Method for generating a medical early warning parameter using a hearing instrument

Abstract

The invention describes a method for generating an early warning parameter (F) using a hearing instrument (4) which has at least one electroacoustic input transducer (12), wherein the input transducer (12) generates an input signal (14) from ambient sound, wherein the hearing instrument (4) is examined for sniffing (50) by a user of the hearing instrument (4) during a first period (T1) when the hearing instrument (4) is worn by the user in one ear, and corresponding instances of sniffing (50) are registered, and wherein an early warning parameter (F) is generated depending on a frequency (H1) and / or intensity (50b) of instances of sniffing (50) by the user during the first period (T1).

Description

[0001] The invention relates to a method for generating an early warning parameter using a hearing instrument which has at least one electroacoustic input transducer.

[0002] Many neurological diseases, such as Parkinson's disease, are incurable according to the current state of medical research. However, the progression of such diseases can often be significantly slowed with modern drug therapies, which can lead to a significantly improved quality of life for those affected. A key point, however, is the early detection of the disease in order to tailor treatment as precisely as possible to the specific disease (and its specific symptoms).

[0003] Neurological disorders are often diagnosed through clinical tests based on specific suspected symptoms. These suspected symptoms often involve a motor or cognitive impairment, such as restricted movement or a loss of clarity in speech. However, by the time such an impairment is sufficiently noticeable to be used by the affected person themselves or someone close to them as a trigger for the aforementioned tests, the disorder may already be latent in the body for years, sometimes causing permanent damage that can significantly reduce the quality of life, which can still be maintained with timely treatment.

[0004] However, low-level changes in motor function in a sick person are often difficult to identify, since such comparatively subtle changes would usually require a large amount of data, which is usually simply not available because it is not collected in the person's everyday life.

[0005] US 2008 / 0 146 890 A1 discloses a wearable device for monitoring various physiological and environmental factors. A non-invasive real-time health and environmental monitor comprises a multitude of compact sensors integrated into small, flat devices, such as earbud modules. Physiological and environmental data are collected and wirelessly transmitted to a wireless network, where the data is stored and / or processed.

[0006] The invention is therefore based on the object of generating an early warning parameter for a person, on the basis of which clinical tests can be carried out on the person in order to detect a possible neurological disorder.

[0007] The stated object is achieved according to the invention by a method for generating an early warning parameter with the aid of a hearing instrument which has at least one electroacoustic input transducer, wherein an input signal is generated from an ambient sound by the input transducer, wherein the hearing instrument examines the input signal generated during the first period for sniffing by a user of the hearing instrument when the hearing instrument is worn by the user on one ear, and corresponding cases of sniffing are respectively registered, and wherein an early warning parameter is generated depending on a temporal distribution, in particular a frequency, and / or intensity of cases of sniffing by the user during the first period.

[0008] According to the method, during a second period, the input signal generated during the second period is examined for sniffing by the user when the hearing instrument is worn by the user on the ear, and corresponding instances of sniffing are registered, wherein the early warning parameter is generated depending on a decrease in instances of sniffing by the user during the second period compared to the first period. Advantageous and partly inventive embodiments are the subject of the dependent claims and the following description.

[0009] The stated object is also achieved according to the invention by a method for generating an early warning parameter with the aid of a hearing instrument which has at least one electroacoustic input transducer, wherein an input signal is generated from an ambient sound by the input transducer, wherein the hearing instrument examines the input signal generated during the first period for sniffing by a user of the hearing instrument during a first period of time when the hearing instrument is worn by the user on one ear, and corresponding cases of sniffing are respectively registered, and wherein an early warning parameter is generated depending on a temporal distribution, in particular a frequency, and / or intensity of cases of sniffing by the user during the first period of time.

[0010] In this variant of the invention, it is provided that an intensity and / or duration as well as a time stamp are registered for each individual case of sniffing detected on the basis of the input signal, wherein the early warning parameter is generated when a minimum number of cases of sniffing with a predetermined minimum intensity and / or minimum duration is undershot, and / or wherein a parameter corridor is defined for the duration and / or the intensity, and the early warning parameter is generated depending on a deviation of the duration or the intensity of registered cases from the parameter corridor.

