WO2017208204A1 - Method and device for detecting apnoea - Google Patents

Abstract

The invention relates to a method and non-invasive device for detecting apnoea, the device comprising: a) means for detecting EMGdi, FC and SpO2 signals; b) a module for conditioning the EMGdi, FC and SpO2 signals, which is connected to the means for detecting the EMGdi, FC and SpO2 signals; c) a signal processing module connected to the module for conditioning the EMGdi and FC signals; d) a processor; e) an analogue-to-digital converter connected to the signal processing module and to the processor; f) a memory connected to the processor; and g) a power source that powers the mentioned components. The method comprises two steps, a step of training and a step of detecting. During the step of training, the processor calculates the data confidence limits in an interval, using the data extracted from the EMGdi, FC and SpO2 signals and stored in the memory. During the step of detecting, the data stored in the memory and extracted from the EMGdi, FC and SpO2 signals are compared to the data confidence limits calculated in the training step. According to the result of the comparison, the detection of obstructive, central or mixed apnoea is indicated, the presence of apnoea being detected early and confirmed by means of SpO2.

Description

 METHOD AND DEVICE FOR APNEA DETECTION

Field of the Invention

The present invention is related to the methods and devices for the detection of apnea by means of the processing of electromyography signals of the surface of the diaphragm muscle, heart rate and is confirmed by the oxygen saturation of the hemoglobin in blood measured from the pulse oximetry , Sp02.

Description of the state of the art.

The definition of apnea is considered as an episode of absence of respiratory flow lasting more than 20 seconds, or of shorter duration, but accompanied by cardiovascular repercussion, in particular bradycardia (less than 100 beats per minute in neonates) and hypoxemia (low concentration of oxygen in the blood). The type of apnea can be classified according to the production mechanism as: Central, Obstructive or Mixed. The first designed apnea detection systems use variables managed by artificial ventilators, such as airflow in the airways or change in the volume of the rib cage in the lungs, and variables related to the behavior of the heart, but These are indirect measurements that may not be expeditiously related to respiratory activity.

The differentiation of the different apnea episodes has been reported in the state of the art through non-invasive methods but associated with artificial ventilators or using direct methods such as capnography, but as in the latter, delays may occur. due to the time it takes for the gas to reach the sensor. The monitoring of diaphragmatic activity through surface electromyography (EMGdi) has been used in the detection of apnea and other respiratory disorders; This method has been mainly used in the detection of sleep apnea. The use of the EMGdi method, with sensors located in the chest brings interference by the electrical activity of the heart, especially caused by the QRS complex.

Several methods for the elimination of the electrocardiogram component present in the electromyographic signal of the diaphragm (EMGdi) are reported in the state of the art. For example, in the article by Corne et al., [1] after obtaining the electromyographic signal of the diaphragm by means of an esophageal probe specially designed for this, the QRS complex (usually 60 ms to 100ms in duration) was initially removed from the signal. The electromyographic signal was then rectified. The 100ms segment removed was replaced with a line that begins with the average electromyographic signal 40ms before the QRS complex, and ends with the average electromyographic signal 40ms after the QRS complex. The electromyography signal was subsequently averaged. The resulting signal called the diaphragm electromyographic signal is of good quality and shows a minimum electrocardiogram content (mainly P wave residues). In 1981, Leo van Eykern patented through US4248240 A a method for electromyographic signal of the transcutaneous diaphragm that works in a similar way. There are significant differences between the two methods, such as delays in processing time.

The use of a non-invasive method of EMGdi is also reported in WO2005096924 A1, this describes the detection of respiratory disorders by processing surface signals of EMGdi, and the elimination of the QRS complex. The elimination of the QRS complex is reported through the calculation of an average amplitude of the QRS complex, in which an envelope of the EMGdi signal is not obtained, nor is there an algorithm that customize the patient's respiratory activity through a previous training time.

A device was not found in the state of the art that alone had the characteristics that are integrated in the present invention and that correspond to:

Elimination of the electrocardiogram wave, present in the EMGdi signal in real time, with the subsequent obtaining of an envelope that resembles the patient's respiratory pattern.

Ability to detect central, obstructive and mixed distress through a non-invasive and indirect method. The central one due to a decrease in the electrical activity of the diaphragm, and the obstructive one indirectly due to an increase in respiratory efforts reflected in a considerable increase in the EMGdi signal; this in addition to taking into account the direct relationship between this signal and a possible bradycardia estimated from the patient's heart rate, and this variable is included in the detection algorithm, which reduces the percentage of false apnea detected by the prototype .

Customization in the detection process from a previous training time, which allows adaptation to the patient's physiology.

 Integration of several detection methods, guaranteeing greater reliability in the results, including the measurement of hemoglobin oxygen saturation in blood from pulse oximetry, Sp02, to corroborate the presence of apnea.

It is necessary to have a device that, in addition to presenting good performance and reliability, is non-invasive and directly related to the respiratory effort signal, which eliminates the ECG wave in real time and generates an envelope of the EMGdi signal; that identifies both central, obstructive and mixed apnea; that has a personalized analysis through an algorithm of training and related to the detection of bradycardia. In addition, that can be used even in premature infants.

Brief description of the graphics. Fig. 1 corresponds to the connection of the three electrodes to a pre-amplification stage.

Fig. 2 corresponds to a diagram of the isolation amplifier of the amplification unit of the signal conditioning module

Fig. 3 corresponds to the circuit diagram and the seventh order transfer function for the Butterworth low-pass filter.

Fig. 4 corresponds to the circuit diagram and the seventh order transfer function for the Butterworth high-pass filter.

Fig. 5 corresponds to the amplification circuit diagram of the filter unit signal. Fig. 6 corresponds to the circuit diagram of a precision full wave rectifier.

