WO2014049438A2

WO2014049438A2 - Method of non-invasive determination of glucose concentration in blood and device for the implementation thereof

Info

Publication number: WO2014049438A2
Application number: PCT/IB2013/002654
Authority: WO
Inventors: Yakov Benediktovich Ulanovsky; Aleksandr Mihaylovich FROLOV; Alena Yakovlevna KOZLOVA
Original assignee: Scienmet La, Inc.
Priority date: 2012-09-25
Filing date: 2013-09-24
Publication date: 2014-04-03
Also published as: WO2014049438A3; RU2506893C1

Abstract

This invention relates to the field of medicine, more specifically to endocrinology, and may be used to control glucose concentration in blood in patients with carbohydrate metabolism disorders. The claimed method of non-invasive determination of blood glucose content makes it possible to determine blood glucose content by the voice of a person.

Description

Method of non-invasive determination of glucose concentration in blood and device for the implementation thereof

FIELD OF THE INVENTION

This invention relates to the field of medicine, more specifically to endocrinology, and may be used to control glucose concentration in blood in patients with carbohydrate metabolism disorders, to conduct differential diagnostics for insulin-dependent and non-insulin-dependent types of diabetes mellitus, and to determine the state of the disease compensation.

STATE OF THE ART

Diabetes mellitus is a chronic disease that originates as a result of insufficient production of insulin by the pancreas or ineffective receptiveness of the cells of the patient to produced insulin.

Diabetes mellitus is a very widespread disease; the number of people suffering from that malaise grows by the year. Today, over 300 million people on Earth have diabetes mellitus already.

Diabetes mellitus requires on-going control of glucose level in the blood, otherwise the disease may lead to serious complications. Only if glucose concentration in the blood is maintained within norm (3.5-6.0 mmol/1), it is possible to forestall the development of complications. Following correct nutrition and physical activity regimens and maintaining a normal or close to normal glucose concentration in the blood makes it possible to prevent the development of diabetic complications.

A blood sugar test, where it is necessary to obtain a drop of blood as a sample, is widespread. For that purpose, special automatic devices are used to puncture the skin. The determination of sugar level is performed at the laboratory. The known methods of performing blood analysis for sugar content in the blood of a patient are based on the property of blood sugar to reduce certain salts in the process of complex chemical reactions; such analyses are biochemical in nature.

The blood for analysis for glucose content is taken by a variety of methods: from a vein (venipuncture), by puncturing the skin of fingertips or ear lobe. In the first case, venous blood is analyzed, while in the second case it is capillary blood. At present, portable devices have been designed for blood sugar determination that may be used by patients at home for determining their blood sugar levels by themselves. That is necessary to correctly adjust the drug dosage for diabetic patients, which significantly increases the efficacy of treatment.

The disadvantage of such methods consists in the fact that they are expensive and, most important of all, require a blood sample to be taken, which implies the likelihood of injuring and infecting the patient.

Non-invasive methods of the determination of glucose concentration in human blood are known that do not require a blood sample to be taken.

For example, a method is known of the determination of glucose concentration in human blood based on the measurement of full electrical resistance of the skin, or one of the components of full electrical resistance of the skin (RU 2230485, 2004).

However, the said method has low sensitivity to glucose concentration determination, since it makes use of a computational formula containing parameters difficult to ascertain. Such parameters, though taking into account the relationship between glucose concentration and electrical parameters of the skin, depend on the ionic content of liquids and perspiration, which differ strongly from person to person.

A method is known of the determination of glucose concentration in human blood and continuous monitoring of glucose concentration in human blood (RU 2342071 , 2007). The method consists in measuring the transmitting functions with the help of two pairs of four- electrode sensors fastened on the surface of the human body.

A disadvantage of the method is low sensitivity to glucose concentration determination, since glucose is electrically neutral. Its concentration in the blood is three orders less than concentration of electrolytes in the blood and in biotissues.