[0011] A hearing instrument generally includes any device that is designed to generate a sound signal from an electrical signal - which can also be provided by an internal signal of the device - and to supply it to the ear of a wearer of this device, i.e. in particular a pair of headphones (e.g. as an "earbud"), a headset, data glasses with a loudspeaker, etc. A hearing instrument also includes a hearing aid in the narrower sense, i.e. a device for treating a wearer with a hearing impairment, in which an input signal generated from an ambient signal by means of a microphone is processed into an output signal and is amplified in particular in a frequency band-dependent manner, and an output sound signal generated from the output signal by means of a loudspeaker or similar is suitable, in particular in a user-specific manner, to at least partially compensate for the wearer's hearing impairment.

[0012] An electroacoustic input transducer includes, in particular, any device configured to generate a corresponding electrical signal from a sound signal. In particular, preprocessing can also occur during the generation of the input signal by the input transducer, e.g., in the form of linear preamplification and / or A / D conversion. The correspondingly generated input signal is provided, in particular, by an electrical signal whose current and / or voltage fluctuations essentially represent the sound pressure fluctuations in the air.

[0013] The early warning parameter can be used to initiate clinical tests in a person that may detect a neurological disorder.

[0014] A neurological disorder can include any neurological condition that, particularly in the long term, can lead to significant limitations in the motor, sensory, and / or cognitive abilities of a patient, including hyposmia and / or anosmia and / or Parkinson's disease. For hyposmia and / or anosmia and / or Parkinson's disease as a neurological disorder, a particularly meaningful early warning parameter can be generated using the described method. In 90% of people with Parkinson's disease, the disease is accompanied by a reduced sense of smell, i.e., hyposmia. Hyposmia and anosmia lead to a reduced frequency of sniffing, with the intensity or depth of an individual sniffing act increasing on average.

[0015] The invention advantageously exploits the fact that neurological disorders, including those mentioned here, are also associated with an initially mild loss of the sense of smell, which gradually worsens. In many cases, this loss of smell occurs years before other neurological symptoms, such as limitations in gross or fine motor skills, appear.

[0016] The loss of smell can lead to a decrease in "sniffing," that is, a sudden or sustained, gasping inhalation through the nose. This type of sniffing also occurs in people without neurological disorders. Typical values ​​are around 20 times per hour, but can be higher or lower depending on external conditions (such as humidity) or personal circumstances (such as allergies or physical activity).

[0017] A noticeably infrequent sniffing episode (compared to statistically expected values, or over the course of a person's life) can then be interpreted as an indication of a possible neurological disorder, meaning that the sense of smell no longer functions as it does in healthy individuals, resulting in a loss of interest in olfactory stimuli. This can now be quantified using the early warning parameter. This allows early detection of suspected neurological disorders and the initiation of appropriate clinical tests, potentially leading to years of therapeutic advantage over conventional detection methods, leading to a sustained improvement in the patient's quality of life.

[0018] An early warning parameter is understood in particular as a quantitative measure (e.g., a scalar numerical value or a group of numerical values ​​as a vector-valued early warning parameter) for a potentially medically relevant (and correspondingly investigated) characteristic, in this case, a sniff, in which a change is to be expected in the case of a neurological disorder, indicates a numerical abnormality compared to an absolute or relative scale. The early warning parameter can then be compared with the corresponding scale to trigger further clinical tests on the person (at the person's own initiative, by a treating physician, by another medical professional, or even by a hearing aid acoustician).

[0019] The absolute scale can be generated, in particular, through statistical surveys of the frequency of sniffing that would typically be expected for the person in question. Demographic factors such as age, weight, or physical fitness, or known allergies can also be taken into account. The relative scale can be achieved, in particular, through a temporal development of sniffing, which is reflected or made visible by the early warning parameter. In particular, the early warning parameter can be compared with a corresponding threshold value (absolute value or threshold value for relative deviation with decreasing frequency in a long-term comparison).

[0020] The first period is chosen to be a time window of at least three days, preferably at least one week, up to four weeks, and preferably up to two weeks. During this first period, while the user is wearing the hearing instrument on their ear, the input signal is examined for sniffing by the user of the hearing instrument using suitable acoustic features, which can particularly include noise analysis and / or detection of a breathing process. This examination is preferably carried out continuously. In particular, a "trigger" can be used for this purpose, for example in the form of a level peak that can occur during sniffing. If the sound level of the input signal exceeds a certain limit, the past seconds (e.g., preferably at least 0.5 seconds, particularly preferably at least one second, preferably up to five seconds, particularly preferably up to three seconds) are examined in detail for the above-mentioned characteristics. Based on this examination, it is then determined when the user sniffs, i.e., in particular, inhales continuously and intermittently through the nose.