Fig. 7 corresponds to the comparator circuit diagram of the signal processing module.

Fig. 8 corresponds to an example of step iii of step a) with 251 time windows of 2 seconds.

Fig. 9 corresponds to the range of confidence limits for the EMGdi signal (a), the range of confidence limits for heart rate (b) for a particular example. Brief Description of the Invention

The present invention relates to a non-invasive device for the early detection of the presence of apnea comprising: EMGdi and FC signal detection means, an EMGdi and FC signal conditioning module, connected to the detection means of EMGdi and FC signal, and Sp02 detection medium; a signal processing module connected to the EMGdi and FC signal conditioning module; a processor; an analog to digital converter connected to the signal processing module and the processor; a memory connected to the processor, to store data and instructions; a power source that provides power to the previous components.

In one embodiment of the invention The EMGdi and FC signal detection means are three surface electrodes with precision modular connectors, Push-Pull type interlocking mechanisms and shielding protection rings for electromagnetic compatibility. Two electrodes are located in the costal margin of the person under observation, one on the left side and the other on the right side of the body, in a line drawn vertically from the nipples of the person under observation, between the sixth and seventh intercostal region; and the third electrode is located in the sternum.

The method for the detection of obstructive, central and mixed apnea consists of two stages: a stage (a) of training and a stage (b) of detection of apnea.

The training stage (a) consists of the following steps; i) acquire the signals of EMGdi, FC, ECG and Sp02 for a time T1; ii) adapt the EMGdi signal so that the result of the processing is a respiratory pattern in the form of a periodic wave; iii) extract the characteristics in the time domain area under the curve, mean ^ and standard deviations (o) for the EMGdi, FC and Sp02 envelope signals; iv) store these characteristics in a memory; and v) calculate the limits of Confidence of stored characteristics taking into account that the minimum number of samples is met according to the central limit theorem, achieving an interval between an upper limit and a lower limit such that the ratio ≤ z≤ μ ₂ ) = ( 1 - a) ^■ 100

The stage (b) of apnea detection consists of the following steps; i) repeat steps i), ii), iii) of step (a) continuously every time T2; ii) store the characteristics obtained in step i) of step (b) in a memory every time T2; iii) compare the characteristics stored in step ii with the confidence interval calculated in step v) of step (a ) every time T2; and, iv) activate the apnea episode alarm if the characteristics compared in step ii above differ from the confidence interval calculated in step v of step (a). In one embodiment of the invention there is a device for displaying the envelope of the EMGdi signal; means for providing a condition message of the person under observation in the event of an apnea episode, said means are connected to the processor; an alarm indicating the state of apnea connected to the processor; Y; Sp02 measuring means, a conditioning module of the Sp02 measuring signal connected to the Sp02 measuring means and means for generating a patient ventilation control signal from the apnea condition connected to the processor. Detailed description of the invention

The present invention corresponds to a method and a non-invasive device for the early detection of the presence of apnea, the device comprises: EMGdi, FC and Sp02 signal detection means; -a signal conditioning module of EMGdi, FC and Spo2 connected to the signal detection means of EMGdi, FC and

Sp02;

 -a signal processing module connected to the EMGdi, FC and Sp02 signal conditioning module;

 -a processor;

 - an analog to digital converter connected to the signal processing module and the processor;

 -a memory connected to the processor; Y,

 -a power source, for example a power source of medical category, which provides power to the above components;

In one embodiment of the invention there is a device to visualize the envelope of the EMGdi signal and the condition of the person under observation;

In the following, each of the parts comprising the invention is described in detail.

EMGdi, FC and Sp02 signal detection means:

The EMGdi, FC and Sp02 signal detection means are responsible for capturing the EMGdi, FC and Sp02 signals of the person under observation.

In one embodiment of the invention, the EMGdi and FC signal detection means are three surface electrodes with precision modular connectors, push-pull interlocking mechanisms and shield protection rings for electromagnetic compatibility. Sp02 signal detection means for example a conventional pulse oximetry clamp that has two LEDs, one infrared and one red and one photodiode to capture the light coming from the LEDs; This clamp is connected to the processor module through a conventional DB9 type connector. Other means of detection of Sp02 are, for example, spectrophotometric measurement meters, and detection means based on photoplethysmography-based measurement (PPG).

The installation of the electrodes is carried out preferably but not exclusively as follows: two electrodes are located on the costal margin of the person under observation, one on the left side and the other on the right side of the body, in a line drawn vertically from the nipples, between the sixth and seventh intercostal region; and the third electrode is located in the sternum.

The way to install the electrodes is chosen due to the ease of replacing them during the measurements, the signal is minimally contaminated by the electrical activity of the intercostal muscles during tidal breathing and the place is close to the insertion of the costal fibers of the diaphragm along the costal margin, obtaining a clear respiratory EMGdi signal during inspiration. Precision modular connectors feature highly durable and polarized pull protection rings to prevent incorrect connections. The metal fabrication of the body ensures EMCdi shielding by electromagnetic compatibility, while the push-pull design saves space and eliminates the risk of accidental cable disconnection due to shock, vibration or cable pull.

In one embodiment of the invention a pair of multipolar (5-way) self-locking connectors with polarization key is used. Self-locking is carried out by means of a simple axial thrust on the Male / Female connectors. The disconnection is carried out by means of axial traction on the outer body of the connector, releasing the interlocking and thus allowing the disconnection of the Male / Female assembly. An aerial male connector and a female connector are used PCB mounting with PEEK insulating material for a cable size of 4.1 to 5mm.

In an embodiment of the invention according to Fig. 1 to make the connection of the three electrodes E1, E2, E3 to a pre-amplification stage This circuit reduces the effective resistance of the electrode by several orders of magnitude, and allows only one amount of safe current flows through it.

