, wrist accelerometer, gps speed, bike pod speed, or bikepod power sensor data, is always transformed to motion based power 1012 through functions implemented in 1004. Since we can have multiple motion sources they can be used to calibrate other sensor through the function 1010.

[0070] Power that person produces can be also calculated from heart rate data. However, since the heart rate adjusts with delay to intensity level changes, this delay must be taken into account in the system. Part 1026 of the system and especially time constant tau ?.sub.hr in function 1028 presents models the delay properties of the human heart relative to motion based data. In this particular implementation of the system model there are two variables that can be allowed to change, fitness index in function 1030 and time constant tau ?.sub.hr in function 1028. The system model can change these values in order to make the heart rate power equal to motion power. When there are small changes in motion power, that is in sport intensity level, then the system model adjusts more strongly the fitness index so that the heart rate and motion based powers are equal. Thus, the system determines the user's fitness index. On the other hand, when there are a lot of intensity changes, fitness index is changed less and the system model adjusts more the time constant tau ?.sub.hr. As described earlier, changes in time constant tau ?.sub.hr during the performance give the fatigue level and thus EPOC of the user.

[0071] In some cases the value and behavior of time constant tau ?.sub.hr is known, for instance when user is fully rested prior to the performance and has done an orthostatic heart rate test. Then the fitness index of the user can be calculated accurately and quickly even in activities where the intensity changes a lot, for instance in hockey, soccer, badminton, tennis, floor ball, etc.

[0072] FIG. 3 shows a system with a wristop computer 220 as a data storage and/or visualization device and a heart rate belt 210 as a data-collecting and processing device. The heart rate belt comprises an ECG sensor and an electronic module that reads and processes the signal provided by the ECG sensor. An acceleration sensor is provided either in the heart rate belt 210 or the wristop computer 220, or both. The heart rate belt 210 is in wireless communication with the wristop computer.

[0073] Alternatively to a heart rate belt 210, a smart garment 230 can be used, providing the same functionality as the heart rate belt 210 discussed above.

[0074] In a still further embodiment, there is provided an additional unit in the system, such as a sensor unit attached to a shoe 240 of the sportsman. The additional sensor unit may comprise an acceleration sensor providing the acceleration signal and is in wireless communication with the heart rate belt 210 or smart garment 230 and/or the wristop computer 220 to provide the acceleration signal or data derived therefrom for further processing or use.

[0075] It should be noted that the wristop computer can be replaced with any other wearable device, such as another wearable sports tracking unit or mobile phone capable of communicating with other components of the system.

[0076] Wireless communication of data between separate devices of a multi-device equipment takes place advantage of a wireless radio-frequency transmitter-receiver or transceiver-transceiver pair. The wireless communication protocol can be one used for communication between computers, and/or between any remote sensors, such as a Bluetooth LE or the proprietary ANT+ protocol. These are using direct-sequence spread spectrum (DSSS) modulation techniques and an adaptive isochronous network configuration, respectively. Enabling descriptions of necessary hardware for various implementations for wireless links are available e.g. from the Texas Instrument?'s handbook Wireless Connectivity which includes IC circuits and related hardware configurations for protocols working in sub-1- and 2.4-GHz frequency bands, such as ANT?, Bluetooth?, Bluetooth? low energy, RFID/NFC, PurePath? Wireless audio, ZigBee?, IEEE 802.15.4, ZigBee RF4CE, 6LoWPAN, Wi-Fi?.

[0077] FIG. 4A depicts one preferred implementation of the system in more detail. There is provided a heart rate belt 510 with a pair of ECG electrodes 513A, 513B connected to an ECG signal preprocessor 514 (typically including at least an A/D converter) contained in an integral or removable electronics module 512A. The ECG data obtained from the preprocessor is processed in digital form in a processing unit 516A running an algorithm for finding the individual heartbeats and optionally further the heart rate. The processing unit 516A can also provide time stamps for the heartbeats. The heartbeat data, i.e., the heartbeats, corresponding time stamps and/or the heart rate, are wirelessly communicated to a wristop computer 520 via a radio transmitter unit 518 in the belt 510 and radio receiver unit 524 in the wristop computer. The receiver unit 524 is functionally connected to a processing unit 522A. In addition, there is provided an acceleration sensor 528 in the wristop computer functionally connected to the processing unit 522A.

[0078] The processing unit 522A is configured to analyze the heartbeat data (heart beat pulses, time stamps or heart rate) received from the heart rate belt 510 and acceleration data from the acceleration sensor 528 to find their temporal correlation, for example using one of the methods described above in more detail. The processing unit 522A also calculates the desired performance parameter using the correlation information.

