Device and method for checking a bodily function of a living being

Abstract

A device (10) for checking a bodily function of a living being is proposed, which device comprises at least one microphone (11) for detecting and recording (101) bodily sounds based on an organ function as audio data (11a).

Description

[0001] The present invention relates to a device and a method for checking a bodily function of a living being, preferably a human or an animal, according to the independent claims.

[0002] Listening to the body, typically with a stethoscope, is generally known from the prior art. This involves recording body sounds, which can be used to diagnose physical conditions such as damage and diseases of organs, particularly the heart or lungs, and the abdomen (and possibly blood vessels). Devices and methods for monitoring a body function are also known from the prior art. For example, US 2018 / 0116626 A1 relates to a detector for detecting cardiac activity, which comprises a multi-channel stethoscope connected to one or more processors.

[0003] It is the object of the present invention to further develop a method and a device for checking a bodily function of a living being in such a way that possible damage and / or diseases of organs can be detected as effectively, reliably, mobilely and universally as possible.

[0004] The aforementioned object is achieved by a device for monitoring a bodily function of a living being, comprising at least one microphone. The living being is, in particular, a human or an animal. The microphone is designed to detect and record bodily sounds based on an organ function of the living being. The bodily sounds are recorded as audio data and preferably stored.

[0005] The device may in particular be a mobile terminal, especially a telecommunications device, preferably a smartphone, which can be placed on the body of the living being, for example the chest, the back, the neck, the groin or on the abdominal cavity of the living being.

[0006] The bodily functions are, in particular, cardiovascular functions and / or respiratory functions and / or functions related to the digestive system of the living being. In other words, bodily sounds related to the heart, lungs, blood vessels, and / or at least one digestive organ, such as the stomach and / or intestines, can be recorded as audio data via the microphone, thus enabling the aforementioned functions to be monitored. For example, heart sounds and murmurs can be detected and recorded using the microphone. In other words, heart auscultation can be digitally recorded.

[0007] With the help of the above-mentioned device, it is thus possible to monitor bodily functions in a very simple, universally applicable manner without the need for complex and time-consuming examinations. Electrocardiography and echocardiography, for example, are known as standard examination methods for monitoring and detecting cardiac diseases. These allow a comprehensive assessment of heart valve disease, heart failure, heart muscle disease, cardiac malformations, and cardiac arrhythmias. However, they require expensive equipment and appropriate preparation, such as the attachment of electrodes to the body's surface.With the present invention, the detection of indications of such cardiac disease conditions can be offered cost-effectively and later validated by standard electrocardiographic and echocardiographic examinations.

[0008] With regard to the condition of blood vessels in, for example, the neck and groin regions, conclusions can be drawn from the audio data based on potential flow noises / turbulences through the device. This may provide clues about, for example, calcifications / atherosclerosis and circulatory disorders.

[0009] Furthermore, the device preferably comprises at least one motion sensor for detecting and recording motion data of a vibration of the living being's body based on an organ function. The motion sensor can be, for example, an accelerometer and / or a gyroscope.

[0010] The motion data is based on a cardiovascular function and / or a respiratory function and / or a function related to the digestive system of the living being, resulting in a measurable movement and / or vibration of the rib cage, which can be measured using the motion sensor, preferably the accelerometer. An accelerometer only records the linear motion components in one or more directions, while a gyroscope measures rotation.

[0011] If movements based on cardiac function are recorded using a gyroscope, the movement data, or in other words, the measurement data, is referred to as gyrocardiography. If an acceleration sensor is used, the measurement data relating to cardiac activity is referred to as seismocardiography.

[0012] Preferably, the device may comprise both an acceleration sensor and a gyroscope for detecting and recording movement data of a corresponding movement, whereby the device is thus preferably designed to record the cardiac activity via both the acceleration sensor and the gyroscope.

[0013] The device comprises, in particular, an evaluation unit, wherein the evaluation unit is configured to temporally synchronize and correlate recorded motion data and audio data. The data are synchronized with each other, in particular, by providing the recorded motion data and audio data with at least one time stamp, preferably at regular intervals. The sampling rates preferably differ with respect to the various sensors. The time stamp at the beginning of the data is then searched for, which is as close as possible to the time stamp of the corresponding other data, so that the data can be evaluated in a synchronized manner.

[0014] Synchronization is particularly important because the sampling rates of the sensors can differ. The closest data points at the beginning of the recordings are searched for, thus defining a common start point. Data points before these can be discarded.

