USABILITY AND SATISFACTION OF A HEARING AID

Abstract

The present disclosure relates to a method of improving usability of, and satisfaction with, a hearing aid. Further provided is a system comprising a hearing aid, wherein the system is configured to perform the method.

Claims

1. A method of improving an usability of a hearing aid and/or a satisfaction with the hearing aid, the method comprising: obtaining data from the hearing aid; determining a prediction score based at least in part on the data, the prediction score indicating a likelihood of a user of the hearing aid being dissatisfied with the hearing aid, and executing a response measure if the prediction score indicates that the user of the hearing aid is dissatisfied, wherein the response measure comprises adjusting a functionality of the hearing aid, or arranging for human support, or a combination thereof.

2. The method according to claim 1, wherein the adjusting the functionality of the hearing aid comprises one or more of: reinstalling software on the hearing aid, updating software on the hearing aid, changing one or more algorithm parameters, performing remote automatic fine-tuning of the hearing aid, and/or updating one or more pre-sets/programs on the hearing aid.

3. The method according to claim 1, wherein the arranging for the human support comprises one or more of: notifying the user of the hearing aid, notifying a hearing care professional, notifying a customer service employee, or any combination of the foregoing.

4. The method according to claim 1, further comprising selecting the response measure before the act of executing the response measure is performed, wherein the response measure is selected based at least in part on the data from the hearing aid.

5. The method according to claim 1, wherein the prediction score is at least partly based on data logged prior to hearing aid returns, data logged from non-returns, or a combination of the foregoing.

6. The method according to claim 1, wherein the prediction score is at least partly based on a comparison between data logged prior to hearing aid returns and data logged from non-returns.

7. The method according to claim 1, wherein the act of determining the prediction score is at least partly performed using machine learning and/or artificial intelligence.

8. The method according to claim 1, wherein the prediction score is determined using a model.

9. The method according to claim 8, wherein the model is built based on data logged prior to hearing aid returns, data logged from non-returns, or a combination of the foregoing.

10. The method according to claim 8, wherein the model comprises a neural network.

11. The method according to claim 1, wherein the act of obtaining the data, the act of determining the prediction score, and the act of executing the response measure are performed automatically.

12. The method according to claim 1, wherein the obtained data comprises: use-time, number of pre-set/program changes, number of power downs, number of re-boots, number of sound environment changes, pattern of sound environment changes, time spent in a type of sound environment, GPS location, temperature, pulse, oxidation saturation, or any combination of the foregoing.

13. The method according to claim 1, wherein the prediction score is also based at least in part on user data.

14. The method according to claim 13, wherein the user data comprises: a type of the hearing aid, a model of the hearing aid, age, gender, socioeconomics, hearing loss profile, user rating, number of contacts to a hearing care professional, number of days since last contact with the hearing care professional, use-time of an app, or any combination of the foregoing.

15. The method according to claim 13, further comprising obtaining the user data remotely.

16. The method according to claim 1, further comprising selecting the response measure based on a similarity between the data and other data for one or more other hearing aid users.

17. The method according to claim 16, further comprising determining the similarity between the data and the other data for the one or more other hearing aid users.

18. The method according to claim 1, further comprising selecting the response measure based on user data.

19. The method according to claim 19, wherein the user data comprises: a type of the hearing aid, a model of the hearing aid, age, gender, socioeconomics, hearing loss profile, user rating, number of contacts to a hearing care professional, number of days since last contact with the hearing care professional, use-time of an app, or any combination of the foregoing.

20. The method according to claim 19, wherein the response measure is selected based on a similarity between the user data and other user data for one or more other hearing aid users.

21. A system configured to perform the method according to claim 1.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In the following, exemplary embodiments are described in more detail with reference to the appended drawings, wherein:

[0064] FIG. 1 is a flow diagram in accordance with exemplary embodiments,

[0065] FIGS. 2-3 shows graphs of data obtained from hearing aids worn by users, and

[0066] FIGS. 4-6 schematically illustrate a system comprising a hearing aid and configured to perform the method of improving usability of and satisfaction with a hearing aid in accordance with exemplary embodiments.

DETAILED DESCRIPTION

[0067] Various embodiments are described hereinafter with reference to the figures. Like reference numerals refer to like elements throughout. Like elements will, thus, not be described in detail with respect to the description of each figure. It should also be noted that the figures are only intended to facilitate the description of the embodiments. They are not intended as an exhaustive description of the claimed invention or as a limitation on the scope of the claimed invention. In addition, an illustrated embodiment needs not have all the aspects or advantages shown. An aspect or an advantage described in conjunction with a particular embodiment is not necessarily limited to that embodiment and can be practiced in any other embodiments even if not so illustrated, or if not so explicitly described.