In one embodiment of the invention, the Sp02 signal detection means is performed by installing the pulse oximetry clamp on a patient's limb, in a slightly translucent region such as a finger, nose, ear or foot in the case of neonatal patients.

EMGdi and FC signal conditioning module:

The EMGdi and FC signal conditioning module is connected to the EMGdi and FC signal detection means. The EMGdi and FC signal conditioning module comprises an amplification unit connected to a filtering unit.

-a instrumentation amplifier with gain between 1 and 15 connected to its output with an isolation amplifier;

 -a isolation amplifier with a gain between 1 and 20 -a variable resistance of offset voltage adjustment connected to the input of the amplification stage;

 -a decoupling capacitor connected to the output of the instrumentation amplifier;

-a decoupling capacitor connected to the output of the isolation amplifier; Y, -a pass-band filter with lower cutoff frequency between 0.1 Hz and 0.5Hz and higher cutoff frequency between 800Hz and 900Hz connected to the isolation amplifier output.

The EMGdi and FC signal conditioning module also comprises an amplification unit connected to a filtering unit. The isolation amplifier of the amplification unit of the EMGdi and FC signal conditioning module has a gain between 1 and 20 gain; The main purpose of the isolation amplifier is to break the ohmic continuity of the input signal with the processing stages, minimizing the risk of electric shock.

In one embodiment of the invention according to Fig. 2, the isolation amplifier corresponds to an integrated AD210 of Analog Devices and the gain of the isolation amplifier is 11, which is obtained from the following formula:

Gisoi = = ¹¹ ; ^R yes =

loofcn, R_S2 = íowi

loofcn, R _S2 = íowi

In this mode the isolation amplifier is powered by a regulated 15V DC source using a medical source; The isolation amplifier provides up to 3500V of insulation, a value that describes the amount of voltage that the insulation barrier can withstand without deteriorating. It also has a bandwidth of 20kHz and gains of 1 to 100 can be obtained using an external resistance configuration.

In an embodiment of the invention, an instrumentation amplifier with a gain of between 1 and 15 connected to its output is connected to the isolation amplifier; In one embodiment of the invention there is a variable resistance of offset voltage adjustment connected to the input of the instrumentation amplifier. This is because the amplifiers have offset voltages associated with both the input (V _I0S ) and the output (V _oos ) thereof, variable resistors are added for the offset voltage adjustment. It is important to adjust the offset voltage to the amplifier input to avoid unwanted signals. Temperature variations will also cause changes in offset voltage. Likewise, the value of the input offset voltage is directly related to the gain by means of the general offset voltage equation referred to the output: V _os {RTO) = (and _10S XG) + V _oos

In one embodiment of the invention, a maximum offset voltage of 100μν and 6mV is presented at room temperature for the input and output, respectively, so that an offset adjustment is made for both the input and output.

After the isolation amplifier, the conditioning and filtering of the signal is carried out by means of a filtering unit where the ECG and EMGdi signals are separated, to only take the frequency range in which the greatest amount of signal information is contained EMGdi, this will eliminate components that act as noise in the signal.

The filtering unit comprises: a first channel that connects to an active high-pass filter with a cut-off frequency of 63Hz connected to the output of the isolation amplifier of the amplification unit of the signal conditioning module;

 - a second channel that connects to an active low-pass filter with a cut-off frequency of 40Hz connected to the output of the isolation amplifier of the amplification unit of the signal conditioning module; this second channel serves to obtain the ECG signal that will serve as a starting point for the elimination of the QRS complexes from the EMGdi signal;

 - a signal amplification circuit with a gain of 1 to 15 and an offset voltage of 1V to 2V connected to the filter outputs;

 - an anti-aliasing filter between 800Hz and 900Hz; Y,

 - a protection voltage limiter with a value of 3.6V, connected to the signal amplification circuit and to the input to an analog to digital converter.

The filtering unit of the signal conditioning module is a pass-through filter with cut-off frequencies between 13Hz and 25Hz.

The purpose of the active pass-band filter is to let the QRS complex pass, and then eliminate the ECG component that is present in the output signal of the high-pass filter of the filtering unit. It is of importance of the phase delay after the filters since synchronization is required to ensure that the segment of ECG that is removed from the signal from the high pass filter corresponds to the QRS complexes detected by the pass filter.

In one embodiment of the invention According to Fig. 3 and Fig. 4, a circuit is provided that provides the seventh order transfer function for the filtering unit. According to Fig. 3 the filter a low pass and According to Fig. 4 the filter passes high of the filtering unit are Butterworth type of seventh order and is constructed in such a way that the circuit transfer function has a Butterworth polynomial in its denominator, cascading circuits based on operational amplifiers, each of which must provide one of the necessary factors for the standardized Butterworth polynomial of so that w _c = 1 rad / s for a filter of order n = 7 is represented by the expression:

(s + l) (s ² + 0.445s + l) (s ² + l, 247s + l) (s ² + l, 802s + 1)

In the case of the seventh order filter, a first order filter and three second order filters are cascaded, however, although in this case the first order is passive with a follow-up amplifier.

To calculate the resistance and capacitor values of each of the filters connected in cascade, the transfer function of the filtering unit must be known, in this embodiment of the invention the following transfer functions are reached:

For the first order low pass filter, you have to:

For the first order high pass filter: s

 H (s) =

For Butterworth low-pass filters of order 2 you get:

H (s) = with [a, b] = [2, 3], [4, 5], [6, 7]

For Butterworth high-pass order 2 you get: H (s) = with [a, b] = [2, 3], [4, 5], [6, 7]

Frequency and magnitude scale change techniques are used to design the different filters with a cutoff frequency other than 1 rad / s, taking into account that ω _ε = 2nF _c . The values after the change of scale corresponding to the premium signed variables R 'and C, depending on the original values, the unsigned premium variables R and C present in the following expressions. R '= k _m R

C = c ^' f ω ₀ 1'

Being k _m and kf the magnitude and frequency scaling factors respectively.

