CN117643471A - Wearable and portable systems and methods for measuring cardiac parameters to detect heart disease - Google Patents

Abstract

The present disclosure relates to wearable and portable systems and methods of measuring cardiac parameters to detect heart disease. A system for measuring cardiac parameters uses a motion sensor to generate a seismogram signal and a cardiac parameter calculation unit. The heart parameter calculation unit is used for generating an envelope signal related to the seismogram signal; identifying signal segments in the envelope signal having a repeating pattern; identifying pairs of consecutive peaks in the signal segment such that a first peak of each pair of consecutive peaks is a systole peak and a second peak of each pair of consecutive peaks is a heart Shu Feng; the systole and diastole of each pair of consecutive peaks is calculated.

Description

Wearable and portable systems and methods for measuring cardiac parameters to detect heart disease

Technical Field

The present disclosure relates to a wearable portable system and method for measuring cardiac parameters to detect heart disease.

Background

In particular, the system may be associated with a patch, adhesive electrode, chest strap, or the like, adapted to the sternum of a person, and configured to provide data related to the health status of the person, e.g., via a mobile device such as a mobile phone, smart watch, or the like.

Today, more and more applications and programs are available, associable with dedicated devices or mobile devices, able to allow monitoring of the health of a person under normal living conditions and at any time of the day. These applications and programs are being investigated to operate without having to connect to complex and expensive machines and without requiring the presence of doctors and/or healthcare personnel.

In particular, these applications and programs aim to provide more and more health information in a personally usable manner, with a good level of accuracy and seamless (seamless). Since they are associated with battery powered devices, they are expected to have low power consumption.

For example, in the case of cardiac activity monitoring, there are already available programs and applications that can be executed by portable devices and allow the execution of Electrocardiography (ECG) and photoplethysmography waves photoplethysmography (PPG). In particular, in the case of an electrocardiogram, analog devices and electrodes connected to the human body are used to detect biopotential associated with heart pulses. In the case of photoplethysmographic waves, optical elements such as LEDs and photodiodes allow monitoring of volume changes in the peripheral circulation. In this way, the heartbeat and its variability can be measured.

Recently, other heartbeat analysis techniques have emerged, but in general, they require complex and expensive equipment and procedures and are therefore mainly only suitable for laboratory and medical practice environments. In fact, wearable and portable applications are very sensitive to disturbances that do not allow reliable results to be obtained.

For example, microphone-based Phonocardiograms (PCM) and ultrasound-based echocardiography have been proposed. Recently, in research, the relationship between phonocardiogram and blood pressure has been highlighted, particularly between PCM-based diastole (diaston) and Pulse Transit Time (PTT), which in turn is related to blood pressure.

Another recently proposed technique for studying heart beat includes a Seismograph (SCG), which allows measurement of chest vibrations associated with heart beat. For this reason, the use of highly sensitive accelerometers to detect heart motion has been proposed.

For example, "A low cost sensing device to detect cardiac timing and Function" by Anh Dinh et al, 2012IEEE 18 ^th International mixed-signals, sensors, and systems test workshop,978-0-7695-4726-8/12 describe the relationship between ECG and SCG (also see fig. 1A and 1B, respectively showing an electrocardiogram and corresponding seismogram), and propose to use accelerometers to detect electrocardiogram movements in association with ECG Sensors and microphones measuring phonocardiograms. In particular, in this context, it is suggested to use the seismogram signal to accurately measure mitral valve (mitral valve) closure, aortic valve (aort valve) opening, ventricular ejection, etc. Furthermore, it is stated in this article that the seismogram signal should allow prediction of events based on expected motion and the duration of cardiac events, however the necessary processing need not be explained in detail, but rather indicates that further investigation is required to determine the relationship between the timing of the SCG waveform and the cardiac timing.

Other references (e.g. "Precision wearable accelerometer contact microphones for longitudinal monitoring of mechano-acoustic cardiopulmonary signals" by P. Gupta et al, npj digital medicine, https:// www.nature.com/optics/s 41746-020-0225-7, 12/2020) describe acquisition of SCG signals by accelerometers in combination with other sensors, or use of accelerometers instead of microphones to detect signal characteristics and compare them to standard values (see e.g. "Accelerometer Type Cardiac Transducer for Detection of Low-Level Heart Sounds" by V. Padmanaban et al, IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL.40, NO., JANUARY 1993).

In "Automatic Identification of Systolic Time Intervals in Seismocardiogram" by g.shafiq et al, https: org/10.1038/srep 37524 and in "Seismocardiographic adjustment of diastolic timed vibrations" by K.Tavakolia et al, DOI:10.1109/embc.2012.6346794 discusses the use of SCG signals for predicting the duration of cardiac signal features such as systole (systole) or diastole; however, these documents do not provide accurate teachings regarding the implementation of devices and/or applications capable of processing such information in an automatic and usable manner by portable and/or wearable devices.

The described solution does not allow to provide portable and wearable devices capable of detecting heart parameters as well as in the case of heart parameters.

In fact, echocardiography is a non-portable and wearable medical device, and wearable and portable devices that use microphones for auscultation are sensitive to ambient acoustic noise that may hide or distort the results.

