DETECTING ATRIAL FIBRILLATION, METHOD OF AND APPARATUS FOR

BACKGROUND OF THE INVENTION

Field of the invention

[0001] The present invention generally relates to a method of and an apparatus for detecting atrial fibrillation by monitoring and analyzing pulse beats using an algorithm that can exclude many rhythm abnormalities that are not atrial fibrillation.

Discussion of the related art

[0002] The heart is the major muscle that functions as the primary pump for blood flow throughout the body. The heart contains two upper chambers called atria and two lower chambers called ventricles. The right atrium receives oxygen-depleted blood while the left atrium receives blood enriched with oxygen from the lungs. When the atria are full, the outlet valves within the heart open and the atria squeeze blood into the ventricles. The right ventricle then pumps oxygen-depleted blood to the lungs while the left ventricle pumps oxygen-enriched blood to all parts of the body. In this fashion, the heart functions primarily as a double sided pump. [0003] The heart's internal pacemaker, known as the sinus node, signals the start of each heart beat. This signal originates in the right atrium in the sinoatrial node and travels simultaneously to the left atrium and down to the interatrial septum to the atrioventricular node. The cycle of electrical stimulation that normally occurs is referred to as normal sinus rhythm. The contraction of the ventricles will be referred to as the heart beats. [0004] Many rhythm abnormalities may be present. Atrial fibrillation is one rhythm abnormality in which the atria do not contract normally. Instead, there is a continuously varying pattern of electrical activation of the atria resulting in a rapid highly irregular pattern of impulses reaching the atrioventricular node. The atrioventricular node acts as a filter and allows a reduced number of these impulses to reach the ventricles which results in a highly irregular heartbeat pattern. This irregular pattern has been shown in

previous studies to be a random pattern (Bootsma et al: Analysis of R-R Intervals in Patients with Atrial Fibrillation at Rest and During Exercise. Circulation 41 : 783, 1970). Whenever the term "irregular' is used in this application it refers to this random pattern of beats found almost exclusively in atrial fibrillation.

[0005] Atrial fibrillation is one of the most common arrhythmias requiring medical attention. Atrial fibrillation may be caused by a number of heart conditions, such as angina, myocardial infarction, heart valve abnormalities, and high blood pressure. These conditions may stretch or scar the atria, thereby causing irregularities in the heart system. Atrial fibrillation may also accompany lung problems or thyroid gland disorders and is also associated with significant morbidity and possible mortality. All persons, young and old, female or male, including the visually and/or sight impaired, may experience atrial fibrillation.

[0006] Atrial fibrillation may occur intermittently or chronically. The most serious complication of atrial fibrillation is formation of a blood clot in the left atrium which may result in a stroke. Many people who develop atrial fibrillation, however, are unaware of their abnormal rhythm. Some in the medical profession have, therefore, advocated self screening of the pulse to detect for the possible occurrence of atrial fibrillation. The literature, however, is generally limited to disclosing instructions for manually taking one's pulse accompanied with additional descriptive information.

[0007] The reason for using the pulse to detect atrial fibrillation is that the pulse usually corresponds to the heartbeat. The contraction of the left ventricle ejects blood from the heart into the aorta and the resulting pressure wave is detected as a pulse in the arterial system. However, when atrial fibrillation is present, the amount of time between beats varies irregularly. [0008] With a longer time interval between beats, there is more time to fill the ventricles with blood and more blood is ejected by the ventricle in the beat following this long interval. This larger volume of blood in the aorta results in a higher systolic pressure for that beat.

[0009] Conversely, when the time interval between beats is short, there is less time for ventricular filling and the volume of blood ejected in the beat following the short time interval is less. This results in a lower systolic pressure for that beat. In some cases, the time interval between beats is so short that the systolic pressure of the following beat is so low that it cannot be palpated as a pulse. A ventricular contraction that cannot be palpated as a pulse in the arterial system results in what is called a "pulse deficit." This is very common in atrial fibrillation. This pulse deficit means that an irregular pattern of heartbeats in the ventricle may result in a less irregular pulse beat pattern since the shortest intervals between heartbeats may not be detected in the pulse. Therefore, any method used to determine the presence of atrial fibrillation by analyzing the time intervals between beats in the ventricles may not be valid when applied to the pulse beats.

