A system for assessing cardiovascular disease risk

Abstract

The application discloses a method for evaluating heart biological rhythm or cardiovascular disease risk of a subject by using at least one of heart rate trough phase and night heart rate difference as a marker. The application further discloses a device for evaluating heart biological rhythm or cardiovascular disease risk of a subject, comprising a data acquisition module, a data preprocessing module and a fitting module, wherein the fitting module evaluates cardiovascular disease risk of the subject by obtaining at least one of heart rate trough phase and night heart rate difference as a marker. The data obtained in the data acquisition module can come from a Holter dynamic electrocardio monitoring system or a wearable device with a heart rate module.

Description

TECHNICAL FIELD

[0001] The present application relates to a marker for assessing cardiac biological rhythm or cardiovascular disease risk and use thereof, and a device and system for assessing cardiac biological rhythm or cardiovascular disease risk using the marker. BACKGROUND

[0002] Physiological and pathological parameters of humans are strictly controlled by circadian rhythm. Circadian rhythm plays an important role in the cardiovascular system, and chronic disruption of the circadian rhythm system, such as shift work, jet lag, social interaction, eating pattern or circadian rhythm disorder, is also closely related to the occurrence and development of cardiovascular disease (CVD). Therefore, assessing the circadian rhythm state of an individual's cardiovascular system is of positive significance for the functional assessment of the cardiovascular system, disease risk prediction, diagnostic analysis and even treatment improvement.

[0003] There are many methods for measuring the circadian rhythm state of an individual, such as dim light melatonin onset (DLMO) and chronotype questionnaire, and researchers have also developed circadian rhythm state analysis methods based on various biological samples. However, these methods such as DLMO and chronotype questionnaire cannot directly reflect the circadian rhythm of the heart system. Heart rate is an important parameter of heart health and is the most easily and accurately measured, and the circadian variation of heart rate can also reflect the rhythmic changes of the function of the cardiovascular system. Previous studies have reported that the circadian difference (dipping) of heart rate is related to the risk of cardiovascular disease, but there is no in-depth correlation with biological rhythm. SUMMARY

[0004] Marker for assessing biological rhythm of a subject and method thereof

[0005] The present application discloses a marker for assessing the cardiac biological rhythm of a subject, the marker being at least one of a heart rate trough phase and a night heart rate difference (A).

[0006] The present application discloses a method for assessing the cardiac biological rhythm of a subject, wherein at least one marker of a heart rate trough phase and a night heart rate difference (A) is used to determine the biological rhythm of the subject.

[0007] The present application discloses a use of a marker in the preparation of a device for assessing the biological rhythm of a subject, wherein the marker is at least one of a heart rate trough phase and a night heart rate difference (A).

[0008] In one specific embodiment, the heart rate trough phase or night heart rate difference (A) obtained by fitting a curve to the heart rate of the subject's night resting state data, wherein the night resting state data refers to the heart rate data during the night sleep after removing the sleep cycle, getting up at night activity and other interference in the night resting state data.

[0009] In one embodiment, the heart rate fitting of the subject's night resting state is obtained by data preprocessing and data fitting using a device capable of collecting continuous heart rate, such as a Holter dynamic electrocardiogram monitoring system or a wearable device with a heart rate module, to obtain the heart rate nadir phase or night heart rate difference (A), which in turn reflects the subject's cardiac biological rhythm.

[0010] In one embodiment, the subject has a cardiac biological rhythm.

[0011] In one embodiment, the device for evaluating biological rhythm comprises

[0012] 1) a data acquisition module;

[0013] 2) a data preprocessing module;

[0014] 3) a data fitting module, comprising

[0015] i) a global fitting module; and optionally

[0016] ii) a local fitting module;

[0017] 4) a reporting module;

[0018] wherein the data acquisition module is used to acquire the heart rate data of the subject, including downloading the heart rate data of the subject and outputting the data to the data preprocessing module;

[0019] the data preprocessing module is used to obtain resting state data, including dividing the heart rate data of the subject by day to determine the night resting state time period and further removing the peak of fluctuation to obtain the resting state data;

[0020] the global fitting module is used to obtain the heart rate nadir phase and / or night heart rate difference, including fitting the night resting state data into a curve using a trigonometric function to obtain the heart rate nadir phase, and / or calculating the night heart rate difference;

[0021] the local fitting module is used to obtain the information of local peaks and local valleys within the night resting state time period in the fitted curve;

[0022] the reporting module is used to evaluate the subject's cardiac biological rhythm or cardiovascular disease risk.

[0023] In one embodiment, the apparatus comprises at least one of an input device and a wearable device operatively attached to a computing device.

[0024] In one embodiment, the apparatus comprises a device capable of collecting continuous heart rate such as a Holter ambulatory electrocardiogram monitoring system or a wearable device with heart rate module.

