Method and apparatus for estimating energy consumption

Abstract

The present invention relates to a method and apparatus for estimating the energy consumption of a person on the basis of heart rate data. In the method, the beat rate of heart is measured with a sensor or previously measured heart rate data are input for providing heart rate data and the energy consumption of a person is determined on the basis of heart rate data. According to the invention, a first threshold value is selected for the mass of the person and in case the mass of the person is larger than the first threshold value, energy consumption is calculated using a formula taking into account the deviation of the person's mass from the said first threshold value. The invention allows getting more accurate energy consumption estimates especially for overweight persons.

Claims

1. A method of estimating a person's energy consumption in a portable electronic device including a data processing unit operably coupled to a heartbeat sensor, the method comprising the steps of: measuring heartbeat with a sensor or taking previously measured heartbeat data for providing heartbeat data of said person; measuring the intensity of the exercise based on the respiratory frequency of the person or said heartbeat data; determining a first estimate of the person's energy consumption on the basis of heartbeat data; selecting a first threshold value for a mass of the person, over which threshold an increasing deviation of the actual mass is rendering said first estimate increasingly inaccurate; selecting a second threshold value for an intensity of the exercise, from which threshold a decreasing deviation of the intensity is rendering said first estimate increasingly accurate; wherein when said person's mass is larger than the first threshold value, multiplying said deviation of mass with the deviation of intensity to produce an effective mass of the person which varies with the intensity of the exercise, from being smaller than the actual mass of the person to approaching the actual mass, as the intensity of the exercise approaches the second threshold value, and by calculating an improved estimate of said person's energy consumption using said effective mass.

2. A method of estimating a person's energy consumption as a Metabolic Equivalent of Task (MET), the method comprising the steps of: selecting a first threshold value for a mass of the person; selecting a second threshold value for an intensity of the exercise; and if the person's mass is larger than the first threshold value, calculating an estimated energy consumption of the person with a formula taking into account a deviation of the person's mass from the first threshold value, wherein an effective mass of the person is a factor in the formula, which is smaller than the person's mass, and wherein the second threshold value is selected so that the second threshold corresponds to an intensity of exercise where the energy consumption of a person is in the range of 1.7 to 2.3 MET, said steps being carried out in a portable electronic device including a data processing unit operably coupled to said heartbeat sensor to produce an improved estimate of said person's energy consumption.

3. The method of claim 2, wherein said second threshold corresponds to an intensity of exercise where the energy consumption of the person is approximately 2 MET.

4. A method of estimating a person's energy consumption, the method comprising the steps of: selecting a first threshold value for a mass of the person wherein said first threshold value is determined using a body weight index depending on the weight of the person and the height of the person; selecting a second threshold value for the intensity of the exercise, wherein the intensity and the second threshold value are determined based on heart rate frequency, respiratory frequency or ventilation, wherein ventilation is calculated on the basis of respiration frequency with the formula ventilation=respiratory frequency*vital capacity*correction factor, wherein the correction factor depends on the intensity of the exercise, and wherein the vital capacity is provided as pre-data, depending on one or more of the person's gender, the person's age and the person's height, and if the person's mass is larger than the first threshold value, calculating the estimated energy consumption of the person with a formula which multiplies the deviation of the person's mass from the first threshold value with the deviation of the intensity of the exercise from the second threshold value, producing an effective mass of the person which is smaller than the actual mass of the person and which approaches the actual mass as the intensity of the exercise approaches the second threshold value, said steps being carried out in a portable electronic device including a data processing unit operably coupled to said heartbeat sensor to produce an improved estimate of said person's energy consumption.

5. The method of claim 4, wherein the first threshold value is selected so that it corresponds with a body weight index of 18 to 25 of the person.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG. 1 illustrates the method according to one embodiment of the invention as a flow chart.

[0046] FIG. 2 illustrates an example of vital capacity according to age for females and males.

[0047] FIG. 3a exemplifies the change of the length of the inter-beat intervals as a descriptor in an exemplary exercise.

[0048] FIG. 3b illustrates the inter-beat interval data processed according to image 3a and Fourier-transformed as well as determining the respiratory frequency.