A non-invasive method is known of the determination of glucose concentration in human blood (RU 2295915, 2005). The method is implemented by means of laser irradiation of the maximum blood vessel concentration area on the mucous membrane, recording of information and instrumental transformation thereof by means of polarization vector orientation and reflected intensity identification, and calculation of the substance concentration in the blood based thereon. For irradiation, a laser beam is used with zero polarization and a wave-length in the range of 0.5-2.1 mkm. The polarizer-analyzer is calibrated beforehand with respect of laser irradiation local absorption points in the mucous tissue, changes in polarization vector orientation of reflected irradiation are recorded in laser irradiation local absorption points according to the angle of rotation thereof, and substance concentration is determined according to polarization vector rotation angle.

However, that method has a low level of accuracy, since the blood vessels, e.g. located in the derma, are about one millimeter deep from the horny layer of the skin. Moreover, superficial layers of the skin possess anisotropy, and the propagation of linearly polarized light leads to an ambivalent interpretation of measurement results.

A non-invasive method is known of the determination of glucose concentration, comprising the measurement of systolic and diastolic arterial blood pressure consecutively on the left and right arms of the patient, glucose blood content being calculated according to mathematical formulae (RU 2368303, 2007).

However, that method is too complex and not accurate enough, since it requires each time to measure blood pressure on both arms of the patient during fasting and after food intake, while glucose concentration in blood is calculated according to mathematical formulae proposed by the applicant.

Researchers from Israel and the USA have developed a method of early Parkinson^'s disease diagnostics, based on the assessment of the finest changes of the human voice, according to The Daily Telegraph. At this time, Parkinson's disease is diagnosed, as a rule, only after the quantity of dead motor neurons has become sufficiently large to cause such symptoms as muscular rigidity, tremor and impaired balance.

Treatment begun at that stage may slow down the disease progression, but not restore movement functions. Timely diagnostics, according to the researchers, may prevent the destruction of up to 60 percent of the nerve cells of the respective brain areas. It is also known that Parkinson's disease affects larynx muscle functions, which sooner or later results in voice hoarseness.

Attempts of early Parkinson's disease diagnostics based on changes of the voice had been undertaken earlier, but were unsuccessful. Professor Shimon Sapir of the University of Haifa managed to accomplish it by using an alternative approach to voice analysis and developing special computer software that identifies changes characteristic of the disease even before they become discernible by the ear.

It needs to be said that the voice itself, irrespective of the words being said, contains a colossal amount of information; the voice allows determining the character of a person and many other things. It may be attributed to the fact that the voice is directly connected to the anatomy and physiology: it depends on the body structure in general and the structure of voice-generating organs, in particular. Sounds are generated during fluctuation of vocal folds, which are stretched in the larynx like strings. They can perform from 80 to 10,000 and more fluctuations in a second, vibrating either with all its mass or with individual portion thereof. It has been established that, under the influence of nerve impulses coming from the central nervous system, the voice folds change their length, thickness and degree of tension. It is the contraction of various portions of the folds that gives rise to the richest array of sounds, similar to the guitar strings when pressed in different places giving rise to different notes.

The voice pitch depends on the length of the folds, and, in turn, their length and thickness depends on the larynx structure. The voice timbre, volume and color depend not only on the length and thickness of the folds, but also on the structure of the so-called vocal resonators.

The larynx, pharynx, mouth, nasal and sinus cavities are upper resonators, and the trachea, bronchi and lungs are lower resonators. Every person has individual specific features, which is what makes voices so different from each other.

As noted above, the voice is related to the anatomy and physiology, therefore almost any disease, in one way or another, influences the way the voice sounds. The voice changes in case of such diseases as bronchitis, tonsillitis, or sinusitis.

A sophisticated electro-acoustic processing of the human voice pronouncing phrases expressing a variety of emotions - joy, grief, fear or anger - has shown that each state of the human being is characterized with a set of distinctive acoustic features. For example, the state of grief is characterized with the greatest length of a syllable, specific ""ascents" and "descents" in the pitch; fear is characterized with sharp fluctuations of the voice volume, distortions of speed and rhythm of the speech, increased pauses etc.

Thus, the voice conveys very accurate information to the surrounding people about the current state of a person. Such reactions are usually poorly controlled by the person himself, so they are very informative.

As the emotional state of a person changes, a great number of his speech characteristics also uncontrollably change. They are, first of all: the change of the main voice tone frequency and frequencies of several major formants, changes in spectral composition of the speech, rise in the energy of high-frequency components, increased speech volume and speed, the appearance of vibration; there also occur some other changes that may be described in a mathematical form and, consequently, calculated with the help of a computer. Thus, the detection of a broad range of characteristics listed above may, with a high degree of reliability, identify changes in the psycho-emotional state of a person.

