as) United States
`a2) Patent Application Publication 1) Pub. No.: US 2006/0084879 A1
`(43) Pub. Date: Apr. 20, 2006
`
`Nazarian et al.
`
`US 20060084879A1
`
`(54) MOTION CANCELLATION OF OPTICAL
`INPUT SIGNALS FOR PHYSIOLOGICAL
`PULSE MEASUREMENT
`
`(75)
`
`Inventors: Richard A. Nazarian, Excelsior, MN
`(US); Lori E, Lucke, Eagan, MN (US);
`Susan S. Alfini, Champlin, MN (US);
`Mark J. Bina, Shoreview, MN (US):
`Don W. E. Evans, St. Paul, MN (US):
`Paul Harris, Delta (CA); Michael W.
`Geatz, Maple Grove, MN (US)
`
`Correspondence Address:
`OPPENHEIMER WOLFF & DONNELLY LLP
`45 SOUTH SEVENTH STREET, SUITE3300
`MINNEAPOLIS, MN 55402 (US)
`
`(73) Assignee: PulseTracer Technologies Inc., Vancou-
`ver (CA)
`
`(21) Appl. No.:
`
`11/250,011
`
`(22)
`
`Filed:
`
`Oct. 13, 2005
`
`Related U.S. Application Data
`
`(60)
`
`Provisional application No. 60/619,253, filed on Oct.
`15, 2004. Provisional application No. 60/681,397,
`filed on May 16, 2005. Provisional application No.
`60/696,858, filed on Jul. 6, 2005.
`
`Publication Classification
`
`(51)
`
`Int. Cl.
`(2006.01)
`AGIB 5/02
`(52) Bee cpecsssserereyeecanncccessesnncsess: snnttescesanciencsveanecs 600/500
`
`ABSTRACT
`(57)
`A pulse rate sensor that
`includes an accelerometer for
`measuring periodic motion and apiezosensor for detecting
`erratic motion is capable of more accurately determining
`pulse rate by accounting for these types of motion. The pulse
`rate sensor in accordance with the present invention dimin-
`ishes pulse rate signal degradation due to erratic motion
`through a combination of algorithms that control signal
`boosting, waveform refinement and signal noise suppres-
`sion.
`
`
`
`
`
`CONDITIONED OPTICAL
`PULSE SIGNALS
`(PHOTODETECTOR)
`504
`
`
`
`
`
`ACCELEROMETER
`PIEZOELECTRIC
`AND/OR DC
`SENSOR
`
`
`
`
`COMPENSATION
`
` 502
`503
`
`
`
`504
`TIME DOMAIN FILTER
`
`
`FREQUENCY DOMAIN -
`
`FOURIER TRANSFORM
`505
`
`
`
`
`PULSE RATE 507,
`
`FREQUENCY DOMAIN-
`FOURIER Trance
`
` BAND
`REJECT
`FILTER
`
`506
`
`oor
`
`Apple Inc.
`APL 1046
`USS. Patent No. 8,923,941
`FITBIT, Ex. 1046
`
`Apple Inc.
`APL1046
`U.S. Patent No. 8,923,941
`
`001
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 1 of 6
`
`US 2006/0084879 Al
`
`SIGNAL
`
`SAMPLE #
`
`FIG. 1A
`
`NOISE
`
`SAMPLE #
`
`FIG. 1B
`
`SIGNAL + NOISE
`
`SAMPLE #
`
`FIG. 1C
`
`002
`
`FITBIT, Ex. 1046
`
`002
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 2 of 6
`
`US 2006/0084879 Al
`
`003
`
`FITBIT, Ex. 1046
`
`003
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 3 of 6
`
`US 2006/0084879 Al
`
`101
`
`120
`
`FIG. 2B
`
`004
`
`FITBIT, Ex. 1046
`
`004
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 4 of 6
`
`US 2006/0084879 Al
`
`FIG. 3
`
`005
`
`FITBIT, Ex. 1046
`
`005
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 5 of 6
`
`US 2006/0084879 Al
`
`CONDITIONED OPTICAL
`PULSE SIGNALS
`(PHOTODETECTOR)
`501
`
`
`
`PIEZOELECTRIC
`
`SENSOR
`
`
`
` 502
`
`TIME DOMAIN FILTER
`504.