[0049] FIG. 4 schematically illustrate an apparatus 20 for determining energy consumption during or after a physical exercise of a person.

DETAILED DESCRIPTION OF THE INVENTION

Definitions

[0050] The term intensity (of an exercise) means the degree of exertion of the exercise. Intensity can be measured via heart rate, respiratory frequency, ventilation or a mathematical derivative or combination of these.

[0051] The term weight index or BMI primarily means the commonly accepted (e.g. World Health Organization, http//apps.who.int/bmi/indcx.jsp?introPagc=intro_3.html) and used definition of m/l.sup.2, wherein m is the mass of the person in kilograms and l the height of the person in meters, but it is not limited at this. As an expert will understand, the body shape, size and/or obesity of a person can be described also via other indexes, such as ones determined by e.g. height and weight, and they are suitable for use in the present invention as well for determining the threshold weight of mass separately for each person.

[0052] The abbreviation HR is used for referring to the absolute heart rate and the abbreviation hrr refers to heart rate reserve. It is the relation of the difference between heart rate and resting heart rate to the difference of maximum heart rate and resting heart rate, i.e. hrr=(HRHR.sub.rest)/(HR.sub.maxHR.sub.rest) (typically the unit is percent of heart rate reserve, i.e. % hrr=100%*hrr), wherein HR is the current heart rate, HR.sub.rest is the resting heart rate and HR.sub.max is the maximum heart rate.

[0053] Inter-beat interval means the temporal distance between two subsequent heartbeats from each other. As patent literature and other literature have disclosed a number of methods for recognizing inter-beat intervals, these methods are not discussed here in closer detail.

[0054] FIG. 2 illustrates an example of carrying out the invention on a relatively general level.

[0055] Heart rate is measured with a heart rate sensor 22, such as a heart rate belt arranged over the chest of an individual, in step 10. As one skilled in the art will understand, also other ways of recognizing heart rate, known in the field, can be used.

[0056] In step 11 the inter-beat intervals of subsequent heartbeats and further the inter-beat interval variation are determined from the heart rate data. The periodicity of the change of inter-beat intervals is indicative of respiratory frequency which is further determined in step 12.

[0057] In step 13, ventilation is determined based on the respiratory frequency and pre-data.

[0058] In step 14, it is determined on the basis of heart rate data and pre-data whether the weight and intensity range is one requiring effective mass correction. In case this is required, the process continues to step 15, in which the effective mass is calculated and further in step 16 energy consumption is calculated using the effective mass. In case the range is not one requiring effective mass correction, energy consumption is calculated directly on the basis of the actual mass of the person in step 18.

[0059] Calculation, i.e. steps 11 to 18 can be carried out in a suitable data processing unit 24, especially a computer, wrist computer or a mobile phone. Real-time energy consumption monitoring is preferably carried out in a wrist computer or a mobile phone. Most typically a computer is used for carrying out a post-analysis of the exercise.

[0060] Preferably the heart rate sensor is in wireless data transfer communication with the data processing unit.

[0061] The essential steps of the invention are discussed in more detail in the following.

[0062] Respiratory Frequency

[0063] According to one embodiment respiratory frequency is essentially determined by the method described in patent FI 121214 (the '117 patent). According to this method, the rate of a person's heart is monitored for providing a heart rate signal, respiratory frequency is determined on the basis of the periodicity of the temporal variation of the heart rate data contained by the heart rate signal so that the periodicity of the temporal variation of the heart rate data is determined in time level using time stamps created on the basis of the heart rate signal. Preferably respiratory frequency is determined so that a series comprising subsequent time points is formed of the time stamps, the periodicity of the series is determined, and the parameter describing respiration is determined on the basis of the sequence of the series. The sequence of the series can be determined by calculating the second derivative of the series and by looking for its zero points. For a more detailed description of the method reference is made to the '117 patent.