[0025] Marker for assessing cardiovascular disease risk of a subject and method thereof

[0026] Disclosed is a marker for assessing cardiovascular disease risk of a subject, the marker being at least one of a heart rate nadir phase and a nocturnal heart rate difference (A).

[0027] Disclosed is a method for assessing cardiovascular disease risk of a subject, comprising the steps of:

[0028] 1) collecting continuous heart rate data of a subject using a device capable of collecting continuous heart rate such as a Holter ambulatory electrocardiogram monitoring system or a wearable device with heart rate module, wherein the heart rate data density is normally 1 data point per minute;

[0029] 2) obtaining a period of time for acquiring resting state data: data preprocessing, removing the peaks of fluctuations caused by interference factors such as sleep cycles and night activities in the nocturnal resting state data; using a trigonometric function (cosine function) with a period of 24 hours to obtain a heart rate fitting curve of the subject in the nocturnal resting state, obtaining a heart rate nadir phase or calculating a nocturnal heart rate difference A; using a Butterworth filter to filter the small peak-valley parameters in the nocturnal resting state data. Wherein the nocturnal heart rate difference A is the amplitude of the heart rate curve fitted from the nocturnal static data.

[0030] 3) judging the cardiovascular disease risk of the subject according to the heart rate nadir phase or the nocturnal heart rate difference A value.

[0031] In one embodiment, the heart rate nadir phase or the nocturnal heart rate difference A value is obtained by fitting the heart rate of the subject in the nocturnal resting state, wherein the nocturnal resting state refers to the period of nocturnal sleep after removing the interference of sleep cycles, night activities and the like in the nocturnal resting state data.

[0032] In one embodiment, when the heart rate nadir phase is between 0 and 5 and / or the night heart rate difference is between 2.75 and 26, the subject is at low risk of cardiovascular disease; when the heart rate nadir phase is < 0, the subject is at high risk of atrial abnormalities such as atrial fibrillation or atrial flutter; when the heart rate nadir phase is > 5, the subject is at high risk of atrial abnormalities such as atrial fibrillation or atrial flutter, or at high risk of ventricular abnormalities such as ventricular fibrillation and / or ventricular flutter; when the night heart rate difference is > 26, the subject is at high risk of atrial abnormalities and conduction block; when the night heart rate difference is < 2.75, the subject is at high risk of sinus tachycardia and QRS. Wherein the subject has a cardiac biological rhythm.

[0033] In one embodiment, the subject is at high risk of cardiovascular disease when the subject has no cardiac biological rhythm or the night average heart rate / day average heart rate is > 1.

[0034] The present application further discloses a use of a marker in the preparation of a device for assessing the risk of cardiovascular disease in a subject, wherein the marker is at least one of the heart rate nadir phase and the night heart rate difference A.

[0035] In one embodiment, the device for assessing the risk of cardiovascular disease comprises

[0036] 1) a data acquisition module;

[0037] 2) a data preprocessing module;

[0038] 3) a data fitting module, comprising

[0039] i) a global fitting module; and optionally

[0040] ii) a local fitting module;

[0041] 4) a reporting module;

[0042] Wherein the data acquisition module is used to acquire heart rate data of the subject, including downloading heart rate data of the subject and outputting data to the data preprocessing module;

[0043] The data preprocessing module is used to acquire resting state data, including dividing the heart rate data of the subject by day to determine the night resting state time period and further removing fluctuation peaks to obtain resting state data;

[0044] The global fitting module is used to acquire the heart rate nadir phase and / or the night heart rate difference, including fitting the night resting state data into a curve using a trigonometric function to obtain the heart rate nadir phase, and / or calculating the night heart rate difference;

[0045] The local fitting module is configured to obtain information of local peaks and local valleys in the night resting state period in the fitted curve;

[0046] The reporting module is configured to evaluate the subject's cardiovascular disease risk according to the heart rate trough phase or the value of the night heart rate difference.

[0047] In one embodiment, the device comprises at least one of an input device and a wearable device operatively attached to a computing device.

[0048] In one embodiment, the device comprises a device capable of collecting continuous heart rate, such as a Holter ambulatory electrocardiogram monitoring system or a wearable device with a heart rate module.