[0021] Such an event, a case of sniffing, is then recorded and preferably stored with time information. The storage can be permanent or temporary in the hearing instrument itself, or in an auxiliary device connected to the hearing instrument (such as a smartphone, via a corresponding app for the hearing instrument). In particular, information on the intensity of the sniffing can also be stored, which can be obtained based on the level and / or duration of a peak in the input signal during sniffing.

[0022] Based on a temporal distribution, particularly the frequency of these recorded events when the user snoops during the first period, the early warning parameter can now be generated. The early warning parameter can be a pure numerical value of a frequency (e.g., "x events / day"), or it can weight the events according to their intensity and normalize them over time. Such an early warning parameter can then be compared to an absolute threshold as described above.

[0023] However, the early warning parameter can also be calculated based on a temporal development of the described frequency (“x events / day” or sum of the events weighted by intensity per unit of time).

[0024] According to the invention, during a second period of time, the input signal generated during the second period of time is examined for sniffing by the user when the hearing instrument is worn by the user on the ear, and corresponding instances of sniffing are registered, wherein the early warning parameter is generated as a function of a decrease in instances of sniffing by the user during the second period of time compared to the first period of time.

[0025] The temporal development of the frequency of sniffing ("x events / day" or the sum of events weighted by intensity per unit of time) can therefore be investigated, in particular, by first recording this frequency for the first period and then for a later, second period. The early warning parameter can then be determined, in particular, by an absolute or relative increase in the number of events or the weighted value, or can be output as a binary value by comparing one of the aforementioned variables with a correspondingly suitable threshold.

[0026] This comparison can also be made by a “running” comparison, where the second period immediately follows the first period or overlaps with it.

[0027] In an alternative design, at least five months separate the first and second periods. This makes it easier to identify significant changes from statistical noise in the collected data on the frequency of snooping.

[0028] To examine the input signal for user sniffing, a breathing event is advantageously identified in the input signal (i.e., based on the acoustic information of the input signal), and the duration of the breathing event and / or the sound pressure generated by the breathing event is determined. This allows sniffing to be estimated or detected particularly efficiently from the input signal.

[0029] Advantageously, a physical activity of the user is detected by at least one physiological sensor and / or at least one motion sensor, wherein an instance of sniffing is also registered depending on the user's physical activity. The motion sensor can in particular be an acceleration sensor. The physiological sensor can be a body temperature sensor and / or a pulse sensor and / or blood pressure sensor of the user. During physical activity, breathing rate and intensity generally increase, so that, for example, sniffing can be given less weight or completely ignored when physical activity is detected.

[0030] It has also proven advantageous to record an intensity and / or duration as well as a time stamp for each individual sniffing case detected based on the input signal. The early warning parameter is generated when a minimum number of sniffing cases with a specified minimum intensity and / or duration is undershot, and / or a parameter corridor is defined for the duration and / or intensity, and the early warning parameter is generated depending on a deviation of the duration or intensity of registered cases from the parameter corridor. In the former case, on the one hand, it is ensured that sufficiently relevant data points are available for the early warning parameter, and on the other hand, the minimum intensity and duration can also be used as (statistically based) absolute scales.In the latter case, the parameter corridor can describe a “normal” breathing behavior of the user, and the early warning parameter can be generated when deviations from this normal behavior become too large.

[0031] It proves to be further advantageous if, in order to examine the input signal for sniffing by the user, the input signal is transmitted to an auxiliary device connected to the hearing instrument, and the examination is carried out on the auxiliary device, and / or detected cases of sniffing are registered in the auxiliary device, and the early warning parameter is generated in the auxiliary device based on the registered cases of sniffing. The auxiliary device can in particular be a smartphone or a smartwatch and preferably comprises a communication device for transmitting the data from the hearing instrument. In the case of a smartphone, the early warning parameter can also be shown on the display (for example, as part of an app for the hearing instrument). By using a hearing system that also includes an auxiliary device as described, less native computing and storage capacity is required in the hearing instrument.

[0032] Ideally, the temporal early warning parameter is generated from the sniffing incidents using an artificial neural network (NN) and / or artificial intelligence (AI). This preferably occurs in the assistive device. For this purpose, it is advantageous if corresponding training data is collected from other users, and the NN or AI is trained using this training data. The data collected in the hearing instrument (e.g., an intensity and / or duration as well as a time stamp of the sniffing, as described above) is then transferred to the appropriately trained NN or AI (e.g., by transferring the data from the hearing instrument to the assistive device via Bluetooth or a comparable protocol and a corresponding signal flow in the assistive device), which then generates the early warning parameter as output. However, the NN or AI can also be implemented natively in the hearing instrument.