[0015] The movement data is preferably collected during the auscultation measurement, specifically during the acquisition and recording of the audio data. The movement data serves primarily to increase the reliability and significance of the subsequent evaluation by the evaluation unit.

[0016] By using the audio data and the movement data, any misuse of the measurement—in other words, the acquisition of the data—can be detected. Since the acquisition is determined by placing the mobile device on the body of the living being, the movement data and the audio data should be correlated with each other. If the correlation is below a previously defined threshold, it can be assumed that a user error has occurred and, therefore, the movement data and / or the audio data are faulty. In such a case, the device is designed to issue a warning and / or an instruction to the user to correctly perform the corresponding measurement again.

[0017] Furthermore, the device can be configured to indicate auscultation points to a user—in other words, points on the body of the living being at which measurements should be taken. For example, the points can be visually displayed to facilitate application for the user. Furthermore, the device can specify a time period for the user or check the time period allocated for the measurement per auscultation point. In particular, the device is configured to indicate to the user whether the corresponding time period has already been reached or whether a new measurement at the same auscultation point is necessary.

[0018] The evaluation unit is further configured to extract one or more audio tracks from the audio data, whereby the audio tracks can differ depending on the organ. For example, audio data can be captured and recorded at different contact points on a body, for example, at different locations on the chest and / or back and / or abdominal cavity of the living being. The device is particularly configured to guide the user to perform appropriate measurements at different locations. Measurements can also be performed at other contact points, such as the abdominal cavity. Depending on the contact point, the evaluation unit is configured to assign an organ to the audio data.For example, audio data recorded on the chest and / or back can be assigned to the heart and / or lungs as organs, while audio data recorded in the abdomen can be assigned to the digestive organs.

[0019] Furthermore, an audio track can be characterized by the fact that the audio data has been filtered, for example, to reduce background noise. The audio track can thus correspond to the filtered audio data.

[0020] In addition, the evaluation unit can be configured to extract one or more motion data traces from the motion data, preferably one trace depending on a previously defined, preferably Cartesian, direction. In particular, one trace per direction is extracted from the motion data of the gyroscope and the motion data of the acceleration sensor, which can then be compared with each other. Above all, a correlation can be determined.

[0021] Furthermore, the evaluation unit is designed in particular to classify at least one section of an audio track and / or at least one section of a movement data track on the basis of the recorded audio data and / or the movement data.

[0022] Within the framework of the classification, information relating to the device and / or the living being can be taken into account in particular. With regard to the device, this can be device settings that can, for example, influence the audio quality. This can be the model, the year of manufacture, the device manufacturer, the operating system, the operating system version, the browser (in the case of an online application), and / or the browser version. The information relating to the living being can be risk factors, for example smoking status. Furthermore, biological sex, age, previous medical conditions and / or medical complaints can be taken into account. Above all, the device can comprise an input module that can preferably query at least one of the aforementioned parameters.

[0023] As part of the classification, a marker can be assigned to at least one, preferably each, section of an audio track and / or a motion data track. This marker is one of the following: conspicuous, inconspicuous, and inadequate. Conspicuous can be understood, in particular, as pathological, while inconspicuous can be understood, in particular, as healthy. Insufficient means that the corresponding section does not have sufficient significance to classify it as either conspicuous or inconspicuous.

[0024] In particular, the corresponding section is marked with a marker. In other words, the period of time in which the corresponding section falls is marked, for example, with color and / or a specific designation. This way, particularly conspicuous areas in the corresponding lane are highlighted.

[0025] The device can in particular comprise a camera for capturing and recording image data based on an organ function, wherein the evaluation unit is designed to determine a pulse of the living being based on the image data. In particular, the image data is based on a cardiac function. In particular, the evaluation unit is designed to determine the heart rate and / or cardiac rhythm based on the image data. Furthermore, at least one medical parameter, such as indications of the underlying systemic blood pressure, can be determined from the correspondingly acquired cardiac information. In particular, the evaluation unit is designed to extract a peripheral pulse curve derivative, in other words, to perform photoplethysmography. In this case, the device has in particular a light source that can detect blood volume changes in the tissue of the living being through tissue.In detail, the skin is illuminated, thus measuring changes in light absorption. For this purpose, the device can primarily comprise a light source, particularly an LED.

[0026] The microphone and / or at least one motion sensor and / or camera are primarily integrated into the mobile device. Therefore, they do not need to be connected to the device, but can be used in the simplest way possible.