[0049] Device and system

[0050] In another aspect, the present application discloses a device for evaluating heart biological rhythm or cardiovascular disease risk, comprising:

[0051] 1) a data acquisition module;

[0052] 2) a data preprocessing module;

[0053] 3) a data fitting module, comprising

[0054] i) a global fitting module; and optionally

[0055] ii) a local fitting module;

[0056] 4) a reporting module;

[0057] Wherein, the data acquisition module is configured to acquire heart rate data of a subject, including downloading heart rate data of the subject and outputting data to the data preprocessing module;

[0058] The data preprocessing module is configured to obtain resting state data, including dividing the heart rate data of the subject by day to determine the night resting state period and further removing fluctuation peaks to obtain the resting state data;

[0059] The global fitting module is configured to obtain a heart rate trough phase and / or a night heart rate difference, including fitting the night resting state data into a curve using a trigonometric function to obtain the heart rate trough phase, and / or calculating the night heart rate difference;

[0060] The local fitting module is configured to obtain information of local peaks and local valleys in the night resting state period in the fitted curve;

[0061] The reporting module is configured to evaluate the subject's heart biological rhythm or cardiovascular disease risk.

[0062] In one embodiment, the device further comprises a data collection module for collecting continuous heart rate data of the subject for more than 24 hours, preferably the data is for about 24 hours, more than 24 hours such as 36 hours, 48 hours, 60 hours, 72 hours, 1 week, 2 weeks, 3 weeks, 4 weeks, 1 month, 2 months, 3 months, 6 months, 9 months, 12 months.

[0063] In one embodiment, the device uses the marker of the nadir phase of heart rate and / or the nocturnal heart rate difference to assess the cardiac biological rhythm or the risk of cardiovascular disease.

[0064] In one embodiment, the value of the nadir phase of heart rate is obtained by using a cos function with a period of 24 hours to globally fit the nocturnal resting state data by least square method.

[0065] In one embodiment, the nocturnal heart rate difference is the amplitude of the fitted heart rate curve of the nocturnal resting state data.

[0066] In one embodiment, in the reporting module, when the value of the nadir phase of heart rate is between 0 and 5 and / or the value of the nocturnal heart rate difference is between 2.75 and 26, the subject is assessed as low risk of cardiovascular disease; when the value of the nadir phase of heart rate is < 0, the subject is assessed as high risk of atrial abnormality such as atrial fibrillation or atrial flutter; when the value of the nadir phase of heart rate is > 5, the subject is assessed as high risk of atrial abnormality such as atrial fibrillation or atrial flutter, or ventricular abnormality such as ventricular fibrillation or ventricular flutter; when the value of the nocturnal heart rate difference is > 26, the subject is assessed as high risk of atrial abnormality and conduction block; when the value of the nocturnal heart rate difference is < 2.75, the subject is assessed as high risk of sinus tachycardia and QRS. The subject is a subject with cardiac biological rhythm.

[0067] In one embodiment, in the reporting module, the subject is assessed as high risk of cardiovascular disease when the subject is without cardiac biological rhythm or the nocturnal average heart rate / daytime average heart rate > 1.

[0068] In one embodiment, the device comprises an input device operably attached to a computing device.

[0069] In one embodiment, the device can be a wearable device.

[0070] In one embodiment, the device further comprises a device capable of collecting continuous heart rate such as a Holter dynamic electrocardiogram monitoring system or a wearable device with a heart rate module.

[0071] In yet another aspect, the present application discloses a system for assessing cardiac chronobiology or cardiovascular disease risk, comprising the device of any one of the preceding aspects, wherein the cardiac chronobiology or cardiovascular disease risk is assessed by fitting at least one of the heart rate nadir phase and the nocturnal heart rate difference.

[0072] In one embodiment, when the heart rate nadir phase is between 0 and 5 and / or the nocturnal heart rate difference is between 2.75 and 26, the subject is at low risk of cardiovascular disease; when the heart rate nadir phase is < 0, the subject is at high risk of atrial abnormalities such as atrial fibrillation or atrial flutter; when the heart rate nadir phase is > 5, the subject is at high risk of atrial abnormalities such as atrial fibrillation or atrial flutter, or at high risk of ventricular abnormalities such as ventricular fibrillation and / or ventricular flutter; when the nocturnal heart rate difference is > 26, the subject is at high risk of atrial abnormalities and conduction block; when the nocturnal heart rate difference is < 2.75, the subject is at high risk of sinus tachycardia and QRS.

[0073] In another aspect, the present application discloses a use of a device capable of collecting continuous heart rate for assessing cardiac chronobiology or cardiovascular disease risk.

[0074] In one embodiment, the cardiac chronobiology or cardiovascular disease risk is assessed by fitting at least one of the heart rate nadir phase and the nocturnal heart rate difference.

[0075] In one embodiment, the device capable of collecting continuous heart rate comprises a Holter dynamic electrocardiogram monitoring system or a wearable device with a heart rate module. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 A flow chart showing the heart rate analysis method.

[0077] Figure 2 A diagram showing the heart rate analysis result. The thin line represents the original heart rate curve; the continuous thick line represents the filtered curve; the dotted line represents the curve obtained by global cos fitting; the black hexagonal star represents the lowest point of the global fitting curve; the × represents the local valley point; the * represents the local peak point; the two thick line segments on the sides represent the heart rate curve at falling asleep and waking up (used to obtain the fast rising and falling slopes).