`
`FREQUENCY DOMAIN -
`FOURIER Theater
`
`
`
`
`
`ACCELEROMETER
`ANDIOR DC
`COMPENSATION
`503
`
`
`
`FREQUENCY DOMAIN -
`FOURIER TRANSFORM
`505
`
`
`
`
`
`
`BAND
`REJECT
`FILTER
`
`506
`
`PULSE RATE 507
`
`FIG. 4
`
`006
`
`FITBIT, Ex. 1046
`
`006
`
`FITBIT, Ex. 1046
`
`

`

`Patent Application Publication Apr. 20,2006 Sheet 6 of 6
`
`US 2006/0084879 Al
`
`
`ANALYZE DC
`COMPENSATION
`SIGNAL
`404
`
`
`
`USE FREQUENCY
`ANALYSIS
`ALGORITHMS
`
`408
`
`
`
`
`
`
`
`USE PEAK DETECTION
`ALGORITHMS
`
`410
`
`
`CALCULATE
`PULSE
`
`RATE
`412
`
`FITBIT, Ex. 1046
`
`007
`
`FITBIT, Ex. 1046
`
`

`

`US 2006/0084879 Al
`
`Apr. 20, 2006
`
`MOTION CANCELLATION OF OPTICAL INPUT
`SIGNALS FOR PHYSIOLOGICAL PULSE
`MEASUREMENT
`
`BACKGROUND OF THE INVENTION
`
`1. Field of the Invention
`{0001]
`{0002] The present invention relates generally to the field
`ofsignal processing. Morespecifically, the present invention
`is related to pulse rate monitors capable of providing accu-
`rate measurement and display of a user’s pulse rate during
`times of physical exercise or other activity.
`
`{0003]
`
`2. Background ofthe Related Art
`
`{0004] The accurate measurement of an active person’s
`pulse rate at the wrist
`is complicated due to the artifacts
`produced by body motion such as when the person is
`running or otherwise engaging in physical activity or exer-
`cise. Therefore, pulse rate monitors presently in the market
`utilize chest bands that are worn close to the heart
`to
`minimize the effect of motion produced by exercise. Arti-
`facts produced by body motion are detected by pulse rate
`sensors as “noise” that masks the heart rate pulse signal
`being measured. In order to mitigate the effects of these
`artifacts, it is necessary to filter out and cancel as much of
`the noise signal occurring in the pulse rate frequency band
`as possible while retaining the desired pulsesignal.
`
`{0005] Signals ofinterest are generated by transmitting a
`light source in the near infrared region into the tissue and
`measuring the returned signal
`intensity. Typically two or
`four light emitting diodes (LEDs) are employed with vary-
`ing intensity to establish the optimum optical window. The
`return signal strength will be modulated by the capillary
`blood flow in the tissue and will vary with the physiologic
`pulse of the subject. This is a well understood and estab-
`lished principal that has been applied to pulse monitoring
`equipment for years. Pulse rate sensing taken at locations
`other than close tothe heart, has not been successful because
`of the relatively low signal strength and relatively high
`“noise” content. The low signal strength can beattributed to
`a number of factors including variations in skin and hair
`density, variations in vascularization, and optical alignment.
`Further, the received signal
`includes several components
`which can be generally classified as “noise” when attempt-
`ing to sense pulse rate. These high “noise” levels are in
`addition to classical noise sources present within almostall
`electrical systems and primarily include inherent optical
`noise sources, interfering light sources, and motion artifact.
`
`To illustrate, FIG. 1A depicts a signal ofinterest.
`{0006]
`FIG. 1B depicts noise caused by motion artifact, interfering
`lights sources, randomnoise andthelike. FIG. 1C illustrates
`how the signal of interest is masked by noise due to low
`signal strength, as an example.
`
`{0007] While conventional signal processing techniques
`may be able to reduce “out of band” noise; that is, noise not
`found within the frequency of interest, they are challenged
`to address noise that mimics the signal ofinterest and that is
`non-random. the most commonof which is motion.