[0027] In particular, the device is designed to output to a user at least one audio track with identification of the classified sections and / or at least one movement data track with identification of the classified sections. Furthermore, the entire audio data and / or movement data can be output. Furthermore, the output can comprise a diagnostic recommendation, in particular based on the classified sections, and / or a confidence value for the diagnostic recommendation and / or clinically relevant parameters determined on the basis of the audio data and / or movement data and / or image data. The clinically relevant parameters can be, in particular, the heart rate, the heart rhythm, indications of blood pressure and / or the heart pulse. In particular, conspicuously marked sections are output. An output can be understood primarily as a visual representation and / or acoustic output.

[0028] Advantageously, the device is designed to output the above-mentioned information individually depending on the user, in particular based on predefined user groups. If the user group comprises, for example, professional users such as doctors or medically trained personnel, more comprehensive above-mentioned information can be output than if the user is a user in the sense of a patient. For example, the device can be designed to output a diagnostic recommendation and / or a confidence value for the diagnostic recommendation only for professional users, while a recommendation to consult a doctor can be output for patients. Furthermore, the audio track and / or the movement data track with the identification of the classified sections can only be output to professional users.

[0029] By outputting confidence values ​​and also by outputting at least one track with the label, particularly the sections classified as conspicuous, the professional user can review a diagnostic recommendation that is also issued. This makes it very easy to subject a diagnostic recommendation from the device to critical review. By outputting the confidence value, the professional user can, in particular, decide whether a diagnostic recommendation should be reviewed, for example, by manually listening to and assessing the sections classified as conspicuous and conducting a corresponding manual evaluation. Because the data on which the diagnostic recommendation is based can be presented in a comprehensible manner, e.g., visually and / or acoustically, and can be reviewed, overall confidence in the output is increased.

[0030] An audio track for the heart can, in particular, be classified according to several functions. For example, the same audio track can be classified according to different functions or medical characteristics. With regard to the heart, these characteristics or functions can be, for example, the heart valves and / or the heart rhythm. With regard to the heart rhythm, an arrhythmic and thus abnormal classification can indicate atrial fibrillation or other cardiac arrhythmias, while a rhythmic classification is considered normal. Another characteristic is the heart phases. In particular, the audio track can be segmented according to the detected heart phases. With regard to breathing, phases such as inhalation and exhalation can also be segmented and the corresponding phases classified. Based on the segmented heart phases, the heart rate and / or heart rate variability can also be determined.

[0031] In detail, the audio data captures the relevant phases of natural cardiac activity, resulting in a repetitive, characteristic pattern consisting of the first heart sound, systole, the second heart sound, and diastole. The first heart sound corresponds to the opening of the semilunar valves—in other words, the aortic and pulmonary valves—while diastole indicates the closing of the semilunar valves. Based on the audio data, heart murmurs can now be perceived as audible sounds between heart sounds, which, unlike the repetitive pattern explained above, indicate an abnormality. Thus, sections in which a heart murmur is detected can be classified as abnormal.

[0032] The phases, i.e., diastole and systole, also typically exhibit a specific frequency-amplitude pattern. If an abnormality is detected, higher frequencies are reached. Thus, at least one predefined threshold can exist, whereby a frequency of the frequency-amplitude pattern can be derived from the determined phases of the audio data or audio track. This frequency can be compared to the frequency threshold, and if the frequency threshold is exceeded, a corresponding section of the audio data or audio track can be classified as abnormal. For example, the threshold for the heart can be 400 Hz, preferably 500 Hz.

[0033] It can also be determined whether detected heart murmurs rise, fall, rise and fall again between the heart patterns (in other words, spindle-shaped), or whether they are band-shaped (in other words, constant). Sections can also be classified using these features.

[0034] The sound quality of the audio data or audio track can also be classified, for example, with at least one term such as rough, rubbing, breathy, pouring, or musical. The duration, for example, early, mid, or late systolic or diastolic, and the radiation of a heart murmur, for example, into the carotid arteries or the axilla, can also be classified. In particular, the audio data from different auscultation points are compared to determine the direction of radiation.

[0035] With regard to heart murmurs between the heart sounds, the evaluation unit can, in particular, comprise a comparison of recorded and detected heart murmurs with previously recorded abnormal heart murmurs. These previously recorded heart murmurs can, in particular, already be classified and, furthermore, preferably, be associated with a valvular heart disease, so that the greatest correlation with a previously recorded heart murmur is determined and its classification can be adopted. Because each heart murmur is associated with a valvular heart disease from a corresponding database, the valvular heart disease can be associated by determining the highest correlation with at least one previously recorded heart murmur and, if necessary, output as a diagnostic recommendation.