At the present time, methods are known of the determination of the psycho-emotional state of a person based on the relationship of the ongoing psychic processes with the dynamics of physiological processes; this is made use of, e.g., in the so-called "lie detectors".

Moreover, parameters of the state of a person may be recorded with the help of external devices not connected directly to the person in question.

That said, parameters of sound fluctuations of the human voice may change not only in the case of change in the emotional state, but due to physiological changes of the larynx and vocal cords, given a change in biochemical properties of the human blood, e.g., a change in the blood glucose level.

According to research, if there is a change in blood glucose level, including the blood flowing in the larynx and the cords, there will occur changes in the elastic properties of the biological tissue of these organs, which in turn will result in changes in spectral characteristics of sound fluctuations of the human voice (which complies with Hooke's law of physics).

Thus, a theoretical analysis of the spectrum of fluctuations of a 2D plate shows that, due to a change in the plate material elasticity ratio, there occurs a change in the spectral composition of plate fluctuations. There occur new overtones, shifts in frequency peaks and changes of intensity of a number of peaks in the fluctuation spectrum. All those changes in the spectrum may be used as quantitative metrics of the change in corresponding parameters depending on changes in the material elasticity ratio.

As applied to the present application, such material is the biological tissue of the larynx and the cord, whose elasticity ratio is suffering change under the impact of changes in glucose level in the human blood.

Therefore, it would be interesting to use the identified correlation between changes in the sound fluctuation spectrum of the voice of a person and changes in the glucose level of his/her blood. Such correlation might be used as a basis of a method of non-invasive determination of glucose concentration in human and animal blood.

A non-invasive method is known of the determination of glucose concentration in blood based on the human voice, comprising the recoding of sound fluctuations of the voice of a person, instrumental transformation thereof for the purpose of obtaining a parameter correlated with the glucose content in the blood, and determination of the glucose content in the blood of the person at the time of such recording. Changes in the frequency of the sound fluctuation spectrum of the human voice were chosen as the parameter correlated with the glucose content in blood (www.fred.ucoz.ru).

However, different people have different frequency characteristics of the larynx and cords, due to gender, age etc. In this regard, a method or device created on the basis of such methodology would have low precision of measurement and would be preferable only as an individual device. In the case of measuring glucose level in another patient with the same device, the error of measurement will be in the order of 40-50%.

Thus, low measurement precision may be said to be a major disadvantage of the state-of- the-art method.

It is also necessary to note that a change in blood glucose level entails not only a change of frequency, but also a change in the intensity of sound fluctuations.

The main frequency area of the human speech is located in the range of from 100 Hz to 1,600-2,000 Hz, with the total intensity of sound fluctuations of this area being about 60 decibel. The difference between speech intensities of different people relative to each other is about 3-4 decibel.

Experiments dedicated to the measurement of peak intensities for selected frequencies in the human voice spectrum (measured with a sound spectrum analyzer) as related to the blood glucose level (measured by a standard glucometer) of different persons have shown that a certain influence of blood glucose level on the intensity of frequency peaks in the human voice spectrum is a reality.

Moreover, for the low- frequency area of the human speech, changes in peak intensities for the selected frequencies induced by changes in the blood glucose level amounted to 5-6 decibel, i.e. actually were in the same order as the differences of speech intensity of different people.

BRIEF SUMMARY OF THE INVENTION

Taking into consideration the above-said, the technical purpose of the invention is the creation of a non-invasive method of measuring glucose concentration in human blood that would allow continuous control of glucose concentration in human blood with sufficient ease and accuracy, and a device for the implementation thereof.

The technical result of the invention is an increase of measurement accuracy due to the selection of the intensity of sound fluctuation frequency peaks of the human voice as a parameter correlated with the glucose content in blood. DETAILED DISCLOSURE OF THE INVENTION

To solve the above problem, in the known method of non-invasive determination of glucose concentration in blood based on the human voice, comprising the recoding of sound fluctuations of the voice of a person, instrumental transformation thereof for the purpose of obtaining a parameter correlated with the glucose content in the blood, and determination of the glucose content in the blood of the person at the time of such recording, the intensity of sound fluctuation frequency peaks of the human voice is selected as a parameter correlated with the glucose content in blood.