[0064] According to an alternative embodiment, respiratory frequency is determined as follows: [0065] the rate of a person's heart is measured by means of a suitable sensor, [0066] the lengths of inter-beat intervals are determined on the basis of heart rate data, [0067] the difference between subsequent inter-beat intervals is calculated and the difference is classified as value A, if the difference is negative, and as value B, if the difference is positive, Typically A=0 and B=1. Thus the execution of Fourier transform in continued analysis can be optimized further. [0068] The Fourier transform of the time series assembled as described above is calculated. If the data consists of values 0 and 1, there is no need for windowing and multiplication. [0069] Values between which the largest value is selected are selected from the frequency response of the conversion of the previous step based on the heart rate data. Its location in the frequency space is selected to be respiratory frequency.

[0070] The largest advantage of such calculation for portable apparatuses is that multiplication is not needed, and the calculation can easily and effectively be implemented with integer calculation. At the end of the disclosure, there is a more detailed example about carrying out the calculation in practice. It is to be noted that the implementation described here is only suitable for cases in which it is desired to find out the periodicity of the data and it does not replace full Fourier transform. Additional advantages are that it is not necessary to separately correct heart rate data prior to analysis and it is not necessary to separately remove heart rate level changes there from. A change of heart rate level would mean an increase of mean heart rate as a result of e.g. increase of running speed. Such changes are seen in the frequency conversion of inter-beat intervals if they are not separately removed.

[0071] According to one embodiment, however, the following heart rate data correction is carried out: [0072] the difference, diff, between subsequent values is calculated, and [0073] if the difference is too large or too small (abs(diff)>quality trigger), 0 is selected as classification result.

[0074] Additionally, if ventilation data is not needed as such anywhere, calculation can be used only when it is observed that the intensity of the exercise in within the intensity range of the present invention, i.e. low enough.

[0075] It should be noted that the new respiratory frequency calculation method presented here is averaging in nature, i.e. the result is in this regard more reliable than the determination of periodicity directly in time level as disclosed in the '117 patent.

[0076] Ventilation

[0077] At its simplest, ventilation is a product of respiratory frequency and depth of respiration (tidal volume). Estimating the respiratory depth requires data about vital capacity. Vital capacity can be estimated on the basis of literature. For example, the publication of American Thoracic Society, Lung Function Testing: Selection of Reference Values and Interpretative Strategies, Am Rev Respir Dis 1991, American Thoracic Society, March 1991, can be used as a source. In this reference, vital capacity has been tabulated as a function of gender, age and height.

[0078] The above-mentioned literature reference contains general values that are most accurate in older age groups. More accurate estimates are available for especially younger age groups and they can be tabulated using reference material as well. FIG. 2 shows a more accurate example of vital capacity as a function of age for males and females produced partly based on the above-mentioned reference and partly based on reference material. The vital capacity according to the invention can, if necessary, be further compensated for height.

[0079] When the vital capacity multiplied by respiratory frequency has been compared with the ventilation values of reference measurements, there was observed a need for a heart rate-dependent correlation function, which can be static and is provided e.g. as an average of reference measurements. According to one embodiment, the factor depends on the value % hrr determined above. In other words,

ventilation=respiratory frequency*vital capacity*correction factor (% hrr).

Thus, when taking the above-mentioned issues into account, ventilation VE in this context depends on many factors, the most important of which are gender, age, height, % hrr, and respiratory frequency.

[0080] Energy Consumption in Resting State and Low Intensity

[0081] According to one embodiment, the level of basic metabolism, i.e. the BMR value per kilogram, is supposed to be constant, whereby the fixed estimate for oxygen consumption is 1 MET=1 ml/kg/min.

[0082] According to a preferred embodiment, a more accurate BMR value and further an oxygen consumption relating to said BMR value are used. For this purpose there are numerous formulae available from literature. For example, historically the most important are the Harris-Benedict equations from 1919:

BMR_males=13.7516*m+5.0033*h6.775*a+66.473

BMR_females=9.5634*m+1.8496*h4.6756*a+655.0955

In the above, m is weight in kilograms, h is height in centimeters and a is age in years.

[0083] It has, however, been noted that near resting state the above-mentioned method gives too high an oxygen consumption estimate for overweight persons.