It has also recently been noted (see, e.g., L.Cheng et al, "Study of the Correlation Between the ratio of Diastolic to Systolic Durations and Echocardiography Measurement and Its Application to the Classification of Heart Failure Phenotypes", international journal of general medicine, 2021:145493-5503) that the ratio between systole (S, hereinafter also referred to as S12) and diastole (D, hereinafter also referred to as S21) is relevant as an indicator of cardiomyopathy and allows for early detection of cardiac dysfunction.

However, the use of echocardiography (and phonocardiograms that are the subject of other studies and publications) is not applicable to portable devices.

Disclosure of Invention

Various embodiments of the present disclosure provide an alternative way to obtain systole (S), diastole (D) and its ratio so that it can be implemented in a portable/wearable device.

In accordance with the present disclosure, a system and method for detecting cardiac parameters is provided.

Drawings

For a better understanding of the various embodiments of the present disclosure, embodiments thereof will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:

FIGS. 1A and 1B show examples of Electrocardiography (ECG) and corresponding Seismography (SCG) signals acquired in a human patient;

FIG. 2 is a general block diagram of a cardiac parameter measurement system for detecting heart disease according to one embodiment;

FIG. 3 is a general block diagram of a cardiac parameter measurement system for detecting heart disease according to another embodiment;

FIGS. 4-7 illustrate examples of signals detected by the system of FIG. 2 or FIG. 3 and their processing according to the present method;

FIG. 8 is a flow chart of a method of cardiac signal processing of the present disclosure; and

fig. 9-11 illustrate some steps of the method of fig. 8 in more detail.

Detailed Description

The present specification relates to a system based on a technique called Seismograph (SCG) that utilizes accelerometers to measure systole and diastole of the cardiac cycle.

Fig. 1A and 1B show examples of electrocardiographic signals obtainable with a portable device and corresponding electrocardiographic signals detected by an accelerometer. In particular, the comparison of fig. 1A and 1B shows the correlation between the two signals and the possibility of detecting the heart cycle S11, systole S12 and diastole S21 by means of the seismocardiographic signals.

As mentioned above, these period periods, in particular their ratio S12/S21, have been considered to be relevant for the calculation of a cardiomyopathy index, which is important for early identification of heart disease.

Fig. 2 shows a system 1 for detecting cardiac parameters, which allows detection of a cardiac cycle S11, systole S12 and diastole S21.

In detail, the system 1 of fig. 2 comprises a detection section 4, a processing section 5 and a power supply section 6.

The detection part 4 here comprises a motion sensor 2. The motion sensor 2 is typically an accelerometer integrated in a semiconductor chip and therefore has a very small size. The accelerometer may be, for example, a three-axis MEMS accelerometer or a six-axis IMU inertial platform, which combines an accelerometer and a gyroscope.

The motion sensor 2 is for example associated, for example in combination or fixedly, with a fixation element 3, which fixation element 3 allows positioning it on the sternum of the patient with a good mechanical coupling, so that the heartbeat can be detected.

For example, the fixing element 3 may be plaster (play), a suction cup element similar to an ECG detection electrode, an elastic tape, or the like.

The motion sensor 2 generates a motion signal SCG which is sent to the processing section 5.

The motion sensor 2 may be connected to the processing section 5 by wire or by a wireless system; in this case it is provided with a suitable interface (not shown) to set up the known transmission circuit.

As an alternative to what has been shown, the motion sensor 2 may be integrated in the processing part 5, for example in a dedicated device, fixed or fixable to the fixation element 3.

The processing section 5 and the power supply section 6 are here arranged within a heart parameter detecting device 10. The heart parameter detecting means 10 are for example formed by a dedicated portable device, a headset, a smart phone, a smart watch, a tablet, a portable computer or other smart device.

The processing section 5 includes a processing unit 15; one or more memories 16 and an I/O unit 17.

The processing unit 15 may comprise, for example, a microcontroller, a microprocessor and/or another CPU (central processing unit).

Memory 16 typically includes non-volatile memory for programs and data to be permanently stored and one or more buffers for data available for processing; in particular, in an embodiment, it includes five FIFO memories, denoted F1-F5 and discussed below. Alternatively, a FIFO memory may be provided in the processing unit 15. The program is executed by the processing unit 15, for example.

The I/O unit 17 comprises units and circuits for external communication, for example for transmitting and/or displaying the heart cycles S11, S12, S21, and may thus comprise radio circuits, wireless communication systems, screens, displays, acoustic annunciators, etc.

Herein, the power supply section 6 includes a battery 20 and a Power Management Integrated Circuit (PMIC) 21, which is coupled to the processing section 5, and is of a known type.

Fig. 3 shows a different heart parameter detection system denoted by 1'.

The system 1' has a general structure similar to the system 1 of fig. 2; accordingly, like parts are denoted by like reference numerals.

In the system 1 'of fig. 3, the detection part, indicated by 4', comprises, in addition to the motion sensor 2 and the fixation element 3, two ECG electrodes 20 for detecting an electrocardiogram signal.

The processing section 5' comprises an ECG preprocessing unit 18 for a first processing (front end) of the electrocardiogram signal.

The ECG electrodes 20 and the ECG preprocessing unit 18 may be of a known type and coupled to each other by a cable 19. The ECG preprocessing unit 18 is also connected to the microcontroller 15 to provide preprocessing signals and allow health predictions to be performed based on ECG and SCG maps.