[0068] FIG. 1 shows a flow diagram in accordance with exemplary embodiments of the method of improving usability of and satisfaction with a hearing aid.

[0069] Modern hearing aids are sophisticated electronic devices, which can record a variety of data such as when, how and where the hearing aid is used, as well as any sensor data from on-board sensors. The “when” may include, but is not limited to, date, time of day, time since last reboot, time since first activation by user, use-time, etc. The “how” may include, but is not limited to, whether the hearing aid is on or off, whether pre-sets/programs are used or changed, whether the hearing aid is turned on or turned off, whether specific parts of the hearing loss compensation software are active, such as sound environment compensation, e.g. focus on a single talker or conversation in a noisy environment (cocktail party effect), etc. The “where” may include, but is not limited to, location, based on for example input from GPS information, accelerometers or a specific meeting room information from a calendar, but also which type of sound environment the user is in. Sensor data may include, but is not limited to, temperature, pulse, and oxidation saturation. Data obtained from a hearing aid can be used to analyse the user and the user's actions, and thereby provide a way for improving the usability of and satisfaction with the hearing aid.

[0070] For example, if using the sound environment as a parameter, one will in general see the effect of dissatisfaction in the way people navigate sound environments, e.g. which environments they linger in and which they try to avoid. This could, for example, mean: increased time in quiet environments, decreased time in noisy environments, and/or decreased time in speech-and-noise environments.

[0071] In FIG. 1, data is obtained from a hearing aid belonging to a user in step S10. If the data is to be analysed on a data processing system outside the hearing aid, the data may be transmitted from the hearing aid to the data processing system via e.g. the internet or a wireless protocols such as Bluetooth, Wi-Fi, NFC, etc. The data processing system may also be comprised within the hearing aid and the data obtained via communication pathways within the hearing aid.

[0072] After obtaining data from the hearing aid, a determination is made of a prediction score in step S20 indicating the likelihood of the user being dissatisfied with the hearing aid, where the determination is made based, at least in part, on the obtained data. The prediction score is an indicator of whether it is likely that the user is satisfied or dissatisfied with their hearing aid and may be the result of a predictive model built from past data.

[0073] Likelihood of dissatisfaction, if predicted using past data, could be given by the likelihood of the customer returning their devices and the likelihood could be indicated by a number that is returned by a machine learning model. A machine learning model is trained on a training set from a data lake, i.e. a repository of data, or from a database and it creates an internal representation of those who return their hearing aids and those who do not based on predetermined interaction parameters. Patterns of user behaviour are compared by the model to its trained internal representation and assigned a likelihood based on how close that comparison is.

[0074] Using past data to build a predictive model, could, for example, be achieved by comparing data recorded for a period of time from the hearing aid of users, who returned their hearing aids to that from the hearing aid of users, who did not return their hearing aids. The differences and/or trends in the data recorded from a significant number of users can be used to build a model forming part of the determination of a prediction score.

[0075] A machine learning model is made specific to the task and its algorithm will learn and improve as new data is fed into it. As more data is added, the model becomes more refined. The model may use raw data, i.e. the data obtained directly from the hearing aid, or processed data. The data obtained from the hearing aid may be processed in a number of known ways such that it is not the raw data that is used to determine the prediction score, but processed data. For example, simple calculations, where data is added, subtracted, etc. may be performed on the raw data. As another example, raw data may be combined to obtain a new type of data, which is not obtained directly from the hearing aid, but produced using raw data.

[0076] In FIGS. 2 and 3 are shown examples of data, which may be utilised in the determination of a prediction score (see further description of FIGS. 2 and 3 below). The five types of interaction parameters shown in FIG. 2 and the parameter shown in FIG. 3 appear to exhibit high confidence in predicting whether the user of the hearing aid returns the hearing aid or not. One example could be to monitor the sequence pattern of data obtained from the hearing aid, for example one or more of the type of data shown in FIG. 2a-e, and determine a prediction score based on the obtained data, where the prediction score then gives an indication of whether the user is likely to return the hearing aid and thereby an indication of user dissatisfaction.

[0077] Using the parameters use-time, number of volume changes, number of re-boots, number of pre-set changes and number of power downs, a machine learning model was achieved, which could in 77% of the cases correctly identify a user, who returned the hearing aid and in 70% of the cases correctly identify a user, who did not return their hearing aid. In that setup, mean sequence data up to the return from some weeks before the return were used, so the dynamic behaviour of the parameters was included.