[0010] An article by Bert K. Bootsma, Adriann J. Hoelen, Jan Strackee and Frits L. Meijler, entitled Analysis of R-R Intervals in Patients with Atrial Fibrillation at Rest and During Exercise, Circulation, Volume XLI, May 1970 describes an analysis of the time intervals between ventricular contractions using the electrocardiogram. The article evaluates the standard deviation divided by the mean of the time intervals between ventricular beats in normal subjects and in those with atrial fibrillation. The article finds that atrial fibrillation can be accurately differentiated from normal sinus rhythm using this formula. However, this was based on ventricular contractions determined by the electrocardiogram and was not applied to the pulse beat intervals. [0011] Due to the presence of a pulse deficit in atrial fibrillation, results based on ventricular contractions determined by the electrocardiogram may not apply to time intervals determined from analyzing the pulse. Furthermore, the extent of the pulse deficit depends on the method used to determine the pulse beats. A method which detects only pulse beats with high systolic pressures will detect fewer pulse beats compared to a more sensitive method. The more sensitive techniques may be better for detecting more pulse beats but they may also give more false positive readings.

[0012] For example, with a photoplethysmograph using a finger probe with a light source and a photoelectric detector, when the sensitivity of the device is increased, the slightest finger movement is detected as a pulse beat. This device at the highest sensitivity setting detects an irregular pulse in those with normal sinus rhythm due to random noise from finger movement. At the highest sensitivity setting, this device would not be useful to detect atrial fibrillation in the home setting. At the lowest sensitivity setting, very few pulse beats would even be detected. Therefore, any device and algorithm which uses the pulse to detect atrial fibrillation must be designed specifically for the purpose of detecting atrial fibrillation.

[0013] Other rhythm abnormalities that are not associated with an increased risk of stroke are found very commonly. For example, premature atrial contractions or premature ventricular contractions are found in many people, even those with no heart conditions. These rhythm abnormalities are intermittent but they are found much more commonly than atrial fibrillation and generally do not require treatment. Therefore, a device that is designed to detect atrial fibrillation should include an algorithm that would not detect premature beats as an irregular rhythm. If a device designed to detect atrial fibrillation does not include an algorithm to reduce the detection of premature beats, the number of false positive readings would be very high and the device would be of very limited value. Those who might use such a device on a regular basis to detect asymptomatic atrial fibrillation would quickly stop using this device if the vast majority of abnormal readings turn out to be due to premature beats and not atrial fibrillation.

[0014] US Patent No. 6,485,429 describes a method and apparatus that can detect arrhythmia using an oscillometric blood pressure monitor. This patent discloses a method for detecting multiple rhythm abnormalities including supraventricular premature contractions, ventricular premature contractions, atrial fibrillation, paroxysmal supraventricular tachycardia, sinus tachycardia, and ventricular bradycardia. A study performed by the inventor of the patent using a commercially manufactured embodiment of the patent found that the device only detected 66.6% of patients with atrial fibrillation

while detecting 85.7% of premature ventricular contractions (Forstner KW, American Society of Hypertension 16th Annual Meeting 2003, page 25). Premature ventricular contractions are found much more commonly than atrial fibrillation. Therefore a device that has a very high rate of detecting premature ventricular contractions and a significantly lower rate of detecting atrial fibrillation would not be a very useful device if the goal is to detect atrial fibrillation exclusively.