Knowing the transfer functions of each circuit in terms of resistors and capacitors, and also knowing the Butterworth polynomial, the respective denominators are matched and one obtains: an equation with two unknowns for the filters of order one, pass-ups and pass- low, and two equations with four unknowns for each of the following sub-circuits. The fixed capacitor values are assumed and the resistance values are obtained, taking into account the frequency scaling factor that is kn = 251.33 and kn = 395.84 to shift the cutoff frequency to 40Hz and 63Hz, respectively. The theoretical values obtained can be seen in table 1 and table 2.

Table 1. Theoretical values of resistors and capacitors for the filter

 Butterworth low-pass with Fc = 40Hz.

251.328 1.802 5.50E-06 3.30E-07 409.16 21317.85 40.00

251.328 1.802 5.50E-06 3.30E-07 409.16 21317.85 40.00

Table 2. Theoretical values of resistors and capacitors for the filter

 Butterworth high-pass with Fc = 63Hz.

Se aproximan los valores obtenidos de las resistencias a valores comerciales según las tablas 3 y tabla 4, y se encuentra la frecuencia de corte de cada sub- circuito con estos nuevos valores.

The values obtained from the resistance to commercial values are approximated according to tables 3 and table 4, and the cut-off frequency of each sub-circuit is found with these new values.

Table 3. Actual values of resistors and capacitors for the filter

After separating the ECG and EMGdi signals by means of the seventh-order Butterworth filters, a DC voltage is amplified and added by means of the signal amplification circuit of the filtering unit.

In an embodiment of the invention according to Fig. 5 for the implementation of the signal amplification circuit of the filtering unit, the integrated LM358 is used, which operates on an individual supply (V ⁺ -GND), thus rejecting the negative voltage The DC voltage adder raises the offset level of the signal to 1.55V, ideal for a high performance microprocessor or a microcontroller After this circuit there is a zener diode of protection with a value of 3.6V, preventing the passage of any higher voltage and damage of the microprocessor or microcontroller. In addition, there is a low-pass filter with Fc = 868.27Hz, which helps to avoid aliasing in the sampling and analog-digital conversion of the signal.

According to Fig 5 the characteristic equation, from Ri to the output of the operational amplifier, is:

According to Fig 5, the series capacitor at the input is used to eliminate the DC voltage before the operational amplifier, in this way it is ensured that the offset that the signal will have at the output of the circuit is stipulated according to the resistance configuration in the characteristic equation. The gain of this circuit is 10.

In one embodiment of the invention in the filtering unit, a combination of seventh-order high-pass and low-pass filters is changed to a combination of second and fourth-order filters, pass-bands and high-pass filters, respectively looking for reduce the remainder of the P wave of the ECG and decrease the delay in the processing of the EMGdi signal. Signal processing module

The signal processing module is responsible for the elimination of the QRS complex present in the EMGdi signal, and obtains an enveloping signal that resembles the patient's respiratory pattern, that is, it resembles a respiratory pattern in the form of a periodic wave.

The ECG elimination form is based on the detection of the QRS complex and the elimination of the sector comprising the QRS complex in the EMGdi signal. The signal processing module comprises: -an ECG elimination unit;

 -a QRS complex detection unit connected to the ECG elimination unit; Y,

 - an EMGdi envelope detection unit connected to the ECG elimination unit;

The signal processing module is connected via two channels to the electromyography signal conditioning module, where a first channel connects the filtering unit of the signal conditioning module to the ECG elimination unit, and a second channel connects the unit of filtering of the signal conditioning module with the QRS complex detection unit. The ECG elimination unit and the QRS complex detection unit of the signal processing module comprises:

-a precision full wave rectifier;

 -a passive low pass filter with a cut-off frequency between 5Hz and 10HZ connected to the precision full-wave rectifier;

-a high-pass filter with a cut-off frequency between 0.1 Hz and 0.5Hz connected to the passive low-pass filter output with a cut-off frequency between 5Hz and 10HZ;

 -a level comparator connected to the filter output passes low with cutoff frequency between 5Hz and 10HZ; Y,

- EMGdi signal precision amplifier in non-inverting configuration and variable gain connected to the input of the full-wave rectifier; where the level comparator provides a voltage pulse as a trigger signal to operate a bilateral switch that controls the passage of the EMGdi signal. In one embodiment of the invention according to Fig. 6, the precision full-wave rectifier, also called the absolute value circuit, is constructed by adding the output of a half-wave rectifier. After the rectifier, a 10Hz passive low pass filter is connected with a 15ms time constant, which gives the signal envelope, and another 0.16 Hz high pass filter that eliminates the low frequency component that can interfere with the true pulse size. The level comparator functions as a threshold detector for the removal of the ECG component from the input signal to the comparator.

In the invention there is a detector immediately after the ECG elimination unit, which identifies the larger wave of the QRS complex normally the R in the classical form of the ECG signal and sends a pulse train to the processor to calculate the heart rate

In one embodiment of the invention according to Fig. 7 the comparator of the signal processing module is an analog-digital circuit where the digital type output signal of this comparator is determined by the analog type input signal, and the The status of the output signal has a low or high digital value that depends on the amplitude of the input signal with respect to a reference level. In this embodiment of the invention the reference signal is given by the voltage divider determined by Ri and R2, and the high and low values of the signal will be the positive and negative feeds of the integrated respectively.