`
`{0008] Various pulse rate detection systems are known in
`the art. U.S. Pat. No. 4,338,950 to Barlow, Jr. et al. discloses
`an instrument comprised of wrist-mounted unit, which con-
`tains a piezoelectric transducer for detecting pulse rate and
`
`an accelerometerfor detecting body motion during physical
`activity. The instrument further includes a processor for
`subtracting the body movement component fromthe signal,
`thus yielding the true heart beat signal.
`
`[0009] U.S. Pat. No. 5,431,170 to Mathews discloses a
`device which may be worn on the wrist or hand during
`physical activity, The device contains a light sensor to
`measure pulse and light sensor or accelerometer for mea-
`suring movement. Mathewsdiscloses that a noise cancella-
`tion circuit takes the values from these sensorsto give a true
`pulse signal that is free of pedometry vibration or noise.
`
`[0010] U.S. Pat. No. 5,807,267 to Bryars et al. discloses an
`apparatus which can be combined in a single unit with a
`wrist watch to display the user’s heart pulse rate during
`physical exercise. A primary piezo sensor detects the heart
`rate pulse of a user and a background piezo sensor detects
`the noise from local body motion. Signals from this back-
`ground sensor are digitally subtracted from the primary
`pulse sensor thus allegedly reducing the effects of random
`body noise.
`
`[0011] ULS. Pat. No. 6,099,478 to Aoshimaetal. disclosse
`a pulse wave detecting means comprising an LED, photo
`transistor, or piezoelectric microphone. Body motion detect-
`ing means detect body motion using an acceleration sensor.
`Aoshima et al. disclose that pulse wave extracting means
`subtract the output of the two sensors to give an accurate
`pulse rate. Assignee related U.S. Pat. No. 5,776,070 pro-
`vides similar disclosures.
`
`[0012] U.S. Pat. No. 6,129,676 to Odagiri et al. discloses
`a pulse rate monitor that can be assembled into a wrist
`watchband and used during activities, such as running. The
`wrist watchband contains an acceleration sensor to detect
`action noise and a piezoelectric microphoneto detect pulse.
`When constant motion such as running is detected,
`the
`action noise spectrum is subtracted from the pulse wave
`spectrum, which is depicted in FIGS. 6A-C. Running speed
`and distance may also be obtained. U.S. Pat. Nos. 5,697,374
`and 6,023,662 provide similar disclosures.
`
`[0013] U.S. Pat. No. 6,361,501 B1 to Amanoet al. dis-
`closes a pulse wave diagnosing device formed of device
`main body having a wristwatch structure, and pulse wave
`detection sensor unit. A body motion component remover
`subtracts corrected body motion data from corrected pulse
`wave data. Body motion waves are detected by an accel-
`eration sensor. This device may be incorporated into a
`pedometer.
`
`to
`[0014] U.S. Patent Appln. Publn. 2005/0116820 Al
`Goldreich discloses a wrist mounted device that detects
`pulse rates. The vibration sensor is a piezo ceramic sensor
`that measures movement of the wrist and may include an
`accelerometer. A physiologic sensor detects the blood pres-
`sure pulse rate, and may befiber optic.Ambient sensors may
`also be present.
`
`[0015] Whatever the precise merits, features, and advan-
`tages of the above cited references, none of them achieves
`or fulfills the purposes of the present invention. For these
`reasons it would be desirable to provide an improved device
`and method for accurately measuring pulse rate during
`physical exercise and other activities.
`
`008
`
`FITBIT, Ex. 1046
`
`008
`
`FITBIT, Ex. 1046
`
`

`

`US 2006/0084879 Al
`
`Apr. 20, 2006
`
`SUMMARY OF THE INVENTION
`
`It is therefore an object of the present invention to
`{0016]
`provide a device and method for accurately monitoring and
`detecting pulse rate.
`
`is an object of the present
`it
`{0017] More particularly,
`invention to provide a solution to that allows for adequate
`removal of inherent optical noise sources, interfering ambi-
`ent
`light sources and motion artifact
`in optical signals
`especially under intense physical activity.