[0015] Other patents that can detect rhythm abnormalities would also be subject to very high false positive rates if they are used to detect atrial fibrillation. US Patent No. 6,095,984 describes an embodiment (section 1-2- 1) where a pulse variation over 0.5% would be detected as abnormal. The most common rhythm abnormality, sinus arrhythmia, by definition (Braunwald, E. Heart Disease A textbook of Cardiovascular Medicine 1992, p 674) has a heart rate that varies by more than 10%. US Patent No. 6,095,984 did not describe methods and apparatus to look for atrial fibrillation and would not be useful for home monitoring of atrial fibrillation, because the algorithm can result in multiple sources of false readings and the apparatus was not specifically set for optimal detection of pulse beats in atrial fibrillation. [0016] An algorithm designed to detect atrial fibrillation by the irregularity of the pulse beat intervals should be designed to reduce the effects of premature beats. One method of reducing the effect of premature beats is to limit the number of beats used in determining the irregularity of the pulse intervals. For example, if premature beats occur on average every twenty beats, then limiting the analysis to only ten beats would reduce the likelihood of a premature beat occurring during the period being analyzed. This method was used in a trial and published in Wiesel et al, PACE, 27:639- 643 (2004). In this study, the rhythm of 450 outpatients of which 54 were documented to have atrial fibrillation was analyzed. The study used an algorithm that calculated the irregularity index, defined as the standard deviation of the time intervals between beats divided by the mean of the time intervals between beats, and analyzed only the last ten beats recorded by an automatic oscillometric blood pressure monitor. All recordings with an

irregularity index greater than 0.06 were considered to be irregular. This study found that 100% of patients with atrial fibrillation were detected while 16% of patients without atrial fibrillation were also found to be irregular (sensitivity for detecting atrial fibrillation 100%, specificity 84%). This is a marked improvement in the rate of detecting atrial fibrillation over the method used by Forstner.

[0017] Though the sensitivity for detecting atrial fibrillation using the irregularity index applied to the last ten beats recorded by an automatic blood pressure monitor is excellent, there is room for improvement in specificity. Premature beats are common rhythm abnormalities that can reduce the specificity of the irregularity index. For example, a ventricular premature beat occurs earlier than the normal beats but results in a pause following the beat. A premature beat usually results in a lower pulse pressure than the normal beats. This premature beat with its lower pulse pressure would often not be recorded by the blood pressure monitor at all. As a result there would be a pause equal to the time interval between two normal beats as a result of this premature beat. For a patient with a heart rate of 60 beats per minute, the pulse interval would be one second between normal beats. A premature ventricular beat would result in a pulse interval of two seconds as measured by a blood pressure cuff. The irregularity index for nine time intervals between beats of one second and one interval of two seconds due to one premature ventricular beat would be equal to 0.29, well above the threshold value of 0.06. If the premature beat occurs 0.5 seconds following a normal beat and it is detected by the blood pressure monitor then there will be a short interval followed by a long interval. With eight intervals of one second, one interval of 0.5 second and one interval of 1.5 second the irregularity index would be 0.24 still very abnormal.

[0018] US Patent 6,519,490 issued to Joseph Wiesel discloses a method to reduce the impact of premature beats by deleting only those beats that occur at shorter intervals than the mean beat to beat interval. This method was found to improve the specificity of an automatic blood monitor. An algorithm that can recognize when a premature beat occurs and eliminates

both the short intervals and the long intervals associated with these beats could further reduce the false positive rate and improve the specificity for the blood pressure monitors used to detect atrial fibrillation. Some patients have very frequent premature beats occurring every other beat. In that case, all the time intervals are either shorter or longer than the mean of the time intervals. If all the beats are eliminated because they fall outside the upper and lower threshold values around the mean time interval then that pattern is not typical of atrial fibrillation and will be considered a regular rhythm. Atrial fibrillation has continuously varying time intervals around the mean time interval such that there are usually at least a few time intervals near the mean time interval that will not be eliminated by this algorithm.

[0019] Occasionally, premature beats will occur frequently with varying intervals. For example, if the normal beats occur at time intervals of one second, there may be premature beats occurring at 0.5 second, 0.75 second and 0.85 second all in the same patient during the period of time recorded by the blood pressure monitor. If we eliminate some of these intervals because they exceed a threshold value, other intervals may still be present that can cause a high irregularity index. Another algorithm can be used to reduce the false positive rate for those patients with frequent premature beats. In atrial fibrillation there are no normal beats and having more than half of the time intervals almost exactly equal in length during the 10 to 40 seconds of a typical blood pressure monitor reading is unlikely. On the other hand, even with frequent premature beats, more than half the beats are usually normal beats. Thus, if we find that most of the time intervals are almost equal during one blood pressure reading, it is much more likely that this is not atrial fibrillation.