[0033] Preferably, a user's speech signal is detected in the input signal, wherein the speech signal is examined by a classifier for critical pronunciation abnormalities, and wherein the early warning parameter is also generated depending on the critical pronunciation abnormality. The classifier can, for example, detect abnormalities in individual phonemes that occur at a lower level than the general deterioration of the overall pronunciation during the course of Parkinson's disease. The classifier can preferably be implemented by an appropriately trained NN or AI and, in particular, implemented in the assistive device. The abnormalities in the phonemes can provide additional evidence of possible Parkinson's disease, depending on which the early warning parameter can be generated (e.g., by a weighted average using the sniff values).

[0034] Likewise, the input signal can be examined for chewing sounds from the user. Detected instances of sniffing that occur within a minimum temporal proximity to a detected chewing sound are not included in the early warning parameter. This prevents data distortion caused by sniffing food, which may not necessarily be related to hyposmia or similar symptoms.

[0035] The invention further relates to a hearing system with a hearing instrument, wherein the hearing system is configured to perform the method described above. Preferably, the hearing system further comprises an auxiliary device connectable to the hearing instrument, which is configured, in particular, as a smartphone and / or a smartwatch. In an alternative, further advantageous embodiment, the hearing system consists only of the hearing instrument.

[0036] The hearing instrument according to the invention shares the advantages of the method according to the invention. The advantages stated for the method and its further developments can be applied analogously to the hearing system.

[0037] An embodiment of the invention is explained in more detail below with reference to the accompanying drawings, each of which shows schematically: Fig. 1 in a block diagram a hearing system with a hearing aid and an auxiliary device designed here as a smartphone, and Fig. 2 shows a block diagram of a method for generating an early warning parameter using the hearing system according to Fig. 1.

[0038] Corresponding parts and sizes are provided with the same reference numerals in all figures.

[0039] In Fig. Figure 1 schematically shows a block diagram of a hearing system 1, which comprises a hearing instrument 4 configured as a hearing aid 2 and an auxiliary device 8 configured as a smartphone 6. The hearing aid 2 comprises an electroacoustic input transducer 12 configured as a microphone 10, which is configured to generate an input signal 14 from ambient sound (not shown in detail). The hearing aid further comprises a control unit 16 with a processor unit 18, which in turn comprises a non-volatile memory, a signal processor, and a working memory addressable by the signal processor (each not shown in detail).

[0040] The control unit is configured, among other things, to process the input signal 14 into an output signal 20 (based on hearing situation-dependent parameters and / or hearing programs or similar stored in the non-volatile memory) based on audiological specifications for a user of the hearing aid (which can be stored in the non-volatile memory, for example, as audiograms or similar information), and in doing so, in particular to amplify and / or compress the signal in a frequency band-specific manner. The said processing of the input signal 14 into the output signal 20 is carried out, in particular, with the aim of correcting a hearing impairment of the user. The hearing aid is further configured to convert the output signal 20 into an output sound (not shown in detail) by means of an electroacoustic output transducer 22, which in this case is designed as a loudspeaker 24, which is fed to the user's ear (i.e., through the auditory canal to the eardrum).

[0041] However, the electroacoustic output transducer 22 can also be configured as a bone conduction transducer or another comparable device. The hearing aid 2 can, in particular, also have at least one further electroacoustic input transducer, which generates a corresponding further input signal, based on the signal contributions of which the output signal 20 is also generated (preferably by directional signal processing together with the input signal 14).

[0042] The hearing aid additionally has an acceleration sensor 26, which is configured to detect the user's movements, as well as a physiological sensor 28, which can be provided, for example, by a temperature sensor and / or a blood pressure monitor (e.g., as a PPG sensor). The physiological sensor 28 is particularly designed such that, based on its data, possibly in conjunction with the motion sensor 26, the user's sporting activity can be determined.

[0043] The hearing aid 2 further comprises a communication device (not shown in detail) (e.g., an antenna) which is configured to establish a corresponding data connection 30 (dashed line) with the smartphone 6 via a suitable protocol, such as Bluetooth, and to exchange data with the smartphone 6 via said data connection 30. For this purpose, the smartphone 6 also comprises a communication device (not shown in detail). The smartphone 6 further comprises a processor unit 32 with a signal processor and a display 34. The signal processor is configured to address a working memory (not shown in detail), which can be implemented in particular in the processor unit 32.