[0036] In particular, the heart rate can be determined based on the audio data or audio track. The determined heart rate can be compared to predefined thresholds. For example, a minimum threshold can be 60 beats per minute, while a maximum threshold can be 100 beats per minute. If the heart rate is below the minimum threshold or above the maximum threshold, the corresponding section can be classified as abnormal.

[0037] An audio track can also be extracted regarding breathing. Furthermore, the movement data can be evaluated accordingly. In particular, a breathing frequency can be determined and compared with a predefined threshold. The threshold for the lungs can preferably be 600 Hz, preferably 700 Hz.

[0038] Furthermore, particularly with regard to breathing, the duration of inspiration and / or expiration and / or their relationship to each other, in other words the inhalation-exhalation ratio, can be compared with corresponding normal ranges during the evaluation. Corresponding thresholds can be defined as follows for adults: e.g., a minimum threshold of 1:1.5 of the inhalation-exhalation ratio and / or a maximum threshold of 1:2. For children, a threshold can be a deviation of 20% from the 1:1 ratio. Sections in which the maximum threshold is exceeded or minimum thresholds are undercut can be classified as abnormal.

[0039] A corresponding classification can also be made if the exhalation time is prolonged. This may indicate a constriction in the respiratory system, for example, in bronchial asthma. This finding can also be considered when making a diagnosis recommendation.

[0040] Furthermore, respiratory sounds that are detected far away from the respiratory phases can be identified, such as fine-bubble rales or large-bubble rustles. Abnormal respiratory sounds such as pouring, humming, whistling, crackles, rales, stridor, pleural rubs, and / or a weakened respiratory sound can be detected. Sections can then be classified as abnormal and assigned to one of the respiratory sounds. Again, a comparison can be made with respiratory sounds from a database, with the respiratory sound with which there is the greatest correlation being assigned. Each respiratory sound in the database can be associated with a disease. For example, fine-bubble rales can indicate pneumonia, coarse-bubble rales can indicate heart failure, humming or pouring can indicate bronchial asthma and COPD, while crackles or pleural rubs can indicate pulmonary fibrosis.Based on the breathing noises, a disease and thus a diagnosis recommendation can be assigned.

[0041] Furthermore, the device can also be used to assess vascular conditions. For example, images can be used to collect information on flow velocities and the occurrence of flow noises / turbulences over the neck (carotid arteries) or in the groin (iliac vessels). These images can then provide clues about, for example, arteriosclerosis or aneurysms.

[0042] In addition, bowel sounds or the absence of bowel sounds detected by the device can provide clues to potential diseases such as ileus (intestinal obstruction).

[0043] In particular, the device comprises at least one neural network, preferably a plurality of neural networks for classifying the audio data and / or motion data and / or image data. The evaluation unit, in particular the at least one neural network, is designed to classify the features from the at least one data, preferably the audio track or the motion data track. The features can in particular be frequencies and / or amplitudes. Furthermore, lengths of sections, for example the heart phases, or inhalation and exhalation phases, can be understood as features; corresponding ratios can also be understood as features. The corresponding features can be compared with previously defined target ranges, wherein in the event of a deviation from a target range, a corresponding section can be classified as conspicuous.

[0044] In particular, the device can comprise a neural network for classifying the audio data with respect to the heart valves. Furthermore, another neural network can be provided for the heart rhythm. Another neural network can be provided for the heart phases, and another network for respiration. Furthermore, another neural network can be provided to filter out background noise in preparation for the other neural networks, thus increasing the quality of the recordings, in particular the audio data. In this way, the same audio track or audio data can be evaluated by different neural networks trained on different classifications. As an alternative to different neural networks, a single neural network can also be provided that fulfills the aforementioned functions.

[0045] The neural network can, in particular, have a final layer, in particular a corresponding activation layer, which is preferably designed as a soft-max activation function and / or sigmoid activation function. This serves to map the corresponding previously performed evaluations to probabilities and thus enable the output of confidences. For this purpose, the aforementioned classes are predefined, with the soft-max function resulting in a sum across all classes of 1. With a sigmoid function, the corresponding probabilities or confidences per class are independent of one another.