[0078] Figure 3 A diagram showing the heart rate chronobiology parameter analysis result. The gray points represent the original heart rate data, and the continuous gray curve represents the filtered curve; the light gray arc curve on the left represents the heart rate nadir curve obtained by cos fitting; the black thick line represents the heart rate falling and rising curve; the black point represents the heart rate nadir.

[0079] Figure 4 A diagram showing the heart rate nadir phase versus the circadian questionnaire result.

[0080] Figure 5 Results showing heart rate nadir phase versus DLMO.

[0081] Figure 6 Holter patient heart rate rhythm groupings are shown.

[0082] Figure 7 Extreme heart rate circadian rhythm parameters population CVD risk is shown. DETAILED DESCRIPTION

[0083] The present application will be further described by a specific description.

[0084] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0085] As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise.

[0086] As used herein, the term "subject" includes any human or non-human animal. The term "non-human animal" includes all vertebrates, e.g., mammals and non-mammals, e.g., non-human primates, sheep, dogs, cats, horses, cows, chickens, rats, mice, amphibians, reptiles, etc. The terms "patient" or "subject" are used interchangeably unless otherwise indicated. In the present application, the preferred subject is a human.

[0087] As used herein, the term "resting state data", also referred to as "night resting state data", refers to heart rate data obtained during a time period of night sleep after calibration time, data partitioning and filtering, and removal of interfering data such as sleep cycles, night awakening activities, etc. The term "night resting state time period" refers to a time period of night sleep state obtained after calibration time, data partitioning and filtering. The night resting state data described in the present application is data after data pre-processing.

[0088] As used herein, the term "heart rate nadir phase" refers to the time point of the lowest point of the curve obtained after fitting a trigonometric function curve to the night resting state data, which time is used to reflect the heart circadian rhythm.

[0089] As used herein, the term "night heart rate difference" refers to the amplitude of the night heart rate fitted curve obtained after fitting a trigonometric function curve to the night resting state data.

[0090] As used herein, the term "no heart circadian rhythm" refers to no rhythm in heart rate data determined using Jonckheer-Terpstra-Kendall rhythmicity analysis (JTK_CYCLE) algorithm.

[0091] As used herein, the term "wearable device" refers to a portable device that is worn directly on the body, or is integrated into a user's clothing or accessory. A wearable device can be a hardware device that can be supported by software for the purpose of data interaction, cloud interaction, or data analysis.

[0092] Embodiments

[0093] The experimental methods in the following examples are conventional methods unless otherwise specified. The present application will be further understood with reference to the following non-limiting experimental examples.

[0094] Example 1 Heart rate analysis method

[0095] The analysis method in the present application uses continuously recorded heart rate data, which can be derived from a Holter dynamic electrocardiogram monitoring system or a wearable device with a heart rate module. The heart rate data density is usually 1 data point per minute.

[0096] Specifically, a total of 211 volunteers (72 males and 139 females) were recruited. The volunteers were required to complete a morning-evening questionnaire (MEQ) and then wear a smart wristband for at least 1 month for data collection. The wearable devices were purchased from two independent suppliers. The 1-minute frequency heart rate (HR) data collected by the smart wristband were retrieved from the manufacturer's cloud server through the application program interface after user authorization and stored in the local database. The study design conforms to the Helsinki Declaration and was approved by the Ethics Committee of Suzhou University (ECSU-201800098).

[0097] Since the daytime heart rate data is affected by various subjective and external factors such as exercise, work intensity, and social interaction, it cannot objectively reflect the individual's biological rhythm on that day. Therefore, the heart rate data in the resting state at night was selected as the source of rhythm parameters. After removing the sleep cycle and night activity interference in the resting state data at night, the core global parameters were obtained using trigonometric function fitting. In addition, the small peak-valley parameters in the resting state data at night were obtained using a Butterworth filter. The overall process diagram is shown in Figure 1 , which specifically includes the following steps:

[0098] 1) Obtain the resting state data time period:

[0099] 1-1) Calibration time

[0100] Time data is the basis for subsequent processing. Considering the particularity and difference of time data storage in different software, the initial time needs to be calibrated before the work begins.

[0101] 1-2) Data division

[0102] Divide the time series data by day, with 14:00 as the dividing point. At the same time, filter out incomplete data.

[0103] 1-3) Filtering

[0104] Plot the heart rate data for each day as a curve. This curve is formed by superimposing a macroscopic biological clock curve and many small noises (jitter), i.e., it is composed of low-frequency components and high-frequency components. The low-frequency components correspond to slow changes in the curve, while the high-frequency components correspond to rapid changes in the curve.