Further, in fig. 3, the power supply portion indicated by 6' is external to the device 10. Alternatively, it may be internal, as shown in FIG. 2.

The system 1 or 1' of fig. 2 or 3 is configured to perform a cardiac parameter measurement method and in particular to extract systole and diastole from the signals SCG detected by the motion sensor 2 of fig. 2 and 3.

Specifically, the measurement method performed by systems 1 and 1' is based on:

acquiring a signal SCG through an accelerometer;

processing the signal SCG to obtain its envelope;

segmenting the sequence for searching for peaks (systole peaks or heart Shu Fengzhi);

verifying the result of identifying the systole peak/heart Shu Qifeng pair; and

parameters are calculated from the peak pairs, including calculating systole S12, diastole S21, and, if necessary, cardiac cycle s11=s12+s21.

The processing of the signal SCG may include filtering, possible signal enhancement processing of the amplified peaks and calculation of the envelope, for example by calculating the effective value (RMS, root mean square).

Sequence segmentation includes dividing signal samples into groups of samples (signal segments) having a repeating pattern according to the periodicity of the pattern of the cardiac signal, and herein includes quantizing the signals (transforming the signals into sequences of "0" and "1") and searching for local maxima (peaks) within each sequence, the local maxima samples having a first logic level (hereinafter, value "1"), each sequence being defined by switching from a second logic level (hereinafter, value "0") to the first logic level and by switching back from the first logic level to the second logic level.

Verification of the results herein includes identifying peak or maximum pairs having features (in this example, amplitude, distance, and duration) that are compatible with systole and diastole.

The calculation of the periods S12, S21 and S11 includes calculating these periods in each signal segment.

Hereinafter, an embodiment of the present cardiac parameter measurement method will now be described with reference to fig. 8, which illustrates a schematic flow chart according to the present disclosure, and fig. 4-7, which illustrate examples of a seismocardiographic Signal (SCG) detected by the motion sensor 2 and an intermediate signal obtainable from the signal SCG.

The heart parameter measurement method of fig. 8 may be performed by the processing unit 15 of fig. 2 or 3 and may be activated automatically, e.g. by the device 10, upon detection of measurement conditions typically present in commercial accelerometers (resting person, receiving a seismogram signal SCG with characteristics similar to heart beat, or after activation controlled by the motion sensor 2 using FSM/MLC (finite state machine/machine learning core) resources), or manually by a command provided, e.g. by the I/O unit 17.

At start-up, step 52, an input signal SCG (seismogram signal) is acquired. The input signal SCG is a digital signal, sampled at a much higher frequency than the heart frequency (e.g. about 2 kHz), and is highly variable.

The acquired signal may have a variable duration (number of samples) starting from a minimum value, for example corresponding to two heartbeats. In a manner not shown, after acquisition of the input signal SCG, verification of the duration (number of samples) of the input signal SCG sufficient for continuous processing may be provided, instead, a warning signal may be generated and the heart parameter measurement method 50 may be interrupted.

In step 54, the input signal SCG is filtered to eliminate unwanted harmonic components and high frequency noise of the input signal SCG, thereby minimizing distortion due to the phase shift, i.e., without changing the peak value of the input signal over time, thereby obtaining a filtered signal s1.

For example, fig. 4 shows a possible pattern of the filtered signal s1.

In step 56, the filtered signal s1 is processed using a nonlinear function, such as a hyperbolic sine (emphasizing the phase), to enhance its characteristics and amplify the amplitude of the peak relative to background noise. In practice, here, the absolute amplitude of the signal is not very relevant; vice versa, the time position of the peak is of interest.

In step 58, the envelope of the filtered signal s1 is calculated, for example by a calculation operation over a movable window (sliding window) of root mean square RMS. Thereby an envelope signal s2 is obtained, for example as shown in fig. 5.

Subsequently, the envelope signal s2 is quantized (i.e. transformed into the sequences "1" and "0", step 60), and its local maximum is searched (step 62, the maximum or peak is searched).

The sequences "1" and "0" obtained in step 60 are quantized signals s3 superimposed on the envelope signal s2, as shown in fig. 6.

The set of quantization step 60 and maximum search step 62 is a segmentation step 64. In practice, in step 64, the envelope signal s2 is divided into groups of repeating patterns based on the heart cycle, and a corresponding peak is associated with each group (identification and storage of clusters, described in detail below).

In step 66, the data processed in the segmentation step 64 is validated based on some rules for evaluating its authenticity (plausibility) and correlating pairs of peaks belonging to the same cardiac cycle (cycle). In fact, knowing the value of the expected period (normal value and maximum shift in the pathological case), any significantly incorrect measurements and artifacts can be discarded by discarding the untrusted peak/peak pairs, as explained in detail below.

In step 68, the searched features are calculated.

Specifically, here, the values S12, S21, and S11 are calculated based on the distance between two consecutive peaks (systole S12) within a pair of peaks, the distance between the second peak of a pair of peaks and the first peak of a pair of consecutive peaks (diastole S21), and the sum thereof (i.e., the distance between the first peaks of two consecutive peak pairs S11, which is the heart rate), respectively, as shown in the enlarged detail of fig. 7, and the ratio S12/S21 thereof is calculated. The first peak of each pair of consecutive peaks is a systole peak and the second peak of each pair of consecutive peaks is a heart Shu Feng.