[0046] In particular, a diagnostic recommendation can be issued based on the areas classified as abnormal. This can take into account the extent of the deviation from a defined threshold and / or the frequency of deviations and thus areas classified as abnormal. The diagnostic recommendation is primarily a diagnostic support, i.e. a decision-making aid for diagnosis by professional users, i.e. medical personnel. The diagnostic recommendation can include an indication of a disease and / or damage to an organ. The diagnostic recommendation is preferably accompanied by a confidence value. In particular, a low confidence can have a value of less than 0.5, a medium confidence a value between 0.5 and 0.75, and a high confidence a value greater than 0.75.

[0047] The neural network relating to the heart valves can be a convolutional network, preferably a convolutional recurrent neural network and / or a temporal convolutional neural network, which can preferably comprise at least three, furthermore at most ten layers, in particular five layers. The layers are preferably provided with a ReLu (Rectified Linear Unit) activation function.

[0048] The neural network for heart phases can convert the corresponding audio track or audio data into a spectrogram, primarily a Mel spectrogram. This, too, is primarily a convolutional network with appropriate ReLu activation. The same model can also be used for respiration. Specifically, at least one neural network is pre-trained using contrastive learning as a so-called shared backbone, serving as a common feature extractor for all subsequent tasks. This reduces the number of required parameters, enabling continuous use on mobile devices, as it requires less computing power and lowers power consumption.

[0049] At least one neural network of the device is preferably trained with ECG auscultation data pairs. The ECG is considered the gold standard for assessing heart rhythm. Training with ECG auscultation data pairs allows specific cardiac arrhythmias, such as atrial fibrillation, to be detected even without an additional ECG. This increases the overall validity of the evaluation.

[0050] Specifically, with regard to heart rhythm, a regular heart rhythm is considered normal. Detecting a deviation from a rhythmic pattern can be achieved, in particular, through training with ECG-auscultation data pairs, so that the valuable information from the ECG is considered the gold standard for assessing heart rhythm.

[0051] At least one neural network of the device is preferably trained with echocardiography findings. Echocardiography is considered the gold standard for assessing valvular heart disease. Training with echocardiography findings can detect valvular heart disease even without the presence of an additional echocardiography finding.

[0052] In a further aspect, the invention relates to a method for testing a bodily function of a living being, wherein the method comprises detecting and recording bodily sounds based on an organ function as audio data using a microphone. In particular, the device is implemented using a device described above. Furthermore, the device is designed to implement a corresponding method.

[0053] Overall, the invention plays a major role in outpatient and inpatient patient care, as medically relevant bodily functions can be monitored and medically evaluated solely on the basis of the device or method, regardless of medical equipment. Furthermore, the effort and costs associated with such functions in the prior art are significantly reduced.

[0054] They show in a purely schematic representation: Fig. 1: a device for checking a bodily function of a living being; Fig. 2: a method for testing a bodily function of a living being; and Fig. 3: data recorded by means of a method and a device.

[0055] Fig. 1 shows a device 10 for checking a bodily function of a living being, which device comprises a microphone 11 for detecting and recording bodily sounds as audio data 11a, wherein the bodily sounds are based on an organ function.

[0056] The device 10 further comprises a motion sensor 12 for recording motion data 12a of a movement of the body of the living being based on an organ function. In particular, the device 10 has two motion sensors 12, namely an accelerometer 13 and a gyroscope 14, both of which serve to record the motion data.

[0057] Furthermore, the device 10 can comprise a camera 16, which serves to capture and record image data based on an organ function. Furthermore, the device comprises an evaluation unit 15, which is designed to synchronize the movement data and audio data with each other in time and to correlate them with each other. Based on the evaluation, sections of an audio track and / or a movement data track can be classified. The at least one track or the sections can be output with appropriate labeling of the classification. A diagnostic recommendation, preferably with a confidence interval, can also be output. In addition, clinically relevant parameters can be output based on the audio data and / or movement data and / or image data.

[0058] Fig. 2 shows a method 100 for testing a bodily function of a living being, which comprises capturing and recording 101 bodily sounds based on an organ function as audio data. Furthermore, the method 100 comprises capturing and recording 102 motion data of a movement of the living being's body based on an organ function. Furthermore, the method may comprise capturing and recording 103 image data based on an organ function.

[0059] The corresponding recorded data is evaluated 104. During the evaluation, the recorded data are, in particular, temporally synchronized 105 and correlated 106. During the evaluation 104, one or more audio tracks are extracted 111 or one or more movement data tracks are extracted 111a. These can be assigned to organs or directions. In particular, the movement data can be divided into different directions.