[0105] To study the biological clock curve of the sleep stage, we need to filter out high-frequency noise, so we use a low-frequency filter to filter or significantly attenuate the high-frequency components of the curve, allowing the low-frequency components to pass. Here we use a Butterworth low-pass filter to obtain the filtered data.

[0106] 1-4) Obtain sleep data (resting state)

[0107] a. Calculate the mean heart rate B_mean and the median heart rate B_prctile for the whole day using the filtered data smooth_filter. Take the minimum of the two as the critical value B_inf.

[0108] b. Find the start point of the sleep data: Take the length interval(1) = 240 (in minutes), and calculate how many points in the time interval [1:interval(1)] have heart rates less than B_inf, i.e., the number of points below B_inf, denoted as count_point.

[0109] c. If count_point = 0, slide the time interval to the right by interval(1), and perform similar operations as in b for the new time interval.

[0110] If 0 < count_point < interval(1), slide 1 unit to the right, and perform similar operations as in b for the new time interval.

[0111] If count_point = interval(1), mark the left end point of the corresponding time interval as the start point of the sleep data, and end the loop.

[0112] d. Find the end point of the sleep data:

[0113] The principle is the same as a, but change the direction from right to left,

[0114] e. If count_point = 0, slide the time interval to the left by interval (1), repeat the operation of d for the new time interval.

[0115] If 0 < count_point < interval (1), slide to the left by 1 unit, repeat the operation of d for the new time interval.

[0116] If count_point = interval (1), mark the right end point of the corresponding time interval as the sleep data termination point, and the loop ends.

[0117] In the case of insufficient sleep time at night, the above method may not be able to effectively obtain the required data. At this time, change the interval length from interval (1) = 240 to interval (2) = 120 (unit: minute), and repeat the relevant steps.

[0118] f. In order to obtain more complete sleep data, consider extending slightly at both ends.

[0119] For the left end (the starting end), if there is an extension margin, consider moving the starting point to the left by continue_len = 60 (unit: minute). If overflow occurs, that is, the starting point exceeds the actual recording range at this time, take 0 <= movement < continue_len.

[0120] For the extended data, do first-order difference. Define the number of rises as the number of differences greater than zero, and define the number of falls as the number of differences less than zero. Calculate the number of rises and falls, and if the number of falls >= 2 times the number of rises, it is necessary to extend, change the starting point coordinate, otherwise no extension is needed, that is, no coordinate modification is needed.

[0121] g. For the right end (the termination end), if there is an extension margin, consider moving the termination point to the right by continue_len = 60 (unit: minute). If overflow occurs, that is, the end point exceeds the actual recording range at this time, take 0 <= movement < continue_len.

[0122] For the extended data, do first-order difference, calculate the number of rises and falls as defined above, and if the number of rises >= 2 times the number of falls, it is necessary to extend, change the starting point coordinate, otherwise no extension is needed, that is, no coordinate modification is needed.

[0123] h. According to the coordinates determined by f and g, cut the sleep data.

[0124] 2) In the resting state period, remove the fluctuating peaks (considered as sleep cycles, night activities and other interference factors) to obtain the preprocessed night resting state data

[0125] To the acquired sleep data, because the whole sleep stage contains many small sleep cycles, which is reflected in the heart rate curve as many small fluctuations, in order to obtain macro sleep resting state data, using sliding window to traverse the data comparison, define the start / end point of local peak / trough and delete it.

[0126] 3) Using a 24-hour cycle cos function, least squares method (adaptive parameter selection) global fitting to obtain the minimum valley phase and heart rate difference;

[0127] Phase is the time corresponding to the minimum point of the fitted cos curve (black hexagonal star horizontal coordinate in the figure)

[0128] Heart rate difference calculation method:

[0129] Night heart rate difference = 2*(night heart rate mean-cos fitting curve minimum position heart rate)

[0130] Diurnal heart rate difference = 2*(all-day heart rate mean-cos fitting curve minimum position heart rate)

[0131] 4) Calculate the parameters related to the rapid decline and rise curve of heart rate when changing between resting and active states;

[0132] Divide the sleep data into three equal parts, marked as left, middle and right.

[0133] Use the left segment data to find the slope value when entering sleep state. Take a time interval of length interval(3)=50, define the statistic "drop score" to judge the quality of the selected time interval at this time (i.e. the effect of taking the slope of the interval). Definition: Take the first-order difference of the interval data, calculate the drop number (the number of negative values in the difference data), the drop amount (the sum of the absolute values of the negative values in the difference data), then take 0.3, 0.7 as the weight to sum them up, called the drop score of the interval. Traverse the left segment to find the highest score interval, find the maximum value max_info and its position max_place in the left half of the interval, and find the minimum value min_info and its position.