`
`It is yet another object ofthe present invention to
`{0018]
`provide a pulse rate monitor that can distinguish between
`noise artifact and an individual’s true pulse from a signal
`representing a composite of pulse and noise artifact as
`illustrated in FIG, 1C,
`
`(0019] These and other objects are accomplished in accor-
`dance with the present invention, a system and method for
`measuring a user’s pulse rate during physical exercise or
`activity, that includes a pulse rate sensor having one or more
`emitters capable of transmitting a light source into body
`tissue, a photo detector for receiving reflected light from the
`body tissue and producing a photo detector output signal
`indicative of the reflected light; an accelerometer for mea-
`suring regular motion of the individual and producing an
`accelerometer output signal indicative of the regular motion:
`a contact type motion sensor for measuring erratic motion of
`the individual and producing a piezo sensor output signal
`indicative of the erratic motion; and a microprocessor for
`receiving the photo detector output signal, the accelerometer
`output signal, and the contact type motion sensor output
`signal, and determining the pulse rate of the individual,
`where the pulse rate is determined by conditioning the photo
`detector output signal and removing portions of the condi-
`tioned photo detector output caused by regular motion and
`erratic motion of the individual.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`{0020] EIG. 1A illustrates a desired optical signal to be
`measured.
`
`(0021] FIG. 1B illustrate noise typically caused by
`motion.
`
`{0022] FIG. 1Cillustrates the combinedoptical signal and
`noise.
`
`{0023] FIG, 2A diagrammatically illustrates a pulse rate
`sensor in accordance with the present invention.
`
`{0024] FIG. 2Billustrates a pulse rate sensor encapsulated
`in a watch module in accordance with the present invention.
`
`FIG.3 illustrates the signal conditioning process
`{0025]
`for an optical signal in accordance with the present inven-
`tion,
`
`FIG,4 illustrates a flowchart depicting the motion
`{0026]
`discrimination process in accordance with the present inven-
`tion.
`
`{0027] FIG. 5 illustrates a flowchart depicting steps fol-
`lowed to calculate pulse rate in accordance with the present
`invention.
`DESCRIPTION OF THE PREFERRED
`EMBODIMENTS
`
`{0028] Whilethis invention is illustrated and described in
`a preferred embodiment, the device may be produced in
`
`many different configurations, forms and materials. There is
`depicted in the drawings, and will hereinafter be described
`in detail, a preferred embodiment of the invention, with the
`understanding that the present disclosure is to be considered
`as an exemplification ofthe principles of the invention and
`the associated functional specifications for its construction
`and is not intended to limit the invention to the embodiment
`illustrated. Those skilled in the art will envision many other
`possible variations within the scope of the present invention.
`
`[0029] FIG. 2A illustrates a pulse rate sensor 101 in
`accordance with the present invention. Emitters 102 (e.g.
`LED, light emitting diodes) transmit a light source in the
`near infrared (IR) region into body tissue and a photo
`detector such as a reflective infra-red sensor 104 that
`receives the reflected light from the tissue. Pulse rate sensor
`101 uses the infrared optical processes from which body
`volume displacement is analyzed to detect and store pulse
`rate data, whichis indicative ofheart rate. Pulse rate sensor
`101 also comprises an accelerometer 108, preferably a
`two-dimensional accelerometer and optionally a three-di-
`mensional accelerometer, that detects periodic or constant
`motion ofthe user, and a contact type motion sensor 106 that
`measures inconsistent or erratic motionofthe user, and other
`movement related sources that effect the optical pulse. The
`contact type motion sensor can be a piezo sensor or other
`types ofsensors capable of measuring erratic motion such as
`vibration. The accelerometer or optical sensor output may
`also be used as input for step counter or pedometer 114.
`Microprocessor 110 performs signal conditioning functions
`on the pulse signals received from the photo detector 104,
`and also samples and filters signals from accelerometer 108
`and piezo sensor 106. Pulse rate detector 112 calculates the
`pulse rate of a user by using conditioned optical pulse
`signals received by microprocessor 110,filtered signals from
`accelerometer 108, and piezo sensor 106. While pulse rate
`detector 112 is shown as being part of microprocessor 110,
`it is understood that these could also be separate compo-
`nents.