According to Fig. 7 In one embodiment of the invention this detector is in a non-inverting configuration to compare the input signal to comparator V¡ with a voltage determined by resistors R1 and R2. It is recommended that the resistance R3 be 15kQ. In one embodiment of the invention, the integrated one that eliminates the sector of the EMGdi signal that contains the QRS complex is a bilateral CD4066 switch, which receives the comparator's pulse as a control or trigger signal, which activates or deactivates the passage of the EMGdi signal.

The level comparator connects an EMGdi signal amplification stage with an amplifier in configuration as a non-inverter. This amplification is variable and is before an absolute value circuit or rectifier. The objective is to deliver a signal with the amplitude that allows to obtain an envelope to be taken to the processor.

In one embodiment of the invention, the absolute value circuit is exactly the same as that used for ECG rectification, but the envelope is obtained by means of a second-order active low-pass filter, with an Fc = 0.72Hz. Finally, the offset of the signal is eliminated with a passive high pass filter with an Fc = 0.048Hz.

The EMGdi envelope detection unit of the signal processing module comprises:

- an envelope detector consisting of an active low-pass filter, with a cut-off frequency between 0.5Hz and 1 Hz. whose input is connected to the precision full-wave rectifier output;

 - a signal offset unit with a passive high pass filter with a frequency between 0.01 Hz to 0.1 Hz; Y,

- a detector in non-inverting configuration connected to the active low-pass filter with a cut-off frequency between 35Hz and

45Hz Processor.

The processor operates in training mode and in apnea episode detection mode, where in the training mode the normality of the EMGdi, FC and Sp02 parameters is identified, and in detection mode the abnormality of the parameters to be detected is identified and classify apnea episodes, indicating these episodes as an alarm signal.

The processor during a training stage, calculates, from data extracted from the EMGdi, Sp02 and FC signals stored in memory during the training stage, in a range the confidence limits of the data, and during a stage of detection, compares the data stored in the memory extracted from the EMGdi, Sp02 and FC signals during the detection stage with the confidence limits of the data calculated in the training stage according to the comparison result, apnea detection is indicated obstructive, central or mixed performing early detection of the presence of apnea.

In one embodiment of the invention the processor is a Microprocessor: Rabbit 4000.

In one embodiment of the invention, the processor is connected to a means to generate a patient ventilation control signal or an alarm signal from the apnea condition.

In one embodiment of the invention, measuring means of Sp02 are counted, a conditioning module of the measuring signal of Sp02 connected to the measuring means of Sp02 and connected to the processor the apnea condition is confirmed by the processor by the concentration of blood hemoglobin oxygen, measured with the Sp02 measuring means. In one embodiment of the invention, the processor is connected to means for generating a patient ventilation control signal or an alarm signal from the apnea condition.

Analog to digital converter

The EMGdi signal previously processed by the hardware of the invention is recorded by means of an A / D channel of the analog to digital converter and a square signal with frequency equal to the heart rate is separated, which is recorded by means of the processor

The analog to digital converter is connected to the signal processing module and to the processor.

When considering an amplified EMGdi signal that can be represented by s (t), which is sampled over a time interval T. the analog-to-digital converter provides a discrete signal that is represented as, x (nT) for n = (1 .... N), over the sampling period Λ / T of the analog to digital converter.

In one embodiment of the invention The sampling frequencies are generally 1 kHz, therefore using a 250ms analysis window results in an input vector with a dimension of 250 for each channel.

Memory.

The invention has a memory connected to the processor, to store data and instructions.

In one embodiment of the invention the memory is 512k of flash memory, 512k of SRAM memory. Device to visualize the envelope of the EMGdi signal.

In one embodiment of the invention there is a device for displaying the envelope of the EMGdi signal; means for providing a condition message of the person under observation in the event of an apnea episode, said means are connected to the processor;

In one embodiment of the invention there is a device for displaying condition messages of the person under observation if an apnea episode occurs, this display device is connected to the processor.

In one embodiment of the invention an alarm indicating the state of apnea connected to the processor; and, a power source that provides power to the previous components.

In one embodiment of the invention the device for display is a reference display TG12864B-05 has a pixel resolution of 128x64 and BackLight.

Method for the detection of obstructive, central and mixed apnea

The method for the detection of obstructive, central and mixed apnea characterized by the stages of: a stage (a) of training and a stage (b) of apnea detection.

It uses control charts also called hypothesis tests in the apnea detection algorithm that includes the physiological variable, heart rate and oxygen saturation, since a consequence of an apnea episode is a noticeable change in the frequency of beats of the heart and a decrease in blood oxygen saturation. The stage (a) of training that consists in establishing comparison characteristics in the time domain through the steps of: i. acquire for a time T1 the signals of EMGdi, FC, and Sp02;

 I. adapt the EMGdi signal so that the result of the processing is a periodic wave-shaped respiratory pattern;

 Neither. extract the following characteristics in the time domain: area under the curve, mean ^ and standard deviations (o), for the EMGdi, FC and Sp02 envelope signals;

iv. store these features in a memory; and V. calculate in a range the confidence limits of the stored characteristics taking into account that the minimum number of samples is fulfilled according to the central limit theorem achieving an interval between an upper limit and a lower limit such that the ratio Ρ ( μ _χ ≤ z≤ μ ₂ ) = (1 - a) ^■ 100; The training stage is implemented to make the personalized equipment, since the signals coming from a person under observation have different EMGdi characteristics in relation to other people under observation and also have a range of heart rates that can be stable for one person and for another no.

Step ii of step (a) comprises the sub-steps of: to. extract the surround signal from the EMGdi signal;

 b. remove from the ECG component of the EMGdi signal; C. obtain a signal indicating the patient's HR;

 d. reconstruct the ECG signal eliminated sub-step b;

 and. remove from the QRS complex present in the EMGdi signal; and f. Acquire the Sp02 signal.