[0079] The processing unit 522A may also be configured to execute software for determining the type of sports based on the acceleration signal and for determining energy consumption using an algorithm and source(s) of information (ECG and/or acceleration) depending on the sports type determined. Alternatively or in addition to automatic sports type determination, the processing unit may be configured to read a stored sports type parameter, typically given by the user or previously automatically determined, from a memory (not shown) of the device and to use that for selecting an appropriate algorithm and source(s) of information. Exemplary methods for automatic sports type determination are described elsewhere in this document.

[0080] FIG. 4B shows an embodiment mostly similar to that of FIG. 4A but having an acceleration sensor 515 in the electronics module 512B of the heart rate belt. The processing unit 516B is configured to read the acceleration signal from the acceleration sensor 515. The processing unit 516B may also carry out one or more of the functions of the processing unit 516A described above and to additionally process the acceleration signal correspondingly to find repetitive movements or movement frequency. The processing unit 516B may also determine the type of sports based on the acceleration signal and to communicate the type, along with the ECG and/or acceleration data, to the wristop computer, which then runs the energy consumption algorithm in its processing unit 522B. According to a second variant, the processing unit 516B of the heart rate belt 510 is also configured to analyze the temporal correlation of the heartbeat and acceleration signals and optionally to determine the desired performance parameter. Likewise, it may also run the energy consumption algorithm either partially or entirely. As an example of partial calculation, energy consumption per a mass unit can be determined based on the heartbeat data or acceleration data in the heart rate belt and then this is multiplied with the user mass in the wristop computer having an interface for inputting the user mass. In one embodiment, the communication interface between the devices allows for transmission of data to the heart rate belt, whereby also the mass information can be sent and utilized in the heat rate belt to obtain a user-specific energy consumption value.

[0081] FIG. 5 shows a flow chart of the present method according to one embodiment. The motion signal is measured using an accelerometer, magnetometer and/or gyro in step 702A and the ECG signal with an ECG sensor simultaneously in a parallel step 702B. The signals are then processed in steps 704A and 704B to determine periodic motions and heartbeats, respectively. The temporal characteristics of the periodic motions and heartbeats are then compared in step 706 to find their potential correlation. If the motions and heartbeats are in synchronization, i.e. their frequencies are equal or one is a multiple of the other, there is full correlation (there may well be a phase shift between the periodic frequencies). A correlation factor of some kind is derived to describe the frequency difference. Then, based on the correlation factor, the desired parameter of interest is calculated in step 708 and displayed and/or stored in step 710.

[0082] According to one embodiment particularly suitable for monitoring running or cycling performances, speed and/or cadence length information is additionally utilized to determine the performance parameter. This option is illustrated with dashed in FIG. 5 as lines as step 709 providing additional input to the computation step 708. Speed can be obtained directly using data from a position sensor, such as a satellite positioning sensor, or velocimeter, such a cyclometer, being part of the system, or using e.g. known or measured step (running) or revolution (cycling) length and the cadence obtained from the motion sensor. Step length can also be approximated using the sensor data obtained from the accelerometer, magnetometer or gyro.

[0083] According to one embodiment, the computing means is programmed to estimate energy consumption of the person during the performance using the heartbeat and/or motion signals. In particular, the equipment may comprise means for determining the type of the sports performance and the computing means for calculating the energy consumption parameter are adapted to utilize the heartbeat signal, the motion signal or both signals in combination in different manner depending on the type of the sports performance determined. The type of the sports performance can be determined either manually by the user but even more advantageously automatically.

[0084] According to one preferred embodiment, the device is capable of distinguishing the sports based on physical rhythmicity, i.e. similarity and temporal stability of successive similar movements, required in particular sports. For example, street or track running is physically very rhythmic sports, i.e. involves a stabile cadence, whereas tennis is more impulsive and irregular. The proposed solution can robustly distinguish between these kinds of sports and apply a corresponding algorithm for calculating energy consumption, which utilize at least partly different sources of information. Energy consumption of running may be better characterized by cadence and tennis by heart rate characteristics. Automatic type determination based on the stability of cadence is also capable of distinguishing between street or track running and cross county running or orienteering, which involve kinematically and/or temporally more versatile motion and in which energy consumption is usually better characterized by heart rate.

[0085] According to one embodiment, there are provided means for receiving and storing a user-specified sports performance type parameter in a memory of the equipment. This is called manual type determination. In addition, the means for determining the type of the sports performance are adapted to read said user-specified sports performance type parameter from said memory in order to determine the type of the sports performance. The means for receiving and storing may comprise user interface means and associated software and hardware means in a wristop computer or mobile phone part of the present system. If necessary, the performance type parameter can be communicated to a heart rate belt or some other unit where energy consumption calculations are carried out.