`
`[0030] Pulse rate sensor 101 may be encapsulated in a
`watch module 120 as shown in FIG. 2B with an optional
`rechargeable battery in one embodiment ofthe present
`invention. However, other carriers/modules are envisioned
`as usable with the present
`invention such as pendants,
`jewelry, bracelets, patches, music players, etc. which may
`also contain standard watch functions to include time/date
`
`and a stopwatch/timer. These other carriers may be posi-
`tioned anywhere on the arm of anindividual or in the case
`of a pendants around the neck ofan individual. The pulse
`rate sensor in accordance with the present invention may
`also include capabilities for firmware updates through the
`implementation of a software boot loader. Other additions
`may include a communication module such as universal
`serial bus (USB) port or an RFreceiver/transmitter to
`automatically upload data to a web portal. Additionally, the
`pulse rate sensor may include radio frequency identification
`(RFID) to uniquely identify each watch sold to the web
`portal.
`
`‘The pulse rate sensor in accordance with the pre-
`[0031]
`ferred embodiment of the present invention consists of an
`optical emitter including of one or more light emitting
`diodes, an optical receiver, active signal conditioning, an
`accelerometer and a piezo sensor, and a microprocessor. The
`microprocessor controls the active signal conditioning
`
`009
`
`FITBIT, Ex. 1046
`
`009
`
`FITBIT, Ex. 1046
`
`

`

`US 2006/0084879 Al
`
`Apr. 20, 2006
`
`provded from filter 12, Amplifier 28 has a fixed gain and is
`output to the microprocesser 110 at 2. The amplification and
`DC adjustment provided by amplifier 28 moves the signal
`level of its output to the midrange of second operational
`amplifier 30 and allows for a second stage of amplification.
`The amplification gain 32 associated with second amplifier
`30 is programmable to allow for adjust necessary to deal
`with varying pulse strengths. The gain 32 is periodically
`adjusted by the microprocessor 110 to maintain signal
`integrity. A programmable direct current compensation sig-
`nal from output 6 is also applied to second amplifier 30 by
`miroprocessor 110.
`It is during this second stage that the
`sensor signal is significantly amplified. The programmable
`direct current compensation at output 6 is adjusted with
`every sample ofthe first stage gain signal or input 2 of the
`analog to digital convertor. The microprocessor calculates
`the direct current value of the optical pulse signal, and
`applies that value using output 6 as the referenceat the input
`of the amplifier stage 30. This is a fine adjustment and allows
`for the quick recovery of the pulse signal during and after
`motion. As significant amplification is applied to the input
`singal at second amplifier 32,
`the programmable direct
`current compensation further helps to center the signal at the
`amplifier input, and prevents second amplifier 32 from
`saturating during periods of motion by the user. Thus, the
`pulse can be more accurately tracked during periods of
`motion.
`
`[0035] Referring again to FIG. 4, the steps used to control
`the direct current compensation signal 6 and the optical
`emitter supply signal 8 are described below,
`
`1) The optical sensor outputsignal,first amplifer
`[0036]
`output signal, and second amplifer output signalare all
`oversampled at inputs 1, 2, and 3 as shown in FIG, 2
`to generate samples x1(i), x2(/), x3(/) where i repre-
`sents the i'” sample.
`filtered to produce a
`[0037]
`2) The signals are all
`smoother signal at a lower sample rate. More specifi-
`cally, the oversampled signals are processed as follows:
`
`¥
`smoothxl( j) = (1/X)* lw: X = number of points
`
`i a
`
`smoothx2( jf) = (1/X)* D200: X = numberof points
`fai
`x
`smoothx3( j) = (L/X) a3; X = number ofpoints
`Pa
`
`applied to the optical sensor by automatically adjusting the
`light emitting diode output to maintain the optimal signal
`strength, controlling the amplification ofthe received signal,
`and automatically removing the direct current bias of the
`received signal. The conditioned optical sensor output and
`the accelerometer and piezo sensor outputs are sampled as
`input to two different pulse rate calculators, one used when
`there is no motion present and the other used whenthere is
`motion present. The piezo sensor is used to detect erratic
`motion while the accelerometer is used to detect periodic
`motion.
`
`{0032] As further described below, pulse rate signal deg-
`radation due to arm motion and skin vibrations is diminished
`through a combination of algorithms that control signal
`boosting, waveform refinement and signal noise suppres-
`sion.
`
`{0033] Referring to FIG.3, the process to achieve signal
`conditioning of the pulse signal output from photo detector
`10 in accordance with the present inventionis illustrated. A
`pulse signal output from photodetector 10, which may be a
`reflective infra-red. sensor, is filtered through low pass filter
`12 and is provided to input 1 of the microprocessor 110,
`which includes an internal analog to digital convertor 14.