Based on these values and in particular on the ratio S12/S21, the system 1 or 1' of fig. 2, 3 may highlight the risk of heart disease, step 70. Specifically, based on the detected features S12, S21, S11, the values of the following parameters: HFrEF (heart failure Heart Failure with reduced Ejection Fraction with reduced ejection fraction); HFmrEF (intermediate range ejection fraction heart failure Heart failure with Mid-Range Ejection Fraction); HFpEF (ejection fraction retaining heart failure Heart Failure with Preserved Ejection Fraction) can be determined and classified. Furthermore, the detected values may be displayed and/or transmitted to the outside, and a warning message or a perfect heart function message may be generated.

Thus, since the accelerometer is used, the ambient noise has no influence on the calculation of the characteristics S12, S21, and S11, and thus has good reliability.

The system 1,1' operates with low power consumption and can therefore be implemented by a portable device; they are also low cost.

Fig. 9-11 illustrate an embodiment of the cardiac parameter measurement method 50 of fig. 8, as described below.

Fig. 9 relates to a preprocessing flow of signal 75, providing an implementation of steps 52-56 of cardiac parameter measurement method 50 of fig. 8.

In detail, in step 80, a sample of the input signal SCG is loaded into the array D of input data, for example contained in a memory (not shown) associated with the motion sensor 2 of fig. 2 and 3. Array D stores N samples D _i 。

In step 82, sample d is processed, in particular in a band-pass filter having a heart rhythm dependent bandwidth (e.g. comprised between 1Hz and 150 Hz) _i Filtering; in step 84, the filtered sample d is filtered, for example by using a third order polynomial Sawitzky-Golay filter _i Further filtering is performed to reduce small local oscillations. Steps 82, 84 implement, for example, step 54 of fig. 8.

In step 84, the filtered sample d in step 82 _i Divided by Ma (theoretical maximum; in step 86 they are multiplied by the spreading factor f to perform normalization of the samples, then in step 88 the normalized samples d are amplified using, for example, a hyperbolic sine function _i . In this way, the signal peaks are amplified more than the lower values, and then searching for local maxima is facilitated.

Note that the theoretical maximum value Ma and the expansion factor f generally vary with the system 1,1', the physiology of the patient, and the degree of mechanical coupling of the motion sensor 2 to the patient. However, they are not critical values and may be preset and may be modified in successive calibration steps after following a command from a high-level user or after the system 1,1' has corrected the correctness/rationality of the acquired signals.

Then, in step 92, the amplified sample d _i Multiplied by a theoretical maximum value Ma and divided by the spreading factor f in step 94 to form a processed signal, again denoted d for clarity _i And (3) representing.

In effect, the set of steps 86-94 implements step 56 of enhancing the signal of FIG. 8.

Steps 86-94 of enhancing the signal are performed on all samples of the input signal SCG.

The method then proceeds to sequence search flow 100, which is described below with reference to FIG. 10.

The sequence search flow 100 implements the envelope calculation 58 and segmentation 64 (including quantization 60 and local maximum search 62) steps.

In particular, the envelope calculation step 58 is performed based on a sliding window, i.e. at the beginning, the sequence search flow 100 receives the processed sample d resulting from step 94 _i And processed with the previous M-1 normalized samples as described below.

In detail, in step 102, a counter i is initialized, for example to zero.

In step 104, the processed sample d just received _i Is loaded into a sample buffer, for example into the memory F1 of fig. 2, hereinafter referred to as processed sample memory F1. The processed sample memory F1 has M memory cells, equal to the number of samples of the sliding window.

At step 106, after verifying that the processed sample memory F1 has stored M samples (corresponding to the sliding window), all the processed samples d present in the normalized sample memory F1 (belonging to the same sliding window) are read _i The method comprises the steps of carrying out a first treatment on the surface of the In step 108, they obey the hanning window function; then, in step 110, they are subjected to an average value removing operation; and, in step 112, they are subjected to a standard deviation calculation operation (RMS calculation) which provides a standard deviation value s _i 。

Steps 108, 110 and 112 allow calculation of a plurality of standard deviation values s _i Envelope of the processed signal s1 forming and constituting the envelope signal s2, e.g.As shown in fig. 5, and implements an envelope calculation step 58.

In step 114, the standard deviation value s just calculated _j Is loaded into an appropriate standard deviation buffer, for example in the memory F2 of fig. 2, also referred to as standard deviation memory F2. The standard deviation memory F2 has N-M memory locations since the first sample loaded has a sufficient number for the processing of the observation window (sliding window).

In step 116, the standard deviation value s just calculated and stored _j Compared to a deviation threshold ths. The deviation threshold ths depends on the detected signal and may be set in a preliminary step of the heart parameter measuring method 50 in a calibration step of the system 1,1' of fig. 2 and 3.

At the standard deviation value s just calculated and stored _i Above the deviation threshold ths (output "yes" from step 116), a logic "1" is stored in the quantized value buffer, e.g. in memory F3 of fig. 2, step 118; at standard deviation s _i If the value is lower than or equal to the threshold value ths (no output from step 116), a logical "0" is stored in the quantized value memory F3, step 120.