In step iii of step a) Extract the characteristics in the time domain area under the curve, mean and standard deviations (or) for the signals; EMGdi, FC and Sp02 envelope. The area under the curve: It is the sum of the heights of the wave in a segment of time given the expression:

^A _t = ∑ \ \

In one embodiment of the invention, the characteristics in the time domain are calculated every 2 seconds and stored in different vectors for each signal, the result of each time window of 2 seconds occupies a column of the vector, obtained at the end of training 2 vectors each of the vectors for each characteristic with 45 positions. Step iii of step (a) comprises the sub-steps of a. extract the area under the curve of the EMGdi envelope from each time T2;

 b. calculate for the ECG signal present in the EMGdi signal, the average time between pulses of the ECG signal obtained every T2 seconds; Y,

C. calculate the means (μ) and the standard deviations (o) for the EMGdi, FC and Sp02 envelope signal each time T2. In step v of step (a) it is carried out in a normal distribution defining a 95% confidence level, so that the probability for the interval meets the expression P (X - 1.96 * σ≤ μ≤ X + 1.96 * σ). In one embodiment of the invention, the training stage (a) takes a time of 1 minute and 30 seconds, during which time it performs the following processes: these 90 seconds are divided into 2-second windows giving a total of 45 time windows. The training stage (b) is carried out through the steps of: i. repeat steps i, i, iii of step (a) continuously every time T2;

 I. storing the characteristics obtained in step i above in a memory every time T2;

 Neither. compare the characteristics stored in step ii with the confidence interval calculated in step v of the stage

(a) every time T2; Y,

 iv. activate apnea episode alarm if the characteristics compared in step iii above differ from the confidence interval calculated in step v of step (a).

Step iv of step (b) comprises the sub-steps of: i. activate an obstructive apnea episode alarm if the characteristics compared in step iii of stage (b) is greater than the upper limit of the range established in step v of stage (a);

I. activate a central apnea episode alarm if the characteristics compared in step iii of stage (b) are less than the lower limit of the range established in step v of stage (a); Y, iii. activate mixed apnea episode alarm if the characteristics compared in step iii fluctuate above and below the upper and lower limits set in step v of step (a).

In step (b) steps iii) to iv) are performed by comparing the values of the EMGdi, FC, and Sp02 envelope signals, with preset limits in case of exceeding these limits, the apnea alarm is activated.

In one embodiment of the invention, the processor declares an apnea episode when two physiological variables exceed the confidence limits in two consecutive time windows of 4 seconds, since the probability of false alarms is almost zero, that is, 4.81%, according to the results produced by the statistical evaluation of the invention.

As an example to achieve a better understanding of how the training is carried out, data from two time windows of 2 seconds each are used, and the characteristics of step iii of stage a) are extracted, yielding the values shown in table 5.

Table 5. Features for two time windows.

Features Window Window Window Window

 1 (EMGdi) 2 (EMGdi) 1 (ECG) 2 (ECG)

T (ms) = ^{tl + t2 + tn ~ 1 + tn}

 136 126

 Heart Rate (HR) =

60000 With these results, step Ni of stage a) is responsible for calculating the means (μ) and standard deviations (o) for each ECG and EMGdi signal, as follows: μ ^ = x = —∑¿ _{= 1} j

¿ = 75.7815 For this example μ _Λ =

= 75.7815

For this e ⁼ 25.22

σ _Λ = ^ ∑ ² _{= 1} (x _¿ - x) ² = 7.07

 Obtained the values for the means and the variances,

step v of step a) calculates an interval of confidence limits taking into account that the minimum number of samples is met according to the central limit theorem. These limits are defined as follows: in samples with mean μ and standard deviation σ, samples of 45 elements are taken and each of these samples in turn has a mean x, and using the central limit theorem you can show that = μ. At the same time, as the number of samples is greater than 30, the distribution of sample means is practically a normal distribution, with mean μ and standard deviation given by the expression% = It is therefore true that:

§ ^ - = Z ~ N (0,1)

In a normal distribution can be calculated in a range falling percentage of observations, such as ≤ z≤ μ ₂₎ = (1 - a) ^■ 100. For this study a confidence level of 95% is defined as which generates the points that delimit the probability for the interval P (X - 1.96 * σ≤ μ≤ X + 1.96 * σ) When using a 95% confidence level, you have a 5% error percentage, a very high percentage if in only 200 seconds the device can detect 5 false apneas, which would be a type I error. To decrease this percentage up to at an acceptable level, the following criteria are used: In order for the program to detect an apnea, both the heart rate and the electrical activity of the diaphragm must be exceeded, in addition, the program must detect values outside twice consecutively the intervals to declare an apnea episode. The above is demonstrated masterly by making a statistical evaluation of the device, that is, what probability of type I error exists in a given number of samples.

To perform this evaluation, EMGdi signals from a healthy subject were recorded by breathing normally, using the A / D acquisition channel of an RCM4510W analog-to-digital converter.

The data recording time to perform stage a) was 8 minutes 37 seconds, which means 251 time windows of 2 seconds each as presented in Fig 8.

With these data, one μ and one σ are calculated, both for the sum of heights (EMGdi), and for the heart rate (ECG), with the results reported in Table 6. Table6. Features for 250 training time windows.

 1.96 * <r 4972.3 .43

μ - 1. 96 * σ 169.23 59.15 According to Fig. 9. The range of confidence limits for the EMGdi signal (a), the range of confidence limits for heart rate (b) in a record of another 14 minutes 26 seconds of the EMGdi signal is presented graphically of the same person under observation.