`Naturally, analog to digital converter 14 could also be a
`component separate from microprocessor 110. Through this
`connection, microprocessor 110 is able to monitorthe output
`from photo detector 10 and to determine the appropriate
`intensity for an infra-red light emitting diode 16, which
`transmits optical signals into bodytissue. The intensity of
`infra-red light emitting diode 16 is programmable using an
`output 8 from a digital
`to analog converter 18 within
`microprocessor 110 (again, digital to analog converter 18
`could also be a component separate from microprocessor
`110), As further discussed below, an internal closed loop
`control function within the microprocessor 110 maintains
`the proper intensity of the infra-red light emitting diode 16
`using the feedback from the photodector 10 at input 1 to
`control output 8. Using this control function, the output
`signal from photodetector 10 is in an appropriate range for
`direct current compensation and other related functions.
`Further, this intensity control allows the microprocessor 110
`to periodically adjust the system to account for different
`environmental and physiological conditions, including vary-
`ing ambientlight levels, long term blood flow changes and
`varying responses from user to user. The pulse signal
`transmitted from photodetector 10 is filtered through a
`lowpassfilter 12 of approximately 10 Hz to reduce inherent
`noise.
`
`{0034] The pulse signal from photodetector 10 generally
`includes a large direct current component that represents
`gross blood flow, and a small alternating current component
`[0038] The filtered sample signal (smoothx1(j)) is used
`that represents true pulse. Because it is desirable to accu-
`for the automatic optical emitter control at signal 8. The
`rately measure true pulse, active signal conditioning includ-
`filtered first amplifier output signal (smoothx2(j)) and
`ing additional amplification ofthe alternating current com-
`
`the output©signalfiltered second amplifier
`
`
`
`ponent and compensationofthe direct current componentis
`(smoothx3(j)) are used for automatic direct current
`necessary. Compensation of the direct current component
`compensation at signal 6 and for pulse detection,
`occurs in two stages. The first stage is accomplished using
`respectively.
`an amplifier 28 to achieve a fixed signal gain. The output
`from this first stage is provided to microprocessor 110 at
`input 2. The second stage is the programmable gain stage
`utilzing a second amplifier 30 with output to microprocessor
`110 shownat3. A fixed direct current compensation voltage,
`shown at output 7, is subtracted from thefiltered pulse signal
`
`3) Automatic control of optical emitter 16 is
`[0039]
`based on the filtered second amplifier output signal at
`input 1, and is applied as follows on a periodic basis.
`[0040]
`a)
`If
`smoothl(j)>expected
`range,
`then
`decrease output at signal 8.
`
`010
`
`FITBIT, Ex. 1046
`
`010
`
`FITBIT, Ex. 1046
`
`

`

`US 2006/0084879 Al
`
`Apr. 20, 2006
`
`smoothl(j)<expected
`If
`b)
`[0041]
`increase output at signal 8.
`
`range,
`
`then
`
`where a non-zero value indicates a peak in the
`[0055]
`filtered signal and a zero indicates no peak.
`
`4) Direct current compensation is applied at
`[0042]
`output 6 as follows at the update rate of the smoothing
`filters as follows.
`
`a) If smooth3(j)>(max value of smooth3(j)-
`[0043]
`threshold),
`then Direct current compensations)=
`smooth2(j)
`
`value of
`smooth3(j)<(min
`if
`b) Else
`[0044]
`smooth3(j)+threshold), then Direct current compen-
`sations)=smooth2(j)
`
`c) Else direct current compensation(j)=direct
`[0045]
`current compensation(j—1).
`
`{0046] During periods of motion, the signal applied for
`direct current compensationat output 6 by the microproces-
`sor is correlated to the motion of the user. As as result, the
`direct current compensation can also be used as an indicator
`of motion as shown in FIG, 5, The direct current compen-
`sation signal is first analyzed in step 404, and this analysis
`is used to detect motion in step 406. If the direct current
`compensation is stable or lacks change this indicates a lack
`of motion and then the peak detection algorithm 410
`described below is used to calculate the pulse rate 412. This
`allows for a fast recovery of the pulse rate after periods of
`motion. If the direct current compensation is correcting the
`signal, this indicates that motionis present and the frequency
`based algorithm408 described below is used to calculate the
`pulse rate 412.