[0020] What is needed is a home monitoring method and apparatus to detect the possible presence of atrial fibrillation and communicate this condition to the user so that the user is alerted to consult a medical practitioner for further testing and/or treatment.

[0021] What is also needed is a method that can differentiate atrial fibrillation from a normal pulse pattern and from common heart rhythm

abnormalities that are not of significant risk such as sinus arrhythmia, atrial premature beats and ventricular premature beats.

[0022] What is further needed is a method of and an apparatus for detecting irregular pulse rhythms during a time period and storing this information so that comparisons may be made with the pulse rate rhythms at later times.

[0023] What is further needed is a noninvasive and relatively simple method and apparatus that monitors pulse rate irregularities to detect atrial fibrillation, and that is suitable for use of all ages, and by the hearing and/or visually impaired and that is relatively easy to use.

[0024] What is still further needed is a monitoring method and apparatus that detects the presence of irregular pulse beats and then displays and stores: i) the number of irregular pulse beats during a pre-selected time interval; and ii) the duration of time between beats during selected intervals.

[0025] Yet another need is for a monitoring method and apparatus that determine whether or not a pulse beat pattern is irregular based on algorithmic or heuristic operations performed on selected pulse beat data.

[0026] What is needed is a method and apparatus for detecting the presence of atrial fibrillation by detecting an irregular pattern of pulses using a sphygmomanometer.

[0027] What is needed is a method and apparatus for detecting the presence of atrial fibrillation by detecting an irregular pattern of pulses using a plethysmograph such as finger probe with a light source and photodetector.

BRIEF SUMMARY OF THE INVENTION

[0028] The present invention provides a method and apparatus that determines the presence of atrial fibrillation by (i) detecting the pulse beat intervals over a short time period to determine whether the intervals form a random, irregular pattern; (ii) determining whether this pulse beat pattern indicates possible atrial fibrillation and then; (iii) communicating this information to the user so that a medical practitioner may be consulted for further testing and/or treatment. The present invention also provides a method

of and an apparatus for detecting irregular pulse rhythms during a time period and storing this information for comparison with the pulse rhythm at later time periods. The present invention may also detect patterns over multiple time periods and compare the patterns over various time periods. [0029] The present invention further provides a noninvasive method of and an apparatus for monitoring the irregularity of the pulse beat pattern to detect atrial fibrillation. The invention may store and display information such as the number of irregular pulse beats during pre-selected time intervals and the duration of time between beats during the selected intervals. The invention also determines the presence of an irregular pulse beat pattern via algorithmic or heuristic operations performed on the relevant data. [0030] Pulse beats may be monitored by use of an inflatable cuff wrapped around a person's appendage, such as an arm, which detects the pulse beats by either oscillometric or auscultatory means. The time intervals between pulse beats can be determined during cuff deflation or while the cuff is inflated at a fixed pressure.

[0031] The present invention allows for the elimination of at least one or more of the pulse beat intervals detected during deflation of a blood pressure cuff if the time interval of the pulse beat is less than a minimal threshold value or greater than a maximum threshold value. This is necessary to prevent the detection of premature beats and to reduce the false positive rates for detecting atrial fibrillation.

[0032] The present invention allows for the detection of a normal rhythm if the percent of pulse intervals that fall within a narrow time period exceeds a threshold percent in order to detect normal sinus beats that may be interspersed with multiple other abnormal beats with varying pulse beat intervals.

[0033] Pulse beats may also be monitored through changes in light transmitted through various body appendages. Each pulse beat changes the light transmission through a location on the appendage. The change in the light transmission corresponds to a pulse beat and the time intervals between pulse beats may be determined.

[0034] Pulse beats may be monitored using other plethysmographic devices, ultrasound devices which measure arterial motion with each pulse beat, ultrasound doppler devices which detect blood flow within an artery or devices that rely on localized compression of the artery to detect the presence of a pulse beat. Using any of these techniques the time intervals between pulse beats can be determined.