According to the claimed method, instrumental transformation includes transforming sound fluctuations of the human voice into a spectrum, sampling sound frequency spectral peaks with uniform intensity in the areas of low and high frequencies, determining intensities of the selected peaks by frequency, and obtaining a ratio between the peak intensities of the selected low and high frequencies.

Besides, the functional dependency of intensity changes of the selected peaks on blood glucose level is determined beforehand with the help of a standard glucometer.

According to the claimed method, the functional dependency of intensity changes of the selected peaks on blood glucose level may be represented by a generalized statistically average functional dependency obtained by means of the recording of sound fluctuations of voices of different individuals, or by an individual functional dependency obtained by means of the recording of sound fluctuations of the voice of one individual.

Preferably, the recording of sound fluctuations of the human voice is performed in the selected low frequency range of from 100 Hz to 1 ,500 Hz and high frequency range of from 7,000 Hz to 10,000 Hz.

A device is also claimed for the implementation of the method according to claim 1, comprising a recorder of sound fluctuations of the human voice, sound spectrum analyzer with filters for the sampling of spectrum peaks in the areas of low and high frequencies, unit of spectrum analyzer data processing and unit of glucose level value determination depending on the intensity of spectrum peaks selected.

Preferably, a microphone is used as a recorder of sound fluctuations of the human voice, and a computer is used as a unit of spectrum analyzer data processing and unit of glucose level value determination depending on the intensity of spectrum peaks selected.

The method is implemented as follows: the subject whose blood glucose level is to be measured, pronounces a standard phrase in the microphone of an e.g. dictaphone, telephone or computer that has a sound spectrum analyzer software installed. Thereafter, with the help of the spectrum analyzer data processing unit and the unit of glucose level value determination depending on the intensity of spectrum peaks selected, sound fluctuations of the human voice are transformed into a spectrum, sound frequency spectrum peaks in the areas of low and high frequencies are sampled, intensities of the selected peaks are determined by frequency, a ratio between the peak intensities of the selected low and high frequencies is obtained, and blood glucose level is determined based on the obtained ratio.

The functional dependency of intensity changes of the selected peaks on blood glucose level is determined beforehand with the help of a standard glucometer, and a generalized statistically average functional dependency is obtained by means of the recording of sound fluctuations of the voices of different individuals, or an individual functional dependency is obtained by means of the recording of sound fluctuations of the voice of one individual.

A methodology has been developed for collecting a bank of measurement data for a large number of people, mostly insulin-dependent diabetics. For each person, at least 10 measurements of glucose content were made with the simultaneous recording of the voice of the individual. A computer with a microphone and installed spectrum analyzer software was used as a bench device for the processing of incoming data and sugar values. The transformation of human voice recordings into a spectrum was performed with the help of the spectrum analyzer, both in the form of digital data and as a spectrum graph. Spectrum graphs represented plots of both separate peaks and continual, merging peaks of different intensity. Peak intensities were plotted on the ordinate, and sound fluctuation frequencies - on the abscissa. At first, the obtained data bases were analyzed visually, by examining qualitative changes on the voice sound spectrum graph depending on the glucose level value. In accordance with such method of analyzing the spectrum graphs and respective glucose levels, frequency areas were found in which peak intensities posted significant changes in response to changes in glucose content for each individual. After that, similar regular peak intensity changes were found in such frequency areas in response to changes in glucose content for all participants of the experiment. Following that, to increase the accuracy of measurement, relative intensity values of harmonic ratio type were used.

To obtain quantitative functional dependencies of peak intensity changes on glucose level values, digital data bases of voice sound spectrum graphs of each individual were used.

The device for measuring blood glucose based on the human voice is schematically shown in the drawing (fig.1) and comprises microphone 1 , sound spectrum analyzer 2 with filters for spectrum peak sampling, in which the voice is disintegrated into a spectrum, computer 3 with units and software for incoming data processing, processor unit 4 that assesses glucose concentration, and telemetric unit 5 that displays blood glucose measurement results.