[0078] The data obtained from the hearing aid may be obtained over a period of time, such as within a short-to-medium time frame, for example during a 90-day trial period. It may also be obtained long after the initial use of the hearing aid to continuously ensure satisfaction with the hearing aid. Even though the user may not be able to return the hearing aid after months or years of using it, the monitoring of data from the hearing aid and the therefrom determined prediction score can continue to provide an indication of user satisfaction. The obtained data may also be data collected within a very short time frame, such as a week, a day, or even hours, minutes or seconds, before the data is used in the determination of a prediction score.

[0079] The factors used in determining the prediction score may be a simple number such as use-time in hours, but it may also be a more complex interaction between the user and the hearing aid such as e.g. the change of the pre-set/program or activation of the volume control in a specific time pattern. Such complex interactions lend themselves to be analysed in a machine learning approach, where patterns in the data are discerned by an artificial intelligence algorithm. A machine learning model such as a neural network that is sensitive to sequence information, e.g. 1D ConvNets, can be trained to distinguish between users, who return their hearing aids to those who do not by learning the trends in the data parameters of those who return their hearing aids. Thus, the step of determining a prediction score may be at least partly performed using machine learning and/or artificial intelligence. For example, the step of determining a prediction score may be at least partly based on a model made using machine learning.

[0080] The prediction score may be a number and the value of the prediction score can be compared with a predetermined critical value, which separates the indication of satisfied from that of dissatisfied. For example, if the prediction score is e.g. higher than a predetermined value, the user may be categorised as dissatisfied. The prediction score may alternatively be expressed in a more complex manner than a single number, for example as several numbers, or as a letter and a number. Any labelling that allows for a decision to be made of whether the user is indicated as being satisfied or dissatisfied may be used.

[0081] If the prediction score indicates that the user is dissatisfied, a response measure is executed in step S30 of FIG. 1. The response measure will comprise adjusting the hearing aid functionality, or arranging for human support. Which response measure is chosen can be based, at least in part, on some or all of the data obtained from the hearing aid. For example, if the user changes pre-sets/programs often, this could indicate, possibly together with other data, that the user is dissatisfied with the programs and an update of one or more pre-sets/programs may be selected as response measure to try and improve the user experience. If the volume of the hearing aid is changed often, this may indicate, again possibly together with other data, that the hearing aid was not calibrated properly to the user's hearing loss and a suitable response measure may be notifying a hearing care professional such that a new calibration may be performed.

[0082] In this way the data collected on the user and the user's interaction with the hearing aid provides a data-driven approach to predict, whether a user is dissatisfied with their hearing aid, allowing for measures to be initiated to improve the usability and satisfaction with the hearing aid without having to directly contact the user to learn whether they are satisfied with their hearing aid.

[0083] In FIG. 2 is shown graphs of mean sequence data of five parameters 12 weeks prior to the last data logging before the hearing aid was returned compared with the same type of data from non-returns. The five parameters are (a) use-time [h], (b) number of pre-set/program changes, (c) number of power downs, (d) percentage of users with at least one volume change, and (e) number of re-boots, all as a function of weeks. The data is based on 4000 non-returns and 2000 returns. For all of the parameters in FIGS. 2a-e, a trend can be ascertained for the return cases versus the non-returns, thus providing a possibility of creating a predictive model.

[0084] In FIG. 3 is shown another example of data, which may be utilised in the determination of a prediction score. Shown is the percentage (%) of hearing aids versus the number of daily pre-set switches 8 weeks prior to the last data logging before the hearing aid was returned compared with the same type of data from non-returns. The data shown is based on 2300 returns and 11000 non-returns. In the graph the data from returns is shown in black and the data from non-returns is shown in grey. It shows that daily pre-set switches are higher for hearing aids that are returned compared to hearing aids that are not returned. Using these data, a machine learning model was achieved, which could in 72% of the cases correctly identify a user, who returned the hearing aid and in 96% of the cases correctly identify a user, who did not return their hearing aid.

[0085] FIG. 4 schematically illustrates a system comprising a hearing aid and configured to perform the method of improving usability of and satisfaction with a hearing aid in accordance with exemplary embodiments. A user 1 is wearing a hearing aid 3, which collects data on the user and the user's behaviour such as e.g. use-time, number of pre-set/program changes, number of power downs, number of re-boots, number of battery charges, number of sound environment changes, pattern of sound environment changes, time spent in a type of sound environment, location, temperature, pulse, oxidation saturation. This data can be used in a number of ways and may be used by a data processing system 9, which is configured to obtain data from a hearing aid, determine a prediction score and execute a response measure.