According to Fig. 9 and To evaluate the performance of the method and the reliability of the system, another 14 minutes 26 seconds of the EMGdi signal of the same subject time were recorded in which no apnea episodes occurred since the subject cannot simulate an episode of real apnea, where you alter your heart rate to the point that it goes out of a normal range. With these data several tests were performed and the following results were obtained from table 7:

Table 7. Proportion of false alarms for each of the tests performed and calculation of the estimation error.

 FA ratio Error

 Estimate

Where:

 • Test R: An apnea episode is detected when the EMGdi signal and the heart rate in the same window are outside the confidence limits for two consecutive time windows.

 • R2 test: An apnea episode is detected when the EMGdi signal and the heart rate in the same window are outside the confidence limits.

· Test R3: An apnea episode is detected when the EMGdi signal is outside the confidence limits during two windows of consecutive time and in the case of heart rate, it is only outside the confidence limits in the second window.

• False Alarm Ratio (FA): It is the probability that type I errors are commented in the training evaluation.

 · Error of estimation: It is the percentage of error of the probability of false alarms.

 • n: It is the sample size used to evaluate the training of the algorithm. The number of false alarms for the R and R3 tests is zero, which shows that there was no apnea episode at any time. Regarding the proportion of false alarms in the R2 Test, it was expected, since a 95% confidence level was defined for each EMGdi and FC heart rate biological variable. The percentage of error of the results obtained for the proportion of false alarms is given by the estimation error, and this is 4.81%.

In one embodiment of the invention, the processor declares an apnea episode when two physiological variables exceed the confidence limits in two consecutive time windows of 4 seconds, since the probability of false alarms is almost zero, that is, 4.81%, according to the results produced by the statistical evaluation of the invention.

Glossary: Based on the production mechanism, apneas can be:

 • Of central origin: absence of airway flow and respiratory movements without evidence of obstruction.

 • Obstructive: absence of airflow with contraction of the respiratory muscles.

 · Mixed: in the same apnea episode a central and an obstructive phase is observed.

• HR: Heart Rate • EMGdi: Diaphragmatic Electromyography Signal.

 • Sp02: Oxygen saturation of hemoglobin in blood.

BIBLIOGRAPHIC REFRENCES

[1] S. Corne, K. W. (2000). Effects of inspiratory flow on diaphragmatic motor output in normal subjects. Journal of Applied Physiology, 481-492. It should be understood that the present invention is not limited to the modalities described and illustrated, since as will be evident to a person versed in art, there are possible variations and modifications that do not depart from the spirit of the invention, which is only found defined by the following claims.