[0044] In a manner to be described below, the input signal 14 is preferably continuously examined in the control unit 16 or, if necessary, by initiating it via an acoustic "trigger" for features that allow a conclusion to be drawn about sniffing (i.e., in particular, a sudden inhalation through the nose) by the user of the hearing aid 2. If such sniffing is detected, it is registered (preferably at least temporarily in the hearing aid itself, for example in a non-volatile memory), with data, for example, on an assumed intensity of the sniffing (based on the available acoustic information) and a time of occurrence preferably also being stored.

[0045] Based on a temporal distribution, for example a significant decrease in the distribution or frequency of such sniffing over a defined comparison period, an early warning parameter can be generated in a manner to be described later, which indicates an abnormality in this distribution that could possibly be related to a neurological disorder such as anosmia or even Parkinson's. The early warning parameter can be generated in the hearing aid 2 itself or in the smartphone 6, whereby for this purpose, data on any sniffing that has occurred is preferably transmitted at regular intervals from the hearing aid 2 to the smartphone 6 (or to another auxiliary device 8) via the data connection 30. The early warning parameter can then, for example,be shown to the user on the display 34 so that the user can initiate appropriate clinical tests for a detailed medical examination to determine whether a serious neurological disorder, such as a degenerative disease, actually exists and, in this case, begin drug therapy as soon as possible.

[0046] In Fig. 2 is a schematic block diagram of the procedure by means of which the aforementioned early warning parameter is determined by the hearing system 1 according to Fig. 1 can be generated.

[0047] The input signal 14 (or possibly other input signals from the hearing aid) is first examined for sniffing. This involves, among other things, detecting a breathing process 40 of the user, and from this a duration 40a and the sound pressure 40b generated during the breathing process 40. The duration 40a and the intensity 40b can then be compared with associated threshold values, or a vectorial distance to a reference value can be defined in order to determine if the determined values ​​are unusual. If this is the case, a possible physical and, in particular, sporting activity 44 of the user is checked based on a signal 26S from the acceleration sensor 26 and a signal 28S from the physiological sensor (which contains, for example, the user's body temperature and / or blood pressure). If it is determined from the signals 26S, 28S that physical activity is present (y), the associated breathing process 40 is classified as not unusual.However, if there is no physical activity 44 (n), it is assumed that the user is sniffing 50.

[0048] The sniffing process 50 is then registered by storing, in particular, the duration 40a of the breathing process 40 as the duration 50a of the sniffing 50 and an intensity 50b of the sniffing 50 derived from the sound pressure 40b generated during the breathing process 40, together with a time marker t0, preferably in a memory of the hearing aid 2.

[0049] In the manner described, the individual cases of snooping 50 are registered over a first period T1, which is preferably on the order of one week. An early warning parameter F could now be formed based on the information available for said cases (dotted arrows and blocks), for example by comparing a frequency H1 of the cases of snooping 50, possibly weighted according to their duration and intensity, with a corresponding absolute reference value R, which may also depend on the user's biometric data, such as their age (and any known pre-existing conditions, allergies, or the like). This comparison is preferably carried out in the processor unit 32 of the smartphone (after transmission of said information 50a, 50b of the cases of snooping 50), particularly preferably via an appropriately pre-trained artificial neural network and / or artificial intelligence implemented there.

[0050] However, a decrease in the frequency and / or intensity of sniffing can also be checked. For this purpose, the cases of sniffing 50 are also recorded over a second period T2, which preferably lies several months and particularly preferably at least five months after the first period T1, along with their respective duration 50a and intensity 50b, as well as the associated time stamp t0. If it is now determined that in the second period T2 the frequency H2 of sniffing 50 (if necessary weighted according to duration 50a and / or intensity 50b) has decreased significantly compared to the frequency H1 in the first period (for example by comparing the increase with a limit value, or using a tabulation that also takes into account the absolute values ​​of the frequencies H1, H2), the early warning parameter F is generated and displayed to the user on the display 34.