[0035] A monitoring method of the present invention includes detecting irregular pulse beats, analyzing the irregularity based on one or more predetermined factors, and communicating this information to a user such as via a screen display, a paper printout, a tone, or auditory, vibratory or other sensory communication.

[0036] The invention may utilize algorithmic or heuristic techniques to determine whether the irregular pulse beats signal the possible presence of atrial fibrillation.

[0037] Other features and advantages of the present invention will become apparent from the following detailed description of the invention with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Fig. 1 is a flow chart of the algorithm (Algorithm I) for a first method of the present invention to determine possible atrial fibrillation.

[0039] Fig. 2 is a flow chart of the algorithm (Algorithm II) for a second method of the present invention to determine possible atrial fibrillation.

[0040] Fig. 3 is a graph showing time intervals for regular sinus rhythm.

[0041] Fig. 4 is a graph showing time intervals for atrial fibrillation.

[0042] Fig. 5 is a graph showing measured time intervals prior to application of Method I.

[0043] Fig. 6 is a graph showing measured time intervals after application of Method I.

[0044] Fig. 7 is a graph showing sinus rhythm with frequent and varying premature beats and pauses.

[0045] Fig. 8 is a graph showing time intervals for the rhythm shown in

Fig. 7 and sorted as in Method II.

[0046] Fig. 9 is a graph showing time intervals as sorted in Method Il and initiating atrial fibrillation.

DETAILED DESCRIPTION OF THE INVENTION

[0047] One embodiment of the invention uses pulse beats that are detected using an inflatable cuff device. The inflatable cuff device may be a known apparatus used to measure blood pressure using oscillometric or auscultatory means.

[0048] The inflatable cuff device is placed around an appendage such as an arm and inflated above systolic pressure. While the cuff is deflated, the pulse beats are detected. The cuff deflation may be stopped and the cuff may remain at a fixed pressure to allow for monitoring of the pulse beats during a constant cuff pressure. The time of each pulse beat is delivered to a processor, which includes instructions that carry out the method described above.

[0049] Further, the processor stores the time of each pulse beat, the intervals between pulse beats and other information in a memory. The memory may include RAM or other device memory or include a hard disc, a floppy disk or other memory devices. The processor may comprise a microprocessor, and applications specific integrated circuit (ASIC), a programmable logic array (PLA) or reduced instruction set chip (RISC). [0050] The processor determines from the pulse beat intervals if the pattern suggests atrial fibrillation or not. The processor then delivers the results to a printer, a display, a vibration generator, and/or an auditory generator, etc. which may include an indication that the pulse beat pattern is regular, irregular, in possible atrial fibrillation, or that a physician should be contacted. Other information, such as the pulse rate, may also be displayed. [0051] This embodiment with an automatic sphygmomanometer that uses the oscillometric method for detecting the pulse beats and blood pressure was developed. To account for the pulse deficit at higher cuff

pressures, and to account for possible respiratory variation in the systolic pressures of normal beats, and to reduce the influence of premature beats, an algorithm was developed that analyzed only the last ten beats detected during cuff deflation. The mean and standard deviation of these last ten beats was calculated and the quotient of the standard deviation over the mean was determined.

[0052] This was compared to a threshold value of 0.06. If the quotient was greater than the threshold value then the pulse pattern was determined to be irregular. As mentioned previously, this method was used in a trial and published in Wiesel et al, PACE, 27:639-643 (2004) using 450 outpatients of which 54 were documented to have atrial fibrillation. The sensitivity using this method with one blood pressure reading was 100% with a specificity of 84% for detecting atrial fibrillation.

[0053] All pulse time intervals recorded by the device are used to determine a mean time interval. For those readings found to be irregular one of two methods may be used to improve the specificity for atrial fibrillation. (See Figs. 3 and 4).