The quantized value memory F3 has N-M memory locations.

Thus, steps 116, 118 and 120 allow for a quantization of the amplitude to be performed (corresponding to step 60 of fig. 8).

Then, in step 122, condition i is verified<Whether N-M applies, i.e. the processed signal D loaded in array D _i Whether or not the last M-1 samples (number insufficient to define a sample evaluation window) have not been processed.

In the affirmative case, YES is output from step 122, the counter i is incremented in step 124, and the method returns to step 104 to load successive normalized samples d _i 。

In the negative case, NO is output from step 122, the method proceeds to the signal maximum search step, corresponding to step 62 of fig. 8.

In practice, at step 104- At the end of the loop defined by 124, the quantized value memory F3 (FIG. 2) contains N-M digital values q of value 0 or 1 _i With a sequence of 1's between zeros (corresponding to low standard deviation values), typically short, corresponding to the time interval in which the envelope signal s2 is valid (except for the spurious values that are eliminated in the continuous process), as shown in fig. 6.

In the maximum value search step, first step 126, counter i is initialized again, for example to zero.

In step 128, the quantized value memory F3 is searched for a first quantized value qj (j > i) equal to "1". Here j actually represents the position (time) where the quantized signal s3 (fig. 6, 7) has a rising edge.

If a value j satisfying the condition of step 128 is found, YES is output from step 130, searching the same quantized value memory F3 for a first quantized value q equal to "0" _k (k>j) Step 132. In fact, here, k represents a position (time) in which the quantized signal s3 has a falling edge continuous with a previously found rising edge.

If a value satisfying the condition of step 132 is found, yes is output from step 134, the standard deviation value s stored in the standard deviation memory F2 and included in the interval between the indexes j and k _j (envelope signal s 2) and its peak amplitude p is obtained in step 136.

In step 138, the clusters formed by the peak amplitude p and the time of the peak (identified by its index j in the standard deviation buffer F2) are stored in a time memory, for example in the memory F4 of fig. 2, also referred to below as cluster memory F4. The cluster is denoted by c (p, j) in fig. 10, and the cluster memory F4 includes K memory locations, where K < M-N.

In step 140, index i is set equal to k, and the sequence search flow 100 returns to step 128 for searching for a new cluster c (p, j).

Thus, steps 116-140 allow for performing a segmentation of the sequence of standard deviation values (corresponding to step 64 of fig. 8).

If No value is found that satisfies the conditions of step 128 (output No from step 130) or step 132 (output No from step 134), i.e. No rising or falling edge is detected in the remaining part of the quantized signal s3 of fig. 6, the method proceeds to step 142, where it is verified whether the cluster memory F4 contains a minimum number of peaks (here equal to 4).

In the negative case, NO is output from step 142, the sequence search flow 100 stops, possibly sending an error message; in the affirmative case, YES is output from step 142, and the method proceeds to parameter verification/calculation routine 150, which will be described below with reference to fig. 11.

The parameter verification/calculation flow 150 of fig. 11 implements the parameter verification 66 and calculation 68 steps of fig. 8. In effect, during verification, cluster c is verified _i Certain plausibility factors are met and for these factors the method proceeds with the calculation of the parameters.

In particular, in the parameter verification/calculation flow 150 of fig. 11, verification of peak amplitude, correct systole peak/heart Shu Feng sequence and correct distance between peaks is performed in order to eliminate false peaks and eliminate segments, for example, where peaks are "lost".

In detail, in the parameter verification/calculation procedure 150, the verification peak has a minimum amplitude, the amplitude ratio between two consecutive peaks is greater than a predetermined peak threshold (i.e., the diastole peak follows the systole peak), and the distance between the two consecutive peaks is included in the tolerance window.

Specifically, in step 160 of fig. 11, the index i is initialized again, for example, to zero.

In step 162, the ith cluster (c) _i ) And successive pairs of clusters (c _i+1 )。

In step 164, cluster c is verified _i Peak amplitude p in (first peak of the pair of consecutive clusters) _i Whether a predetermined amplitude threshold thp, set for example in a calibration step, is exceeded.

In the negative case, the output NO from step 164, counter i is incremented, step 166, and it is verified whether the incremented value of counter i is below K-1 (i.e., whether there are additional clusters in cluster memory F4), step 168.

If the verification in step 168 gives a positive result (output YES), the parameter verification/calculation flow 150 returns to step 162 to analyze two other consecutive peaks, typically the second peak and subsequent peaks of the previous verification.

If the verification in step 164 gives a positive result (output YES), the peak amplitude p is verified in step 170 _i And continuous peak amplitude p _i+1 Whether the amplitude ratio between is above the peak ratio threshold thpf.

In the negative case, NO is output from step 170, two more consecutive peaks are verified, and the process returns to step 166; in the affirmative case (representing peak p _i Is a peak p that is continuous due to systole of the same cardiac cycle _i+1 Due to the fact that they are diastole), it is verified that they are placed at an acceptable distance in time. For this purpose, in step 172, a time T1 is determined by dividing the indices i and i+1 by the sampling frequency fs of the input signal SCG _i And T2 _i (referred to as the i-th peak and the successive i+1th peak, respectively).