[0084] According to the invention, this problem can be solved by determining threshold mass, m.sub.0 (so-called zero mass), for overweight persons. The accurate body weight index, BMI, for providing this data, can be determined by means of e.g., reference measurements and the difference between the values produced by the method and the reference values. It is also possible to use a BMI value on the range of normal weight 18.5 to 25. In testing it has been found that a relatively good value is a BMI value of about 19.

[0085] More specifically, the idea of the correction based on BMI is to select a threshold value for both low intensity (second threshold value) as well as the threshold mass (first threshold value) and to interpolate the zero mass so as to be the correct mass, when the intensity of the exercise changes from zero to this threshold value. This can be done through effective mass m.sub.eff. In mathematical terms the effective mass at low intensities is

m.sub.eff=m.sub.0+a*(mm.sub.0)*(II.sub.0).

In the above, m is weight and intensity can be described by e.g., the above-mentioned unit % hrr, ventilation or other unit describing the intensity. I.sub.0 is the selected threshold value for intensity. The actor a is a scaling constant.

[0086] When the intensity of the exercise is lower than the chosen threshold value for intensity (I<I.sub.0), if BMI is larger than the threshold value (due to which m>m.sub.0), effective mass m.sub.eff is used as basis for calculation, as is described below in more detail. On the other hand, if BMI is lower than the selected threshold value, (m<m.sub.0), actual mass is used directly as mass.

[0087] Finally, the units m.sub.eff and ventilation described above are used for calculating an estimate for momentary energy consumption at low intensities.

E=vo.sub.2(ventilation)*m_eff/200.

[0088] The function vo.sub.2 about ventilation can be, for example

vo.sub.2 (ml/kg/min)=0.385 (ml/I)*ventilation (I/min)/m.sub.real.

[0089] This function will change accordingly (shape, factors) to fit the data better, as the reference database gets more accurate.

[0090] If intensity is higher than intensity_0, effective mass correction is preferably not made as described above, but instead the method described in e.g., the '117 patent is used.

[0091] If no inter-beat interval data are available, BMR and BMI correction are applied directly to the calculated vo.sub.2 value (i.e. not ventilation corrected) at low intensities. The difference in these is that in resting state, the heart rate reacts to other than performed work and can thus be seen as energy consumption in the basic method. This can be compensated by selecting in the basic method the effective mass so that in relation to the reference measurements the results are unbiased (i.e. averages are the same but regression is not as good as in a method improved with ventilation).

[0092] Referring to FIG. 4, in a preferred embodiment of the present invention, an apparatus 20 for determining energy consumption during or after a physical exercise of a person includes the heart rate sensor 22, the data processing unit 24 and a memory 26. The heart rate sensor 22 is configured to measure the heartbeat of the person or to receive a heartbeat signal from an external source. The heartrate signal is representative of the heartbeat of the person. The data processing unit 24 is operably coupled to the sensor 22, the data processing unit 24 is configured to determine the length of inter-beat intervals from the heartbeat data and configured to determine the energy consumption of the person based upon the heartbeat data. The memory 26 is operably coupled to the data processing unit 24. The memory is configured to save pre-data relating to the person. The pre-data includes the person's mass, the person's body weight index, or a combination thereof and at least a first threshold value describing the mass or body weight index of a person. The data processing unit 24 is arranged to determine on the basis of the pre-data and the first threshold value whether the mass or body weight index of the person corresponds to a larger mass or weight index than the first threshold value. In the event the person's mass is larger than the first threshold value, the data processing unit 24 calculates energy consumption using a formula taking into account the deviation of the person's mass from the said first threshold value.

[0093] The memory 26 is arranged to save a second threshold value describing the intensity of the exercise. The data processing unit 24 is arranged to determine on the basis of heart rate data whether the intensity of the exercise is lower than the first threshold value. In case the intensity of the exercise is lower than the second threshold value and the mass of the person is larger than the first threshold value, the data processing unit 24 is arranged to determine the energy consumption using a formula taking into account the deviation of the mass and the intensity of the exercise from the first and second threshold value, respectively.

Example

[0094] This example illustrates, with computer code shown in tables 1 to 5, a practical execution of the invention in a simple manner having a small power consumption.