The device functions as follows: The audio recording is transmitted from microphone 1 to sound spectrum analyzer 2, wherein voice transformation into a spectrum takes place. Digital data are transferred to the computer and telemetric unit 5, the latter displaying the results of glucose level measurement.

EXAMPLES OF METHOD IMPLEMENTATION

The claimed method was experimentally tried on five type 1 diabetics, two type 2 diabetics and three healthy individuals. For each trial subject, one day of testing was used for the calibration of the method. The other test days (from one to three for different individuals) were used for the restoration of glucose in the blood of the subjects. During test days, tests were conducted for half a day on an hourly basis, with changes in blood glucose concentration taking place in both directions (downward and upward). At the same time, for the sake of control, invasive measurements of blood glucose were conducted with the help of the Accu-Chek Active glucometer. Results of some of the experiments are shown in fig.2-4.

Fig.2 shows charts of human voice intensity changes depending on blood sugar levels for five insulin-dependent diabetics. The measurements were made with the help of Accu-Chek, a sound spectrum analyzer and a special calculation methodology developed for the determination of the functional dependency of the selected peak intensity on blood glucose. Blood glucose levels are plotted on the ordinate, and relative dimensionless values of voice spectrum peak intensities are on the abscissa. Human voice recordings were made from mobile telephones with the simultaneous recording of blood glucose levels for the same time moment. The trial was conducted for the period of one month. All subjects were diabetics since 5 to 10 years of a variety of ages from 40 to 75. The length of the trial was determined by the need to obtain blood sugar readings from maximum values to normal ones typical of the normal healthy individual.

Fig.3 shows test results of a bench version of the non-invasive device. Rhombus-like points on the chart correspond to blood glucose concentrations measured with the help of the invasive Accu-Chek glucometer. The line with square-like dots shows the results of the noninvasive bench device. Blood sugar levels are plotted on the ordinate, time (in hours) is plotted on the abscissa. Measurements were conducted on insulin-dependent diabetic M. by the two devices concurrently on an hourly basis beginning at 9am in the morning. Fig.4 shows charts of measured blood glucose levels for another insulin-dependent diabetic subject.

Claims

What is claimed is:

1. A method of non-invasive determination of glucose concentration in the blood based on the human voice, comprising the recoding of sound fluctuations of the voice of a person, instrumental transformation thereof for the purpose of obtaining a parameter correlated with the glucose content in the blood, and determination of the glucose content in the blood of the person at the time of such recording, wherein the intensity of sound fluctuation frequency peaks of the human voice is selected as a parameter correlated with the glucose content in the blood.

2. The method according to claim 1, wherein instrumental transformation includes transforming sound fluctuations of the human voice into a spectrum, sampling sound frequency spectral peaks with uniform intensity in the areas of low and high frequencies, determining intensities of the selected peaks by frequency, and obtaining a ratio between the peak intensities of the selected low and high frequencies.

3. The method according to claim 2, wherein the functional dependency of intensity changes of the selected peaks on the blood glucose level is determined beforehand with the help of a standard glucometer.

4. The method according to claim 3, wherein the functional dependency of intensity changes of the selected peaks on the blood glucose level is represented by a generalized statistically average functional dependency obtained by means of the recording of sound fluctuations of voices of different individuals.

5. The method according to claim 3, wherein the functional dependency of intensity changes of the selected peaks on the blood glucose level is represented by an individual functional dependency obtained by means of the recording of sound fluctuations of the voice of one individual.

6. The method according to claim 1 , wherein the recording of sound fluctuations of the human voice is performed in the selected low frequency range from 100 Hz to 1 ,500 Hz n BbicoKHx HacTOT OT 7,000 Hz RO 10,000 Hz.

7. A device for the implementation of the method according to claim 1 , comprising a recorder of sound fluctuations of the human voice, sound spectrum analyzer with filters for the sampling of spectrum peaks in the areas of low and high frequencies, unit of spectrum analyzer data processing and unit of glucose level value determination depending on the intensity of spectrum peaks selected.

8. Device according to claim 7, wherein a microphone is used as a recorder of sound fluctuations of the human voice, and a computer is used as a unit of spectrum analyzer data processing and unit of glucose level value determination depending on the intensity of spectrum peaks selected.