In step 174 it is verified whether the distance between the two peaks i and i+1) is comprised between the lower distance threshold value Thtmi and the higher distance threshold value Thtma.

In the negative case, NO is output from step 174, the continuous peak is verified, and the process returns to step 166; in the affirmative case, YES is output, time pair T1 _i ，T2 _i (shown as pair T (T1 in FIG. 11 _i ，T2 _i ) For example, in the memory F5 of fig. 2, hereinafter referred to as the peak time memory F5). The peak time memory F5 has H memory locations.

Then, at step 178, counter i is incremented by one unit and parameter verification/calculation process 150 returns to step 166 to further increment counter i to search for successive peak pairs p _i And p _i+1 。

In effect, steps 160-178 of FIG. 11 implement the sequence verification step 66 of FIG. 8.

If the verification in step 168 gives a negative result (output no), that is,if no additional peak pair p is present in cluster memory F4 _i ，p _i+1 The parameter verification/calculation process 150 proceeds to step 180 where the counter i is again initialized and reset.

In step 182, the ith time pair T (T1) is read from the peak time memory F5 _i ，T2 _i )。

Then, in step 184, parameters S12, S21, and S11 of the i-th pair are calculated, for example:

S12 _i ＝T2 _i －T1 _i

S21 _i ＝T1 _i +1－T2 _i

S11 _i ＝T1 _i +1－T1 _i

and the parameters just calculated are provided for their further processing and/or display, as shown in the verification step 70 of fig. 8.

In step 186, it is verified whether all time pairs T (T1) present in the peak time store F5 have been taken into account by verifying whether the counter i has reached the last element of the peak time store F5 _i ，T2 _i ) The method comprises the steps of carrying out a first treatment on the surface of the In the negative case, YES is output from step 186, index i is incremented at step 188, and the parameter verification/calculation process 150 returns to step 182 to read successive time pairs.

If all elements of the peak time store F5 have been considered, then a "NO" output from step 186, the parameter verification/calculation process 150 stops.

As an alternative to the above, the calculated parameters may be provided for their display/processing at the end of the parameter verification/calculation flow 150.

In the case where the cardiac parameter measurement method 50 is performed by the system 1' of fig. 3, at the end of the parameter verification/calculation procedure 150, it may include steps for acquiring an electrocardiogram signal from the EI electrode 20 for further processing.

The systems and methods described herein have a number of advantages.

In particular, they allow reliable measurement of systole, diastole and cardiac cycle, since the accelerometer is not affected by external noise and can suppress noise due to respiration. In practice, this is at a very different frequency than the heartbeat and can be eliminated by an initial filtering step, thus not affecting the measurement.

The measuring system 1,1' is compact and can therefore be implemented in a portable device or device.

Furthermore, since accelerometers are already present in many wearable and portable devices, the described systems and methods allow for the addition of cardiac function analysis functions to current devices already present in the market at very low cost.

This approach has proven to be reliable and effective and can provide early warning for performing further, deeper medical analyses.

Finally, it is clear that modifications and variations may be made to the system and method described and illustrated herein, without departing from the scope of the present disclosure.

For example, considering that recent generations of accelerometers are typically equipped with algorithms for detecting activity (rest, walking, running …), embedded in the system and capable of executing the steps of the method in parallel, they may integrate the hardware and software blocks/modules (yet be encoded) for implementing the method.

Furthermore, given the ability to detect patient status (lying or standing) in the nearest accelerometer, this detection can be used to activate/suspend the measurement.

In the foregoing description, the processing and analysis of the signals is performed in a "discontinuous" mode, i.e. based on pre-acquired signals.

In this mode of operation, the processing unit 15 acts on two processes: the first procedure takes data either when possible (stationary person) or after triggering (as described above) and records it in array D for X seconds, e.g. 8 seconds; subsequently, the second process analyzes the previous X seconds of data and performs calculations in a rather short time, such as parameters found in 100ms (depending on the computing power of the processing unit 15); finally, a second process may analyze the continuous data sets present in array D.

This mode involves a trade-off of selection time: if a short acquisition time is chosen, it is possible to have a fast response, but the risk of analysis is split (the heart beat at the end of one sequence is not correlated with the heart beat at the beginning of the successive sequence, the adjacent heart beats are used to perform sequence control); conversely, if a long acquisition time is selected, it is possible to have continuous analysis of invalid values/clusters and better control, but it has a delayed response time.

Alternatively, the described streams may be modified in a simple manner so as to operate "on-line" without causing discontinuities or fragmentation of the data stream.

According to the present disclosure, according to a first example, a system for measuring cardiac parameters comprises:

a motion sensor (2) configured to generate a seismogram Signal (SCG);

a heart parameter calculation unit (15) receiving the seismogram signals and comprising:

an envelope determination device (58) configured to generate an envelope signal (s 2) related to the seismogram signal;

-a segmentation device (64) configured to identify signal segments having a repeating pattern in the envelope signal;

a verification device (150) configured to identify pairs of consecutive peaks between the signal segments such that a first peak of each pair of consecutive peaks is a systole peak and a second peak of each pair of consecutive peaks is a heart Shu Feng; and

A parameter calculation device (68, 70) configured to calculate a systole (S12) and a diastole (S21) for each pair of consecutive peaks.