TABLE-US-00001 TABLE 1 Initializing exemplary inter-beat interval data (values in milliseconds) function sample_fDft % % % dataHere = [920 843 799 816 861 845 845 856 801 759 738 731 735 733 713 ... 709 708 710 719 705 689 699 719 755 740 758]; fPwd = fDft(dataHere);

TABLE-US-00002 TABLE 2 Initialization of variables and classification of inter-beat intervals function fPwd = fDft(d) % % % Here the resolution in time domain is 50 ms. With N = 400 this means % that there is 20 s of data in buffer. Below is the formula of the % discrete Fourier transformation. % % N % X(k) = sum x(n)*exp(j*2*pi*(k1)*(n1)/N), 1 <= k <= N. % n=1 % This formula is used in the implementation below. global sin_n cos_n % Initialise variables. Cos_n and Sin_n are conot anto in real % implementation. F_s = 1/0.050; % 1/(50 ms) N = 400; freq=(0:N1)*(F_s/N); n = 0: (N1); cos_n = cos(2*pi*n/N); sin_n = sin(2*pi*n/N); data = zeros(400,1); % Take the first 20 s of Incoming data in this example and classify the % differences of the consecutive values. If the newest value is greater % than the previous, fill the buffer with value A (here A = 1). Otherwise % the buffered value is B (here B = 0). d_prev = d(1); index_prev = 0; s = 0; for i=1:max(size(d)), B = s + d(i); If s < 20000, index = mod{ floor(s/50), 400 }; if d(i) > d_prev, for k=index_prev+1:index; data(k) = 1; end end index_prev = index; else break; end d_prev = d(i); end

TABLE-US-00003 TABLE 3 Calculating the Fourier transform and determining and outputting the respiration fequency % The guidance to watch the correct frequency range can came front % outside or it can be calculated based on the current incoming data. % Here constant limits of 0 and 30 bpm are used. ii = find(freq*60>0 & freq*60<30); lowerFreqIndex = ii(1) 1 upperFreqIndex = ii(end) 1; fPwd = getPwd(lowerFreqIndex,upperFreqIndex,data); % Calculate the respiration rate. Resolution can be enhanced by % calculating the center of the mass of the power density peak. Here the % location of the highest value is considered to be the respiration rate. [m,iMax] = max(fPwd); fprintf(Respiration rate is %d breaths per minute.\n, 60*(iMax1)*F_s/400);

TABLE-US-00004 TABLE 4 Plotting the curves % Plot the data and the power density function of the diference of that % data subplot (2,1,1);plot(cumsum(d(1:i1))/1000,d(1:i1),x); title(\bf{Inter-beat intervals to be analyzed}); xlabel(Time [sec]); subplot(2,1,2);plot(freq(ii)*60,fPwd(ii)); title(\bf{Power density of the difference signal}); xlabel(Respiration rate [1/m

TABLE-US-00005 TABLE 5 Simplified Fourier transform function function fPwd = getPwd(lowerFreqIndex,upperFreqIndex, d) % % This implementation is valid only for values A=1 and B=0 (See the % general explanation) . Typically the calctiiation load here in this % example is about (upperFreqIndex lowerFreqIndex) * (N/2) i.e. about % 2000 summations (half of the values are zeroes). This is about the % same as using FFT with the same data. The complexity of the FFT is % O(N)=N*log(N), here this is about 2400. In FFT, one has to use, in % general, multiplications, too. Furthermore, no windowing is used here. % Also, fixed point arithmetic can be used easily in this kind of an % implementation. % global sin_n cos_n f=zeros(200,2); for i=0: (max(size(d)) 1), for j=lowerFreqIndex:upperFreqIndex, if d (i+1) ~= 0, indexHere = mod( i*j, 400 ); f(j+1,1) = f(j+1,1) + cos_n(indexHere+1); f(j+1,2) = f(j+1,2) + sin_n(indexHere+1); end end end fPwd = f(:,1).{circumflex over ()}2+f(:,2).{circumflex over ()}2;

[0095] As can be seen from FIG. 3b, the peak value or center of mass of the curve and thus respiration frequency is at the point of 18 respirations per minute.

[0096] While the preferred embodiments of the invention have been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention. Accordingly, it will be intended to