[0134] The slope of the sleep state is the number of heart rate drops per minute (taking the negative) in this time interval. The end of the drop time is the time corresponding to the last time of the sleep time interval, and the end of the drop heart rate is the heart rate corresponding to the end of the drop time.

[0135] The slope value of the wake-up stage is calculated using the right segment data. A time interval of length interval(3) = 50 is defined as the "rise score" to determine the quality of the selected time interval (i.e. the effect of calculating the slope using the interval). Definition: the first-order difference is calculated for the interval data, the rising number (the number of values greater than zero in the difference data) and the rising amount (the sum of values greater than zero in the difference data) are calculated, and then the two are summed with weights of 0.3 and 0.7 to obtain the rise score of the interval. The highest score interval is found by traversing the right segment. The minimum value min_info and its location min_place of the left half of the interval are found, and the maximum value max_info and its location max_place of the right half of the interval are found. The slope value of the wake-up stage is the number of heart rate rises per minute in the time interval. The rise start time is the time corresponding to the first time of the wake-up time interval, and the rise start heart rate is the heart rate corresponding to the rise start time.

[0136] 5) Calculate multiple local wavelet parameters (possibly related to sleep cycles) in the resting state time interval based on the Butterworth filter.

[0137] Local peaks and troughs are calculated using filtered sleep data. The time interval is divided into three equal parts, and the mean heart rate of the left, middle, and right parts of the time interval is calculated, as well as the maximum or minimum heart rate of the left and right parts. If the mean heart rate of the middle part is greater than the mean heart rates of the left and right parts, and the difference between the maximum value of the middle part and the minimum value of the left half, and the difference between the maximum value of the right half and the minimum value of the middle part are both greater than the threshold, then the maximum heart rate in the time interval is taken as the local peak point. If the mean heart rate of the middle part is less than the mean heart rates of the left and right parts, and the maximum value of the left half and the difference between the maximum value of the right half and the minimum value of the middle part are both greater than the threshold, then the minimum heart rate in the time interval is taken as the local trough point.

[0138] Cases where the night amplitude is lower than the minimum 5% are classified as arrhythmia, and cases where the night phase differs from the average phase of the population by ±10-14 hours are classified as reverse phase. These two cases are no longer considered for further parameter determination. If neither of these two cases occurs, the following six parameters with clear rhythm significance are output: night phase, night heart rate amplitude, night heart rate mean, night heart rate rapid decline slope, night heart rate rapid rise slope, and sleep symmetry index. The results are output to the customer.

[0139] The fitting of the heart rate curve is shown in Figure 2 The schematic diagram of the analysis results of the biological rhythm parameters is shown in Figure 3 The list of analyzed parameters is shown in Table 1, where the most critical parameters are:

[0140] 1) Heart rate trough phase: The time point corresponding to the highest peak of the rhythmic data is conventionally defined as the phase (peak phase) of the rhythm. However, since the daytime heart rate data is greatly disturbed by activities, our algorithm selects the time point corresponding to the lowest trough of the heart rate fitting curve during the night sleep to reflect the heart rate rhythm phase, i.e. the heart rate trough phase.

[0141] 2) Night heart rate difference: Amplitude is also an important parameter in biological rhythm. Conventionally, the difference between the highest peak and the lowest trough of the fitting curve reflects the oscillation intensity of heart rate in 24-hour cycle. Since we mainly analyze the night data, we define the night heart rate difference.

[0142] Table 1. List of heart rate biological rhythm parameters

[0143]

[0144] Example 2 Accuracy of heart rate biological rhythm analysis results

[0145] To prove that the heart rate trough phase can be used as a biomarker of biological rhythm, the inventors used the mature tools commonly used in the prior art for biological rhythm analysis (morningness-eveningness questionnaire MED and melatonin concentration change under dim light DLMO) for comparison. The analysis steps and results are as follows:

[0146] 1. Comparison of heart rate trough phase and morningness-eveningness questionnaire results

[0147] The morningness-eveningness questionnaire (MEQ) is a mature tool for analyzing individual chronotypes. We conducted a volunteer experiment to compare the heart rate trough phase obtained by heart rate biological rhythm analysis and the MEQ chronotype. According to the results obtained from 211 volunteers, the heart rate trough phase and the EMQ chronotype showed good correlation (Fig. Figure 4 A). Further correlation analysis of heart rate trough phase value and chronotype questionnaire score (the higher the score, the later the chronotype) showed a significant negative correlation (Fig. Figure 4 B, significance P<0.0001).

[0148] 2. Comparison of heart rate trough phase and DLMO results

[0149] Melatonin concentration change under dim light (DLMO) is considered a gold standard for determining individual circadian rhythms. We also conducted an experiment to compare the heart rate trough phase obtained by heart rate biological rhythm analysis and DLMO.