The system according to example 1 may further comprise a filtering module (54), the filtering module (54) comprising a band pass filter (82), the band pass filter (82) being coupled to the motion sensor (2) and the envelope determination device (58) and configured to filter the seismogram Signal (SCG) and to generate a filtered signal (s 1).

In a system according to example 1 or 2, the envelope determination device (58) may include a processor configured to calculate a plurality of standard deviation values (s _j ) Is a computing device (86-94).

In a system according to any of the preceding examples, the segmentation device (64) may comprise a quantization module (60; 116-120) configured to generate a sequence of quantized samples (s 3), the sequence of quantized samples (s 3) having a first logic level when the envelope signal (s 2) exceeds a threshold value and having a second logic level when the envelope signal (s 2) is below the threshold value.

In a system according to the foregoing example, the segmentation device (64) may further comprise a maximum value search module (62; 128, 132, 136) configured to search for a maximum value of the envelope signal (s 2) between a switching time of the quantized samples between the first and second logic levels and a successive switching time within the quantized sample sequence (s 3), and a cluster generation module (138) configured to associate the switching time with a respective maximum value of the envelope signal (s 2) forming a cluster.

In a system according to the previous example, the maximum value search module (62; 128, 132, 136) may be configured to search for a peak start timing within the quantized sample sequence (s 3), wherein quantized samples of the quantized sample sequence (s 3) are switched from a first logic level to a second logic level, and the maximum value search module (62; 128, 132, 136) may be configured to search for a peak end timing after the peak start timing, wherein quantized samples of the quantized sample sequence (s 3) are switched from the second logic level to the first logic level, and the maximum value search module may further comprise a maximum amplitude search module (136) configured to search for a local maximum value (p) in the envelope signal (s 2) between the peak start timing and the peak end timing, and wherein the cluster generation module (138) is configured to associate the peak start timing with the maximum value of the envelope signal (s 2).

In a system according to any of the preceding examples, the verification device (150) may comprise a peak verification module (164) and a characteristic comparison module (170, 174), the peak verification module (164) being configured to verify that the value of the clustered envelope signal (s 3) exceeds a threshold value, the characteristic comparison module (170, 174) being configured to verify the characteristic (p _i 、T _i ；T _i1 、T _i2 ) A predetermined relationship is satisfied.

In a system according to the foregoing example, the continuous peak comparison module (164, 70, 172) may include at least one of:

an amplitude verification module configured to verify that a first peak of the pair of consecutive peaks has an amplitude greater than an amplitude threshold;

an amplitude ratio verification module configured to verify that a first peak of the pair of continuous peaks and a second peak of the pair of continuous peaks have an amplitude ratio satisfying a ratio relationship; and

a peak distance verification module configured to verify that a distance between a first peak of the pair of consecutive peaks and a second peak of the pair of consecutive peaks satisfies a distance relationship.

In the system according to any of the preceding examples, the parameter calculation device (68) may comprise a heart cycle determination module (S11) and a ratio determination module (70) configured to calculate a ratio (S12/S21) between systole (S12) and diastole (S21) for each pair of consecutive peaks.

In a system according to the foregoing example, the duration determination module (184) may be further configured to calculate the duration of the cardiac cycle from the duration of the systole peak and the duration of the diastole peak in each pair of consecutive peaks.

In a system according to any of the preceding examples, the motion sensor may comprise an accelerometer (2).

According to another example, a method of measuring a cardiac parameter, comprises:

generating a seismograph signal;

generating an envelope signal related to the seismograph signal;

splitting the envelope signal into signal segments having a repeating pattern;

identifying pairs of consecutive peaks between the signal segments such that a first peak of each pair of consecutive peaks is a systole peak and a second peak of each pair of consecutive peaks is a heart Shu Feng; and

systole and diastole are calculated for each pair of consecutive peaks.

The method according to the preceding example may further comprise filtering the seismogram signal with a band pass filter and generating a filtered signal, wherein generating the envelope signal comprises calculating an envelope of the filtered signal.

In a method according to the preceding example, generating the envelope signal may comprise calculating a plurality of standard deviation values (s _j )。

Furthermore, splitting the envelope signal may comprise: a sequence of quantized samples having a first logic level is generated when the envelope signal exceeds a threshold value and a sequence of quantized samples having a second logic level is generated when the envelope signal is below the threshold value.

In the method according to the previous example, dividing the envelope signal may comprise: searching for a maximum value of the envelope signal (s 2) between a switching time of the quantized samples between the second and first logic level and a successive switching time from the first to the second logic level within the sequence of quantized samples (s 3), and associating the switching time with a corresponding maximum value of the envelope signal (s 2) forming the cluster.

In a method according to any of the preceding examples, identifying pairs of consecutive peaks between the signal segments may comprise verifying that the value of the clustered envelope signal (s 3) exceeds a threshold and verifying that the characteristic (p) of the envelope signal (s 3) of the characteristic consecutive clusters _i 、T _i ；T _i1 、T _i2 ) A predetermined relationship is satisfied.

The method according to the preceding example may further comprise calculating a cardiac cycle (S11) and calculating a ratio (S12/S21) between systole (S12) and diastole (S21) for each pair of consecutive peaks.