[0086] According to one embodiment, the means for determining the type of the sports performance are adapted to determine the type of the sports performance automatically based on characteristics, in particular rhythmicity, of the motion signal. According to a further embodiment, this is implemented such that the means for determining the type of the sports performance are adapted to calculate a parameter depicting rhythmicity of the performance based on the motion signal and the device is adapted to use that parameter for determining the type. In one approach, the means for calculating the energy consumption parameter are adapted to utilize only the ECG signal in calculations if the rhythmicity of the performance is below a predefined level and only the motion signal if the rhythmicity of the performance is above the predefined level. Thus, rhythmic sports are characterized by the acceleration data, for example and irregular sports by the heart rate data, allowing for more accurate estimations for energy consumption.

[0087] According to one option, the means for determining the type of the sports performance are adapted to determine rhythmicity of the motion signal by detecting repeating features in the motion signal, associating time stamps to the repeating features, and calculating correlation between intervals between successive time stamps, preferably at all sensor axes. Correlation value per axis forms characterization vector that is utilized to differentiate between types of sports. Such correlation analysis can be implemented in a wearable device efficiently in the time domain.

[0088] In an alternative option, the means for determining the type of the sports performance are adapted to apply Fourier analysis on the motion signal to determine rhythmicity of the motion signal. Discrete Fourier analysis can also be relatively efficiently implemented in small devices.

[0089] FIG. 6 shows an embodiment otherwise similar to that illustrated in FIG. 5 but having additional sports type determination. The motion and ECG signals are measured in steps 802A, 802B and used for motion and heartbeat detection in steps 804A, 804B, respectively. The motion data is further used in step 806 for determining the type of the sports being performed. Methods for this are described elsewhere in this document in more detail. If the sports type is rhythmic, it makes sense to continue with temporal comparison of the motion and heartbeat data in step 810 and to characterize the performance based on the comparison in step 812. In addition, energy consumption is calculated using or putting more weight on the motion signal in step 807A. If the sports is irregular by nature, the comparison phase is skipped and energy calculated on the basis of the heartbeat signal in step 807B. The results of the calculations are displayed and/or stored in step 814.

[0090] The dashed lines in FIG. 6 indicate that the result of temporal comparison of the motion and heartbeat signals (step 810) or the parameter derived based on the comparison (step 812) can also be used in the energy consumption calculation (step 807A). Thus, the deviation from the anaerobic threshold and therefore different metabolic states can be taken into account. This is, however, optional since in certain rhythmic sports, such as running, the energy consumption can also relatively accurately be determined based on the motion data only.

[0091] According to one embodiment, for determining the type of sports, the motion signal is analyzed so as to detect cyclic, i.e. regular, motion characteristic to a cyclic sports, such as running, in it. Rhythmicity can be determined in the time domain for example by time stamping impulses detected in the signal using correlation analysis to determine whether the inter-impulse interval remains constant (to a predefined degree) over time, which would be indicative of a performance of regular nature. If the correlation between intervals is low, the performance is irregular. In an alternative embodiment, the motion based sensor signal is converted to frequency domain using a discrete Fourier transformation, and the presence of strong (peaked) frequency components is detected. There is a peak in the frequency data if a lot of cyclic motion takes place at constant frequency over a chosen time interval. The rhythmicity determination may be carried out continuously or repeated at predefined intervals during the performance so that the energy consumption algorithm can be changed, if the type of sports changes. This can occur for example if the user carries out gymnastic exercises every now and then during a jog.

[0092] As concerns the energy consumption part, the determination of the type of sports can have more than two outcomes (not just between rhythmic and irregular), and there may be more than two algorithms corresponding to these outcomes implemented.

[0093] According to one embodiment, it is determined if the user has specified a particular sports type manually (and wishes that to be used as a basis for algorithm selection). In the affirmative, the algorithm is chosen accordingly without automatic determination. In the negative, automatic type determination is initiated, as described above.

[0094] The rhythmicity analysis should be carried out over a time period of significant length, so that random and occasional variations in the frequency of motion or measurement errors do not result in a wrong outcome. If the rhythmicity stays relatively constant over the period chosen, a first algorithm is chosen for further calculations and on the opposite case a second algorithm is chosen.