`
`{0047] The direct current compensation(j) signal is used
`for motion discrimination in block 406 and determines
`whether peak detection algorithms can be used, for example
`during periods of no motion, or if frequency analysis is
`necessary for example during periods of motion. Stated
`alternatively, the direct current compensation(j) signal
`is
`used to determine whether to apply peak detection algo-
`rithms in block 408 or frequency analysis in block 410 to
`determinethe pulse rate, More specifically, if direct current
`compensation(j) has not changed for the past 3x(1/pulse
`rate) seconds during which pulse rate is being measured,
`then the peak detection algorithm described belowis used.
`If direct current compensations) has changed during the past
`3x(1/pulse rate) seconds during which pulse rate is being
`measured the frequency analysis algorithm described below
`is used.
`Peak detection is calculated as follows:
`
`1) A first derivative calculated as diffl(i)=
`[0048]
`smoothx2(i)-smoothx2(i-1) is taken over the filtered
`signal.
`
`2) A second derivative, diff2(i)=dilTl (i)-diffl (i-
`[0049]
`1), is calculated or computed from the first derivative.
`
`3) Peak detection is analyzed using the first and
`[0050]
`second derivatives to find the peaks within the filtered
`signal.
`
`[0051]
`
`If diffl (i)=0 and diff2(i)<0 then
`
`[0052]
`
`peak(ij=i
`
`[0053]
`
`If diffl (i) does not=0 and diff2(ij>0 then
`
`[0054]
`
`peak(i)=0
`
`4) The instantaneous pulse rate is calculated in
`[0056]
`block 412 as the difference between peaks inthe filtered
`signal.
`Instantaneous pulse rate(ij=sample rate (in hz)/Period-
`(peak(i)-previous non-zero(peak(i)))*60 sec/minute.
`
`the pulse detection
`[0057] Referring back to FIG. 3,
`system also includes an accelerometer 20 and piezo sensor
`22. Signals from accelerometer 20 and piezo sensor 22 are
`filtered through lowpass filters 24, 26 at analog to digital
`inputs 5 and 4, respectively before being sampled. The
`accelerometer 20 is used in the motion mitigation fast
`fourier transform (FFT) algorithms and the piezo sensor is
`an indicator of erratic motion as shown in FIG. 4, Generally
`speaking, the accelerometer 20 and piezo sensor 22 are used
`to remove the motion artifacts from the optical sensor output
`present during motion of the user. When substantial erratic
`or momentary motion is detected using the piezo sensor 22,
`the information is used to filter the optical sensor output in
`the time domain so that the signal is not used for pulse rate
`calculations during the erratic motion 504. Additional fil-
`tering may also be appliedto the instantaneous pulse rate to
`provide a more stable pulse rate output. Periodic motion is
`detected by analyzing the accelerometer output signal in the
`frequency domain 505 and this information is used to
`generate a band reject filter which is applied to the optical
`sensor output in the frequency domain 506.
`
`If it is determined that motion is present, a fre-
`[0058]
`quency algorithm as shownin FIG. 4 is used. The signals
`that undergo frequency analysis are the smoothed first
`amplifier output signal (smoothx2) and the smoothed second
`amplifier output signal (smoothx3) in step 501. Signals from
`the accelerometer 20 and direct current compensation are
`also provided in step 503 so that further frequency analysis
`can be done. Frequency analysis may includeall or a subset
`ofthe foregoing signals. The frequency bins will be ana-
`lyzed to determine whether motion frequencies are present.