[0150] Twelve volunteers gathered in two light-protected rooms from 18:00 to 24:00. Saliva samples were collected every 30 minutes, stored in a refrigerator, and tested using an ELISA kit (IBL International, Switzerland). The dim light melatonin onset (DLMO) was determined for each person using the hockey-stick method in MATLAB (Danilenko KV, Verevkin EG, Antyufeev VS, Wirz-Justice A, Cajochen C: The hockey-stick method to estimate evening dim light melatonin onset (DLMO) in humans. Chronobiol Int 2014, 31 :349-355.).

[0151] According to the results obtained in 9 volunteers, the correlation analysis between the heart rate trough phase and the DLMO proved that both were linearly correlated (r = 0.89, p < 0.05). Figure 5

[0152] The comparative analysis of the above two aspects confirmed that the heart rate trough phase could accurately reflect the individual's biological rhythm state and was a good marker.

[0153] Example 3 Correlation between heart rate biological rhythm and cardiovascular disease

[0154] ​To further analyze the relationship between heart rate circadian parameters and the incidence of cardiovascular diseases, we collected 11, 074 Holter datasets from the Department of Cardiology, the First Affiliated Hospital of Soochow University, covering relevant patients from September 2010 to July 2014. Then, the Holter datasets were screened to ensure the integrity of the data and exclude patients using artificial pacemakers, so as to perform an unbiased analysis of the time pattern. A total of 10, 095 datasets were further analyzed. The simplified HR data of 1-minute frequency and diagnostic conclusions were exported using the custom module of the Holter software ECGLab (Shenzhen Biomedical Instrument Co.). This retrospective study design was approved by the hospital ethics review committee (application number: 2019025). Among the 10, 095 Holter heart rate data of clinical patients, the age of the patients was 8 to 97 years old, and most of the population (25th-75th percentile) was between 48 and 68 years old. The Holter data contained complete electrocardiograms, from which we exported the simple heart rate data (1 data point per minute) and analyzed the circadian parameters of the heart rate of these patients using the method described in Example 1. In addition, the Holter data also contained diagnostic conclusions, and the Holter dataset contained the diagnostic conclusions given by experienced cardiologists according to the clinical guidelines (AHA / ACCF / HRS recommendations for standardization and interpretation of electrocardiography). We also extracted the clinical indicators related to cardiovascular diseases from them, and divided them into 7 categories and 13 subcategories of CVD indicators (Table 2).

[0155] Table 2. CVD indicator classification

[0156]

[0157] We first filtered the data with a Butterworth filter to remove noise. Then we used a sliding window to automatically distinguish between rest and activity periods, replacing the fixed day and night with the slope of the HR descent and ascent. The first time point of the sliding window, where all HRs are below the threshold, was considered as the start of the night period by comparing it with the threshold calculated from the filtered HR data. Similar processing was done to determine the end of the night. Then, we used the least square method to fit a cos function to the night HR data. The minimum fitted HR value and the time of the trough were determined automatically. In addition, the average daily HR value, the average resting HR value, and the minimum fitted HR value were used to calculate the subject's HR night and circadian variation during the night and the entire circadian cycle. Using the algorithm of K-means clustering, we automatically searched for the onset of the transition from low stable state to high state. The first active point ttonsetland the first time point ttonset2, where the difference between the data point and the average of the previous three points reached a certain threshold, were two candidate onset times after 4 am. The earlier of the two candidate points was selected as the onset time. Similarly, all parameters were retrieved from the wristband-based HR data.

[0158] We first analyzed the relationship between heart rate rhythm and CVD. The Jonckheer-Terpstra-Kendall rhythm analysis (JTK_CYCLE) algorithm was used to determine whether the heart rate data had a rhythm. According to the results, the Holter patients were divided into two categories: non-rhythmic (32.5%) and rhythmic (67.5%). The rhythmic group was further divided into anti-phase group (night average heart rate / day average heart rate≥1) and positive phase group (night average heart rate / day average heart rate<1). Figure 6 Linear regression was used to evaluate the correlation, and one-way ANOVA with Bonferroni test in GraphPad Prism 8 was used to evaluate the association between disease occurrence and chronotype by inputting the baseline characteristics and chronotypes of CVD index in ECG data: arrhythmic type, anti-phase type, and rhythmic type. In all statistical analyses, a two-sided P value <0.05 was considered statistically significant. Using Bonferroni analysis, we found that the anti-phase group had significantly higher atrial events, ventricular events, sinus tachycardia, conduction block, and QRS risk compared to the normal rhythm group; while the non-rhythm group had significantly higher atrial events, sinus bradycardia, conduction block, and QRS risk (Table 3).