[0095] In a still further embodiment, the computing unit is programmed to filter out movement-induced components of the heartbeat signal based on the motion signal for providing a filtered heartbeat signal and to use to filtered signal for deriving the performance parameter. This embodiment solves the problem of generation of movement-induced repeating errors in the heartbeat signal. In particular at the beginning of a performance, when the measurement electrodes of the measurement device and also the skin is dry, variations in the contact pressure between the electrodes and the skin due to cadence result in measurement artifacts. For the same reason, also static electricity may accumulate close to the electrodes and cause undesired peaks in the signal measured.

[0096] FIG. 7 illustrates ECG signal formation. In an ideal case, the heartbeat of a runner, for example, produces a clean ECG signal with easily detectable heartbeat waveforms 1120 as illustrated by the uppermost graph of FIG. 7. However, in practice, there is always random noise 1135 present, as well as systematic noise, i.e., artifacts 1130 induced by the movement of the runner. In the case of rhythmic cadence, the artifacts are typically repeated periodically, following the more or less constant rhythm of the movement. The random noise 1135 and artifacts 1130 are illustrated by the middle graph of FIG. 7. Thus, the real measured signal contains the sum of the real heartbeat signal and the noise and artifact signals, which is illustrated by the lowermost graph of FIG. 7. From this sum signal, it is much more difficult to detect the heartbeats robustly than from the ideal ECG signal.

[0097] According to one embodiment, the computing unit is programmed to filter out frequency components from the heartbeat signal that are common to the motion signal and the heartbeat signal. Preferably, this is carried out in the frequency domain by computing a Fourier transform of the heartbeat signal and a Fourier transform of the motion signal and suppressing frequency components of the muscular activity signal that correspond to some or all of the frequency components of the motion signal.

[0098] Preferably, the heartbeat sensor and the motion sensor are located in the same device unit, such as a heart rate belt or smart garment. This ensures that the motion signal corresponds to actual movement of the heartbeat sensor. It is however possible that the sensors are located in different device units potentially located on different body parts, because their rhythmic movements are typically not independent and therefore sufficient information for removing rhythmic artifacts can be obtained.

[0099] FIG. 8 shows one exemplary flow schematic for signal filtering. The heartbeat raw signal is measured in step 300 from a suitable location of the body of the person using a suitable sensor. In step 302, it is checked whether rhythmic filtering feature of the system has been activated. If not, the heartbeats or muscular activities are detected in step 312B based on the raw signal using a suitable algorithm, which are known per se. If the filtering is set active, motion signal is measured in step 304 to assist in the filtering process. A discrete Fourier transform (DFT) of the motion signal is computed in step 306, preferably using a suitable fast Fourier transformation algorithm. The same is carried out for the heartbeat signal in step 308. The DFT of the motion signal indicates the frequencies at which there may be motion-induced artifacts present in the heartbeat signal. Some or all of these frequencies are then suppressed from the heartbeat signal, preferably in the frequency domain to provide a filtered heartbeat signal in step 310. Individual heartbeats are then detected using the filtered signal in step 312A. In step 314, the results of the detection are then utilized in further computations to characterize the sports performance or the person and/or stored in a memory of the system and/or visualized. Typical uses of the heartbeat data include energy consumption estimation and training effect estimation.

[0100] The steps of the FIG. 8 can naturally take place in any suitable order, which may deviate from the one illustrated. Moreover, if the present signal cleaning method is set to take place by default, steps 302 or 312B are not needed.

[0101] It is to be understood that the embodiments of the invention disclosed are not limited to the particular process steps, units, components, devices, materials or products disclosed herein, but are extended to equivalents thereof as would be recognized by those ordinarily skilled in the relevant arts. It should also be understood that terminology employed herein is used for the purpose of describing particular embodiments only and is not intended to be limiting.

[0102] Reference throughout this specification to one embodiment or an embodiment means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases in one embodiment or in an embodiment in various places throughout this specification are not necessarily all referring to the same embodiment.

[0103] As used herein, various embodiments and examples of the present invention may be referred to herein along with alternatives for the various components thereof. It is understood that such embodiments, examples, and alternatives are not to be construed as de facto equivalents of one another, but are to be considered as separate and autonomous representations of the present invention.

[0104] Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the description, numerous specific details are provided, such as examples of shapes and dimensions etc., to provide a thorough understanding of embodiments of the invention. One skilled in the relevant art will recognize, however, that the invention can be practiced without one or more of the specific details, or with other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the invention.

[0105] While the forgoing examples are illustrative of the principles of the present invention in one or more particular applications, it will be apparent to those of ordinary skill in the art that numerous modifications in form, usage and details of implementation can be made without the exercise of inventive faculty, and without departing from the principles and concepts of the invention. Accordingly, it is not intended that the invention be limited, except as by the claims set forth below.