[0051] Although the invention has been illustrated and described in detail by the preferred embodiment, the invention is not limited to the disclosed examples and other variations may be derived therefrom by those skilled in the art without departing from the scope of the invention. List of reference symbols 1 hearing system 2 hearing aids 4 Hearing instrument 6 smartphones 8 Auxiliary device 10 Microphone 12 input converters 14 Input signal 16 Control unit 18 Processor unit (of the hearing aid) 20 Output signal 22 output converters 24 speakers 26 / 26S Accelerometer / Accelerometer signal 28 physiological sensor / physiological sensor signal 30 Data connection 32 processor unit (of the smartphone) 34 Display 40 breathing process 40a / b Duration / generated sound pressure of the breathing process 44 physical activity 50 sniffs 50a / b Duration / intensity of sniffing F Early warning parameters H1 / 2 Frequency R reference value t0 time marker T1 / 2 first / second period

Claims

[1] Method for generating an early warning parameter (F) with the aid of a hearing instrument (4) which has at least one electroacoustic input transducer (12), wherein an input signal (14) is generated from an ambient sound by the input transducer (12), wherein the hearing instrument (4) during a first period (T1) examines the input signal (14) generated during the first period (T1) for sniffing (50) by a user of the hearing instrument (4) when the hearing instrument (4) is worn by the user on one ear, and corresponding instances of sniffing (50) are respectively registered, wherein an early warning parameter (F) is generated depending on a frequency (H1) and / or intensity (50b) of cases of sniffing (50) by the user during the first period (T1), wherein during a second period (T2) the input signal (14) generated during the second period (T2) is examined for sniffing (50) by the user when the hearing instrument (4) is worn by the user on the ear, and corresponding instances of sniffing (50) are respectively registered, and wherein the early warning parameter (F) is generated as a function of a decrease in instances of sniffing (50) by the user during the second period (T2) compared to the first period (T1). [2] Method according to claim 1, wherein at least five months lie between the first period (T1) and the second period (T2). [3] Method for generating an early warning parameter (F) with the aid of a hearing instrument (4) which has at least one electroacoustic input transducer (12), wherein an input signal (14) is generated from an ambient sound by the input transducer (12), wherein the hearing instrument (4) during a first period (T1) examines the input signal (14) generated during the first period (T1) for sniffing (50) by a user of the hearing instrument (4) when the hearing instrument (4) is worn by the user on one ear, and corresponding instances of sniffing (50) are respectively registered, wherein an early warning parameter (F) is generated depending on a frequency (H1) and / or intensity (50b) of cases of sniffing (50) by the user during the first period (T1), wherein an intensity (50b) and / or duration (50a) as well as a time mark (t0) are registered for each individual case of sniffing (50) detected on the basis of the input signal (14), and where - the early warning parameter (F) is generated when a minimum number of sniffing cases (50) with a specified minimum intensity and / or minimum duration is not reached, and / or - wherein a parameter corridor is defined for the duration (50a) and / or the intensity (50b), and the early warning parameter (F) is generated as a function of a deviation of the duration (50a) or the intensity (50b) of registered cases from the parameter corridor. [4] Method according to one of the preceding claims, wherein a time window of at least three days up to four weeks is selected as the first period (T1). [5] Method according to one of the preceding claims, wherein, in order to examine the input signal (14) for a sniffing (50) of the user, a breathing process (40) of the user is identified in the input signal (14), and a duration (40a) of the breathing process (40) and / or a sound pressure (40b) generated by the breathing process are determined. [6] Method according to one of the preceding claims, wherein a physical activity (44) of the user is detected by at least one physiological sensor (28) and / or at least one motion sensor (26), and wherein a case of sniffing (50) is also registered depending on the detected physical activity (44) of the user. [7] Method according to one of the preceding claims, wherein the early warning parameter (F) is generated from the sniffing cases (50) using an artificial neural network and / or artificial intelligence. [8] Method according to one of the preceding claims, wherein, for the examination of the input signal (14) for sniffing (50) of the user, the input signal (14) is sent to an auxiliary device (8) connected to the hearing instrument (4), and the examination is carried out on the auxiliary device (8), and / or wherein detected cases of sniffing (50) are each registered in the auxiliary device (8), and the early warning parameter (F) is generated in the auxiliary device (8) on the basis of the registered cases of sniffing (50). [9] Method according to one of the preceding claims, wherein the input signal (14) is further examined for chewing sounds of the user, and Whereby detected cases of sniffing (50) which are in a minimum temporal proximity to a detected chewing sound are not used for the early warning parameter (F). [10] Method according to one of the preceding claims, wherein the speech signal is examined by means of a classifier for a critical abnormality in the pronunciation, and where the early warning parameter (F) is also generated depending on the critical conspicuousness of the pronunciation. [11] Hearing system (1) with a hearing instrument (4), wherein the hearing system (1) is configured to carry out the method according to one of the preceding claims. [12] Hearing system (1) according to claim 11, which further comprises an auxiliary device (8) connectable to the hearing instrument (4).

Images