[0159] To further evaluate the correlation between the heart rate chronobiological parameters and the CVD indicators, we performed a progressive partial correlation analysis to determine the critical points of the heart rate trough phase (HR Trough Phase) and the nocturnal variation (Nocturnal Variation) in the rhythmic group (including the positive phase and the reverse phase). We found that the people with the heart rate trough phase between 0 and 5 (89.7% of the rhythmic group) and the nocturnal variation between 2.75 and 26 (91.9% of the rhythmic group) had a lower risk of CVD indicators. After determining the normal range according to the critical points, we further studied the correlation between the heart rate chronobiological parameters and the CVD risk indicators in the abnormal range using Pearson correlation analysis. The results showed that:

[0160] Table 3. Comparison of CVD risk in different heart rate chronobiological groups

[0161]

[0162] Statistical significance

[0163] *: P < 0.05; **: P < 0.01; ***: P < 0.001; ****: P < 0.0001

[0164] In the rhythmic group, extreme heart rate trough phase ( or ) was closely related to atrial abnormal events (atrial fibrillation and atrial flutter) ( Figure 7 A), while ventricular abnormal events (ventricular fibrillation and ventricular flutter) were only related to the severe delay of heart rate trough phase ( Figure 7 A).

[0165] In the reverse phase group, the nocturnal variation of heart rate was most closely related to atrial abnormal events and sinus bradycardia ( Figure 7 B).

[0166] In the rhythmic group, excessive nocturnal variation of heart rate (A ≥ 26) was closely related to atrial events and conduction block ( Figure 7 C), while the excessive small nocturnal variation of heart rate (A ≤ 2.75) was significantly related to sinus tachycardia and QRS ( ​ C).

Claims

1. An apparatus for assessing a subject's cardiac bio-rhythm or risk of cardiovascular disease, characterized by, Comprising: 1) a data acquisition module for acquiring heart rate data of a subject, including downloading heart rate data of a subject and outputting data to a data preprocessing module; 2) a data preprocessing module for acquiring resting state data, including segmenting heart rate data of a subject by day to determine a night resting state time period and further removing fluctuation peaks to obtain night resting state data; 3) a data fitting module, including: i) a global fitting module for acquiring a heart rate nadir phase and / or a night heart rate difference, including fitting the night resting state data into a curve using a trigonometric function to acquire a heart rate nadir phase, and / or calculating a night heart rate difference; and optionally ii) a local fitting module for obtaining local peak points and local valley points within the night resting state time period in the fitted curve; 4) a reporting module for assessing a subject's cardiac biological rhythm or cardiovascular disease risk; wherein segmenting heart rate data of a subject by day to determine a night resting state time period and further removing fluctuation peaks to obtain night resting state data includes calculating a mean heart rate for the whole day and a median heart rate for the whole day, taking the minimum of the two as a threshold, obtaining a start point and an end point of sleep data according to the threshold, using a sliding window to traverse and compare the data, defining the start point and end point of local peak and valley and deleting them; in the reporting module, when the value of the heart rate nadir phase is between 0 and 5 and / or the value of the night heart rate difference is between 2.75 and 26, the subject is assessed as a low risk of cardiovascular disease; when the value of the heart rate nadir phase is ≤0, the subject is assessed as an atrial abnormality; when the value of the heart rate nadir phase is ≥5, the subject is assessed as an atrial abnormality; when the value of the night heart rate difference is ≥26, the subject is assessed as a high risk of atrial abnormality and conduction block; when the value of the night heart rate difference is ≤2.75, the subject is assessed as a high risk of sinus tachycardia and QRS abnormality, and the subject has a cardiac biological rhythm.

2. The apparatus of claim 1, wherein, Further comprising a data collection module for collecting heart rate data of a subject for more than 24 hours.

3. The apparatus of claim 1, wherein, In the global fitting module, a cos function with a period of 24 hours is used, and the night resting state data is fitted by least squares.

4. The apparatus of claim 1, wherein, The night heart rate difference is the amplitude of the heart rate curve fitted from the night resting state data.

5. The apparatus of claim 1, wherein, In the local fitting module, the local peak points and local valley points are calculated based on a Butterworth filter.

6. The apparatus of claim 1, wherein, In the reporting module, when the subject has no cardiac biological rhythm or the night average heart rate / daytime average heart rate ≥1, the subject is assessed as a high risk of cardiovascular disease.

7. The apparatus of claim 1, wherein the apparatus comprises an input device operably attached to a computing device.

8. The apparatus of claim 1, wherein the apparatus can be a wearable device.

9. The apparatus of claim 1, wherein the apparatus further can comprise a device capable of collecting continuous heart rate.

10. A system for assessing a cardiac biological rhythm or a cardiovascular disease risk, comprising: including the apparatus of any one of claims 1 to 9, wherein the cardiac biological rhythm or the risk of cardiovascular disease is further assessed by fitting at least one of a heart rate nadir phase and a nocturnal heart rate difference.

Images