[0054] In a first preferred method of the present invention (Fig. 1), all recorded time intervals recorded by the automatic sphygmomanometer are used to determine a mean time interval. (See Fig. 5). All intervals more than about 25% above the mean or less than about 25% below the mean are eliminated. (See Fig. 6). The new mean and standard deviation are, then calculated from the remaining time intervals. A new irregularity index is calculated as the ratio of the new standard deviation to the new mean. If the irregularity index is less than about 0.066 then the rhythm is considered regular. If the irregularity index is equal to or greater than about 0.066 then the rhythm is considered irregular and likely to be atrial fibrillation. If there are no time intervals remaining when those above or below the threshold values are eliminated then the pattern is not consistent with atrial fibrillation and is typical of premature beats occurring every other beat. The rhythm is then considered regular. Fig. 1 is a flow chart schematic of the method described herein. Using this algorithm on the 247 irregular readings in the previously

mentioned study (Wiesel et al, PACE, 27:639-643 (2004) resulted in all patients with atrial fibrillation continuing to be detected as irregular but now approximately half of the previously irregular patients not in atrial fibrillation were reclassified as regular. This improved the specificity to around 92% while keeping the sensitivity at 100%.

[0055] In a second preferred method of the present invention (Fig. 2), all pulse time intervals recorded by the device are sorted from shortest interval to longest interval. Specifically, all the time intervals recorded by the sphygmomanometer are sorted from the shortest interval to longest interval. (See Figs. 7 and 8). Beginning with the shortest interval, the first three time intervals are used to calculate a mean and standard deviation. (See Fig. 6). The ratio of the standard deviation to the mean is the irregularity index for these three intervals. If this irregularity index is less than about 0.01 that is used as an indication that these three intervals are almost equal. The time interval following the shortest interval is then used to calculate an irregularity index for the next three beats. This is continued for the next time interval until the irregularity index for the three longest time intervals is calculated. If about 50% or more of the irregularity indices are less than about 0.01 then the rhythm is considered regular. (See Fig. 9 for irregularity). When this algorithm was applied to the 247 irregular readings from the study mentioned above, all the patients with atrial fibrillation were still detected as irregular while two-thirds of the previously irregular readings are now classified as regular. This results in a specificity of about 94% with a sensitivity remaining at about 100%. Fig. 2 is a flow chart schematic of the method described herein.

[0056] There is another embodiment of the invention in which the pulse rate is monitored through changes in light transmitted through a body appendage, such as a finger. Light is transmitted by a light source through the finger, or other appendage of an individual, and is received by a detector, which measures the change in light transmitted through the appendage to detect a pulse beat. The detector may comprise a conventional pulse

measurement device. The detector delivers the time of each measured pulse to the processor, which carries out the operations described above. A light power supply detector sensitivity adjuster may be used to adjust the sensitivity of the detector in a known manner with respect to detecting the light from the light source. To detect the pulse beat, the finger or other appendage is placed between the light source and the detector and the light source is activated to shine light through the finger to reach the detector. [0057] Advantageously, the invention provides a method and apparatus that easily detect the presence of irregular heartbeats from a plurality of heartbeats, pulses or other measurements.

[0058] As a further advantage, the invention differentiates atrial fibrillation from non-atrial fibrillation rhythms including normal and other abnormal rhythms.

[0059] A still further advantage is that the invention provides relatively simple, non-invasive home monitoring.

[0060] An apparatus and method is encompassed by the invention to include means for sorting the pulse beat intervals by length from shortest to longest or longest to shortest; means for calculating an irregularity index U for the first n time intervals, the irregularity index Ii being a quotient of a mean and a standard deviation of a first n time intervals; means for calculating an irregularity index I ₂ for the n time intervals beginning with the second time interval, the irregularity index I ₂ being a quotient of a mean and a standard deviation of n time intervals beginning with the second time interval; means for continuing to calculate an irregularity index l _m where m ranges from one until N-(n-1), where N is the total number of pulse intervals; means for determining what percent, P, of the said irregularity indices U, I ₂ ,...IN-( _π -I), are less than a threshold value T to determine possible atrial fibrillation in that if P exceeds a cutoff value P _cu t _off , the rhythm is not atrial fibrillation and if P is less than or equal to Pcu _toff , the rhythm is atrial fibrillation.

[0061] Although the present invention has been described in relation to particular embodiments thereof, many other variations and modifications and other uses may become apparent to those skilled in the art. It is preferred,

therefore, that the present invention be limited not by this specific disclosure herein, but only by the appended claims.