`Again, motion frequencies are determined by the acceler-
`ometer output or the direct current compensation signal. The
`motion frequencies are then removed fromthe sensor sig-
`nals, smoothx? and smoothx3 to allow for discrimination of
`the pulse rate as follows:
`
`to
`ten second window filter
`a) Apply a
`[0059]
`smoothx2, smoothx3, accelerometer, and direct current
`compensation
`
`b) Removedata as necessary if erratic motionis
`[0060]
`detected in the time domain at step 504
`
`c¢) Apply frequency translationto all or a subset
`[0061]
`of windowed smoothx2, smoothx3, accelerometer, and
`direct current compensation in step 505;
`
`d) Bandpassfilter all signals in the range of 0.5
`[0062]
`Hz to 4 Hz at step 506:
`
`smoothx2,
`in
`Identify frequency peaks
`e)
`[0063]
`smoothx3, accelerometer, and DC compensation;
`
`f) Using the frequency peaks from the acceler-
`[0064]
`ometer of direct current compensation to determine the
`frequencies to reject, using band reject filter,
`from
`smoothx2 and smoothx3 at step 506;
`
`011
`
`FITBIT, Ex. 1046
`
`011
`
`FITBIT, Ex. 1046
`
`

`

`US 2006/0084879 Al
`
`Apr. 20, 2006
`
`g) Remove peaks from smoothx3 that are not
`[0065]
`commonwith those in smoothx2; and
`
`h) Pulse rate equals the lowest frequency peak in
`[0066]
`smoothx3 (block 507)
`
`If no peaks are remaining, the pulse rate is not updated.
`{0067]
`tis anticipated that the present invention may also
`be used as a pedometer. To use the pulse rate monitor wrist
`watch as a pedometer, the data from the two-axis acceler-
`ometeris filtered. If the motiondetectedis a “step,”1.e. foot
`motion, the filter applied to the Y-axis accelerometer data
`should be one-half that applied to the X-axis accelerometer
`data. Peak detection, i.e. detecting the intervals between the
`peaks of the signals from the X and Y accelerometers, is
`used to determine the fundamental frequencies of motion.
`Peaks are detected using the first and second derivative
`methods described above.
`
`{0068] The pulserate is output using a digital signal which
`transitions at the rate of the pulse rate. The pedometer is
`output using a digital signal which transitions once for each
`step detected.
`
`{0069] A system and methodforthe effective implemen-
`tation of motion cancellation of optical
`input signals for
`physiological pulse measurement
`in accordance with the
`present invention has been disclosed herein. While various
`preferred embodiments have been shown and described,
`it
`will be understood that
`there is no intent
`to limit
`the
`invention by such disclosure, but rather,
`it
`is intended to
`cover all modifications and alternate constructions falling
`within the spirit and scope of the invention, as defined in the
`appended claims.
`Weclaim:
`
`1. A pulse rate sensor for measuring a physiological
`parameter of an individual comprising:
`
`an emitter for transmitting a light source into body tissue:
`a photo detector for receiving reflected light from said
`body tissue and producing a photo detector output
`signal indicative of reflected light;
`
`an accelerometer for measuring regular motion of the
`individual and producing an accelerometer output sig-
`nal indicative of the regular motion;
`a contact type motion sensor for measuring erratic motion
`of the user and producing a contact type motion sensor
`output signal indicative ofthe erratic motion; and
`
`a microprocessor for receiving the photo detector output
`signal, the accelerometer output signal, and the contact
`type motion sensor output signal, and determining the
`pulse rate ofthe individual, wherein the pulse rate is
`determined by conditioning the photo detector output
`signal and removing portions ofthe conditioned photo
`detector output caused by regular motion and erratic
`motion of the individual.
`2. The pulse rate sensor of claim 1 wherein said trans-
`mitted light source lies in the near infrared region.
`3. The pulse rate sensor of claim | wherein said emitter
`comprises a plurality of light emitting diodes.
`4. The pulse rate sensor of claim 1 wherein said photo
`detector is a reflective infra-red sensor.
`§. The pulse rate sensor of claim 1 wherein said acceler-
`ometer is a two-dimensional accelerometer.
`
`6. The pulse rate sensor of claim 1 wherein said sensor is
`encapsulated in a watch module for wrist-based sensing.
`7. The pulse rate sensor of claim 6 wherein said watch
`module includes a universal serial bus port to automatically
`upload data, a radio frequency identification tag, to uniquely
`identify said watch module, wherein said watch module is
`further capable of receiving firmware updates.
`8. The pulse rate sensor of claim 1 wherein said contact
`type motion sensor comprises a piezo sensor and said
`contact type motion sensor output signal comprises a piezo
`sensor output signal.
`9. A method of measuring a user pulse rate during
`physical exercise or activity, said method comprising:
`
`transmitting a light source into body tissue;
`
`the
`from