The method according to the preceding example may further comprise calculating a ratio between the duration of the systole peak and the duration of the diastole peak, and generating the heart health information based on the ratio.

The various embodiments described above may be combined to provide further embodiments.

These and other changes can be made to the embodiments in light of the above-detailed description. In general, in the following claims, the terms used should not be construed to limit the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the present disclosure.

Claims (20)

1. A system for measuring cardiac parameters, the system comprising:

A motion sensor configured to generate a seismogram signal; and

a processor configured to:

receiving the seismograph signals;

generating an envelope signal cross-correlated with the seismograph signal;

identifying signal segments in the envelope signal having a repeating pattern;

identifying pairs of consecutive peaks between the signal segments such that a first peak of each pair of consecutive peaks is a systole peak and a second peak of each pair of consecutive peaks is a heart Shu Feng; and

systole and diastole were calculated for each pair of consecutive peaks.

2. The system of claim 1, further comprising:

a band pass filter configured to filter the seismographic signals and generate a filtered signal.

3. The system of claim 1, wherein the processor is configured to calculate a plurality of standard deviation values.

4. The system of claim 1, wherein the processor is configured to generate a sequence of quantized samples having a first logic level when the envelope signal exceeds a threshold and a second logic level when the envelope signal is below the threshold.

5. The system of claim 4, wherein the processor is configured to:

Searching for a maximum value of the envelope signal between a switching time and a successive switching time of the quantized samples between the first logic level and the second logic level within the quantized sample sequence; and

the switching times and the successive switching times are associated with respective maxima of the envelope signals forming clusters.

6. The system of claim 5, wherein the processor is configured to:

searching for a peak start timing and a peak end timing within the quantized sample sequence, a quantized sample of the quantized sample sequence switching from the first logic level to the second logic level at the peak start timing, and a quantized sample of the quantized sample sequence switching from the second logic level to the first logic level after the peak start timing and at the peak end timing;

searching for a local maximum in the envelope signal between the peak start timing and the peak end timing; and

the peak start timing is associated with the local maximum of the envelope signal.

7. The system of claim 5, wherein the processor is configured to:

Verifying that the value of the envelope signal of the cluster exceeds a threshold; and

verifying that characteristics of the envelope signals of successive clusters satisfy a predetermined relationship.

8. The system of claim 1, wherein the processor is configured to calculate a cardiac cycle and calculate a ratio between the systole and diastole of each pair of consecutive peaks.

9. The system of claim 1, wherein the motion sensor comprises an accelerometer.

10. A method for measuring cardiac parameters, comprising:

generating a seismograph signal;

generating an envelope signal cross-correlated with the seismograph signal;

splitting the envelope signal into signal segments having a repeating pattern;

identifying pairs of consecutive peaks between the signal segments such that a first peak of each pair of consecutive peaks is a systole peak and a second peak of each pair of consecutive peaks is a heart Shu Feng; and

systole and diastole were calculated for each pair of consecutive peaks.

11. The method of claim 10, further comprising: filtering the seismocardiographic signals with a band pass filter and generating a filtered signal, wherein generating the envelope signal comprises calculating an envelope of the filtered signal.

12. The method of claim 10, wherein partitioning the envelope signal comprises generating a sequence of quantized samples having a first logic level when the envelope signal exceeds a threshold and a second logic level when the envelope signal is below the threshold.

13. The method of claim 12, wherein partitioning the envelope signal comprises searching for a maximum value of the envelope signal between a switching time of the quantized samples between the second logic level and the first logic level and a consecutive switching time from the first logic level to the second logic level within the sequence of quantized samples, and associating the switching time and the consecutive switching time with respective maximum values of the envelope signal forming a cluster.

14. The method of claim 10, wherein identifying pairs of consecutive peaks between the signal segments comprises verifying that values of the envelope signals of the clusters exceed a threshold value, and verifying that characteristics of the envelope signals of consecutive clusters satisfy a predetermined relationship.

15. The method of claim 14, further comprising calculating a cardiac cycle, and calculating a ratio between the systole and diastole of each pair of consecutive peaks.

16. An apparatus, comprising:

a sensor configured to generate a seismogram signal; and

a processor configured to:

generating an envelope signal of the seismograph signal;

splitting the envelope signal into signal segments having a repeating pattern;

Determining successive pairs of peaks in the signal segment, respectively; and

systole and diastole are determined based on the pairs of consecutive peaks.

17. The apparatus of claim 16, wherein each pair of consecutive peaks comprises a systole peak and a heart Shu Feng.

18. The apparatus of claim 16, wherein the systole is determined based on a distance between a first peak and a second peak of the pair of consecutive peaks.

19. The apparatus of claim 16, wherein the heart Shu Qiji is determined by a distance between a peak of a first of the pairs of consecutive peaks and a peak of a second of the pairs of consecutive peaks.

20. The apparatus of claim 19, wherein

The systole is determined based on a first distance between a first peak and a second peak of the pair of consecutive peaks,

the core Shu Qiji is determined at a second distance between the peak of the first one of the pairs of consecutive peaks and the peak of the second one of the pairs of consecutive peaks, an

The processor is configured to determine a heart rate based on a sum of the first distance and the second distance.