Claims

CLAIMS. 1. A non-invasive device for the early detection of the presence of apnea that comprises: a. EMGdi, FC and Sp02 signal detection means; b. an EMGdi, FC and Sp02 signal conditioning module connected to the EMGdi, FC and Sp02 signal detection means; c. a signal processing module connected to the signal conditioning module EMGdi, FC and Sp02; d. a processor; and. an analog to digital converter connected to the signal processing module and the processor; F. a memory connected to the processor; g. a power source that provides power to the above components; where the processor, during a training stage, calculates, from data extracted from the EMGdi, HR and Sp02 signals stored in memory during the training stage, in an interval the confidence limits of the data, and during a stage detection, compares the data stored in memory extracted from the EMGdi, HR and Sp02 signals during the detection stage with the confidence limits of the data calculated in the training stage according to the result of the comparison apnea detection is indicated obstructive, central or mixed, performing early detection of the presence of apnea. 2. The device claimed in claim 1 characterized in that the EMGdi, FC and signal detection means are three surface electrodes with precision modular connectors, type interlocking mechanisms Push-Pull and shielding protection rings for electromagnetic compatibility. 3. The device claimed in claim 2, characterized in that: - two electrodes are located on the patient's costal margin, one on the left side and the other on the right side of the body, in a line drawn vertically from the nipples, between the sixth and seventh intercostal region; and, - the third electrode is located on the sternum. 4. The device claimed in claim 1 characterized in that the EMGdi and FC signal conditioning module comprises an amplification unit connected to a filtering unit. 5. The device claimed in claim 4 characterized in that the amplification unit of the EMGdi and FC signal conditioning module comprises: a. an isolation amplifier with a gain between 1 and 20; b. an operational amplifier in differential configuration with gain between 1 and 15 connected to its input with the isolation amplifier; c. a variable offset voltage adjustment resistor connected to the input of the amplification stage; d. a decoupling capacitor connected to the output of the instrumentation operational amplifier; and. a decoupling capacitor connected to the output of the isolation amplifier; and, f. a band-pass filter with a lower cutoff frequency between 0.1 Hz and 0.5Hz and an upper cutoff frequency between 800Hz and 900Hz connected to the isolation amplifier output. 6. The device claimed in claim 4 characterized in that the filtering unit comprises a. a first channel that connects to an active high-pass filter with a cutoff frequency of 63Hz connected to the output of the isolation amplifier of the amplification unit of the signal conditioning module; b. a second channel that connects to an active low-pass filter with a cutoff frequency of 40Hz connected to the output of the isolation amplifier of the amplification unit of the signal conditioning module; c. a signal amplification circuit with gain of 1 to 15 and an offset voltage of 1V to 2V connected to the filter outputs; d. an anti-aliasing filter between 800Hz and 900Hz; and, and. a protection voltage limiter with a value of 3.6V, connected to a signal amplification circuit and to the input to an analog-to-digital converter. 7. The device claimed in claim 4 characterized in that the filtering unit of the signal conditioning module is a band-pass filter with cut-off frequencies between 13Hz and 25Hz. 8. The device claimed in claim 1 characterized in that the signal processing module comprises: a. an ECG removal unit; b. a QRS complex detection unit connected to the ECG elimination unit; and, c. an EMGdi envelope detection unit connected to the ECG removal unit. 9. The device claimed in claims 1, 4 and 8 characterized in that the signal processing module is connected through two channels to the electromyography signal conditioning module, where a first channel connects the filtering unit of the signal conditioning module. signal to the ECG elimination unit, and a second channel connects the filtering unit of the signal conditioning module with the QRS complex detection unit. 10. The device claimed in claim 8 characterized in that the ECG elimination unit and the QRS complex detection unit of the signal processing module comprises: a. a precision full wave rectifier; b. a passive low-pass filter with a cut-off frequency between 5Hz and 10HZ connected to the precision full-wave rectifier; c. a high-pass filter with a cut-off frequency between 0.1 Hz and 0.5 Hz connected to the output of the passive low-pass filter with a cut-off frequency between 5 Hz and 10 Hz; d. a level comparator connected to the output of the low-pass filter with a cutoff frequency between 5Hz and 10HZ; and, and. precision amplifier of the EMGdi signal in non-inverting configuration and variable gain connected to the input of the full-wave rectifier; where the level comparator provides a voltage pulse as a trigger signal to operate a bilateral switch that controls the passage of the EMGdi signal. 11. The device claimed in claim 8 characterized in that the EMGdi envelope detection unit of the signal processing module comprises: to. an envelope detector made up of an active low-pass filter, with a cut-off frequency between 0.5Hz and 1 Hz whose input is connected to the output of the precision full-wave rectifier; b. a signal offset unit with a passive high-pass filter with a frequency between 0.01 Hz to 0.1 Hz; and, c. a detector in non-inverting configuration connected to the active low-pass filter with a cut-off frequency between 35Hz to 45Hz. 12. The device claimed in claim 1 characterized in that the processor operates in training mode and in apnea episode detection mode, where in the training mode the normality of the Sp02, EMGdi and HR parameters is identified, and in Detection identifies the abnormality of the parameters to detect and classify apnea episodes, indicating these episodes as an alarm signal. 13. The device claimed in claim 1 characterized in that the processor is connected to a means for generating a patient ventilation control signal from the apnea condition; 14. The device claimed in claim 1 characterized in that the processor is connected to a display device to display the envelope of the EMGdi signal, and display the apnea condition messages. 15. The device claimed in claim 1 characterized in that the processor is connected to a means that indicates the apnea state. 16. The device claimed in claim 1 characterized in that the processor is connected to Sp02 measurement means. 17. Method for the detection of obstructive, central or mixed apnea, characterized by the following stages: a. a training stage (a) that consists of establishing the comparison characteristics in the time domain in a memory through the steps of: Yo. acquire the EMGdi, HR, Sp02 signals for a time T1; Yo. adapt the EMGdi signal in such a way that the result of the processing is a respiratory pattern in the form of a periodic wave; Neither. extract the following features in the time domain: area under the curve, mean^ and standard deviations (o), for the EMGdi, HR and Sp02 envelope signals; iv. store these characteristics in a memory; and V. calculate in an interval the confidence limits of the stored characteristics taking into account that the minimum number of samples is met according to the central limit theorem, achieving an interval between an upper limit and a lower limit such that the relationship Ρ( μ _χ ≤ z≤ μ ₂ ) = (1 - a) ^■ 100; b. a stage (b) of apnea detection through the steps of: v. repeat steps i, i, iii of step (a) continuously every time T2; saw. storing the characteristics obtained in step i above in a memory every time T2; vii. comparing the features stored in step ii with the confidence interval calculated in step v of step (a) each time T2; and, viii. activate apnea episode alarm if the characteristics compared in step iii above differ from the confidence interval calculated in step v of step (a). 18. The method for detecting apnea claimed in claim 17 characterized in that step iv of step (b) comprises the substeps of: iv. activate an obstructive apnea episode alarm if the characteristics compared in step iii of step (b) is greater than the upper limit of the range established in step v of step (a); v. activate a central apnea episode alarm if the characteristics compared in step iii of step (b) are less than the lower limit of the range established in step v of step (a); and, saw. activate mixed apnea episode alarm if the characteristics compared in step iii fluctuate above and below the upper and lower limits established in step v of step (a). 19. The method for detecting apnea claimed in claim 17 characterized in that step ii of step (a) comprises the substeps of: a. extract the envelope signal from the EMGdi signal; b. eliminate the ECG component of the EMGdi signal; c. obtain an indicator signal of the patient's HR; d. reconstruct the ECG signal removed in substep b; and. eliminate the QRS complex present in the EMGdi signal; and, f. acquire the Sp02 signal. 20. The method for detecting apnea claimed in claim 17 characterized in that step III of step (a) comprises the substeps of: d. extract the area under the EMGdi envelope curve of each time T2; and. calculate for the ECG signal present in the EMGdi signal, the average of the times between pulses of the ECG signal obtained every T2 seconds; and, F. calculate the means (μ) and standard deviations (o) for the EMGdi, HR and Sp02 envelope signal at each time T2. 21. The method for detecting apnea claimed in claim 17 characterized in that in step (a) step v is carried out in a normal distribution defining a confidence level of 95%, so that the probability for the interval meets the expression P(X - 1.96 * σ≤ μ≤ Χ + 1.96 * σ). 22. The method for detecting apnea claimed in claim 17 characterized in that in step (b) steps iii) to iv) are carried out by comparing the values of the envelope signal of EMGdi, HR and Sp02 signal with pre-established limits. , and if these limits are exceeded, the apnea episode alarm is activated. 23. The method for detecting apnea claimed in claim 17 and 22 characterized in that the apnea episode alarm is confirmed by comparing an Sp02 signal.