[0060] A classification 112 is performed, in particular of at least one section of an audio track and / or corresponding motion data track. In detail, features can be extracted from the at least one audio track and / or motion data track and compared with previously defined target ranges and / or thresholds 1. A marker can be assigned 115 to each section of the audio track and / or motion data track, and this section can be marked 116 with the corresponding marker.

[0061] Based on the image data, a pulse of the living being can be determined 117. An output 120 of the audio track and / or the movement track can be provided with an identification of the classified sections and / or a diagnostic recommendation and / or a confidence value of the diagnostic recommendation or other relevant parameters.

[0062] Fig.Figure 3 shows recorded data using a method 100 and a device 10. The first row shows audio data 11a, the second row shows recorded movement data 12b of a gyroscope in the X direction, the third row shows recorded movement data 12c of an acceleration sensor in the X direction, and, at the bottom, a peripheral pulse signal 17. This data is incorporated into the evaluation. List of reference symbols 10 Device 11 Microphone 11a Audio data 12 Motion sensor 12a Movement data 12b Movement data of a gyroscope 12c Motion data from an accelerometer 13 accelerometers 14 Gyroscope 15 Evaluation unit 16 Camera 17 Pulse signal 100 methods for testing a bodily function of a living being 101 Capturing and recording body sounds based on an organ function as audio data 102 Recording and recording of movement data 103 Acquisition and recording of image data 104 Evaluation 105 Synchronization 106 Correlation 111 Extracting the audio track 111a Extracting a motion data trace 112 Classification 113 Extraction of features 114 Comparison with the previously defined target values ​​and / or thresholds 115 Assignment of markers 116 Marking 117 Determining the pulse 120th issue QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 2018 / 0116626 A1

[0002]

Claims

[1] Device (10) for checking a bodily function of a living being, characterized by that the device (10) comprises at least one microphone (11), wherein the microphone (11) is designed to detect and record (101) body sounds based on organ function as audio data (11a). [2] Device (10) according to claim 1, characterized by , that the device (10) comprises at least one motion sensor (12) for detecting and recording (102) motion data (12a) of a movement of the body of the living being based on an organ function. [3] Device (10) according to claim 2, characterized by that the device (10) comprises an accelerometer (13) and a gyroscope (14). [4] Device (10) according to any one of the preceding claims, characterized by, that the device (10) comprises an evaluation unit (15) wherein the evaluation unit (15) is configured to temporally synchronize (105) and correlate (106) recorded motion data (12a) and audio data (11a). [5] Device (10) according to claim 4, characterized by , that the evaluation unit (15) is designed to extract one or more audio tracks (111) from the audio data (11a). [6] Device (10) according to claim 5, characterized by that the audio tracks differ according to organ. [7] Device (10) according to claim 4 or 5, characterized by , that the evaluation unit (15) is designed to extract one or more movement data traces (111a) from the movement data (12a). [8] Device (10) according to claim 7, characterized by that the movement data traces differ according to direction. [9] Device (10) according to any one of claims 5 to 8, characterized by, that the evaluation unit (15) is designed to classify at least one section of an audio track and / or a motion data track on the basis of the recorded audio data (11a) and / or the motion data (12a) (112). [10] Device (10) according to any one of claims 5 to 9, characterized by , that the evaluation unit is designed to extract features from the at least one audio track and / or one motion data track for classification (113) and to compare them with previously defined target ranges and / or thresholds (114). [11] Device (10) according to any one of claims 5 to 10, characterized by , that a marker, preferably from the set conspicuous, inconspicuous and inadequate, is assigned to a section of the audio track and / or a motion data track (115), wherein the corresponding section is marked with the marker (116). [12] Device (10) according to any one of the preceding claims, characterized by, that the device (10) comprises a camera (16) for capturing and recording (103) image data based on organ function, wherein the evaluation unit (15) is configured to determine a pulse of the living being on the basis of the image data (117). [13] Device (10) according to any one of claims 5 to 12, characterized by , that the device (10) is designed to output to a user at least one audio track and / or one motion data track with identification of classified sections and / or a diagnostic recommendation and / or a confidence value for the diagnostic recommendation and / or certain clinically relevant parameters based on the audio data and / or motion data and / or image data (120). [14] Method (100) for checking a bodily function of a living being, characterized by, that the method (100) comprises the acquisition and recording (101) of body sounds based on organ function by means of a microphone (11) as audio data (11a). [15] Method (100) according to claim 14, characterized by that the method (100) is carried out by means of a device (10) according to one of claims 1 to 13.

Images