[0086] In the embodiment shown in FIG. 4, the data processing system 9 is comprised in a remote server 5 and the hearing aid 3 is configured to communicate with the remote server 5 such that data transmission 7 between the hearing aid 3 and the remote server 5 is possible. The data transmission 7 between the hearing aid 3 and the remote server 5 may take place via software, for example an app, running on an external device such as e.g. a mobile phone.

[0087] The data processing system 9 obtains data via the data transmission 7 and determines a prediction score, which is at least in part based on the obtained data, but can also be based in part on user-specific data. The user-specific data could be, for example, type of the hearing aid, model of the hearing aid, age, gender, socioeconomics, hearing loss profile, user feedback rating provided, number of contacts to a hearing care professional, number of days since last contact with a hearing care professional, and use-time of a linked app. Such user-specific data could be obtained remotely, i.e. from outside the hearing aid, for example from one or more databases or external devices. In the embodiment shown in FIG. 4, user-specific data could be available on the remote server 5. Such user-specific data obtained remotely could link the information to the hearing aid ID and thereby link it to data obtained from the hearing aid 3.

[0088] Further, data from the hearing aid generated during test and/or manufacturing may also be used in determining the prediction score. The prediction score indicates the likelihood of the user being dissatisfied with the hearing aid and if the prediction indicates dissatisfaction, a response measure is executed.

[0089] The data transmission 7 may be performed regularly or sporadically. When using a predictive model based on past data, for example from comparing data recorded for a period of time from the hearing aid of users, who returned their hearing aids to those from the hearing aid of users, who did not return their hearing aids, to determine the prediction score, the predictive model may be continuously or periodically updated. The remote server 5 can be connected to a plurality of hearing aid users from which it receives data such that the predictive model can improve over time. The remote server 5 may comprise a machine learning algorithm, which analyses the data, for example by looking for trends in the data parameters of those users, who return their hearing aids, compared to those users, who do not. Alternatively, the remote server 5 may be connected to a system comprising a machine learning algorithm.

[0090] FIG. 5 schematically illustrates another system comprising a hearing aid and configured to perform the method of improving usability of and satisfaction with a hearing aid in accordance with other exemplary embodiments. As in FIGS. 4 and 6, a user 1 is wearing a hearing aid 3, which collects data on the user and the user's behaviour. In the embodiment shown in FIG. 5, a data processing system 9 is comprised in the hearing aid 3 and the data processing system 9 obtains data via communication pathways within the hearing aid 3. The data processing system 9 in the embodiment shown in FIG. 5 is configured to obtain data from the hearing aid 3, determine a prediction score, which is at least in part based on the obtained data, and execute a response measure. Executing the response measure may mean that the hearing aid arranges for a response measure to be implemented.

[0091] The prediction score may be a result of using a predictive model based on past data, for example based on a model obtained by comparing data recorded for a period of time from the hearing aid of users, who returned their hearing aids to those from the hearing aid of users, who did not return their hearing aids. The data processing system 9 within the hearing aid 3 may comprise software, which executes the predictive model. The predictive model may be updated regularly or periodically either by a software update or by a machine learning algorithm comprised in the data processing system 9.

[0092] To update the software or the machine learning algorithm, or to gather data from other hearing aid users, for example for use in creating a predictive model, the hearing aid 3 may have a means of wired or wireless communication 13 with an external system, for example wireless communication with a remote server 5 as shown in FIG. 4 or with an app running on an external device 15, where the external device may communicate with another system such as a remote server 5.

[0093] FIG. 6 schematically illustrates yet another system comprising a hearing aid and configured to perform the method of improving usability of and satisfaction with a hearing aid in accordance with other exemplary embodiments. As in FIGS. 4 and 5, a user 1 is wearing a hearing aid 3, which collects data on the user and the user's behaviour. In the embodiment shown in FIG. 6, as in FIG. 5, a data processing system 9 is comprised in the hearing aid 3 and the data processing system 9 obtains data via communication pathways within the hearing aid 3.

[0094] To acquire data from other hearing aid users, the hearing aid 3 has a means of wired or wireless communication 13 with an external system, for example wireless communication with a remote server 5. The remote server 5 has a database 11 comprising data from other hearing aid users, which the data processing system 9 may use in its determination of a prediction score.

[0095] The data from other hearing aid users may be data, which is or can be separated into data from users, who returned their hearing aids, and users, who did not return their hearing aids.

[0096] In all embodiments, a response measure is executed if the prediction score indicates that the user is dissatisfied and the response measure comprises adjusting the hearing aid functionality, or arranging for human support.

[0097] For example, a response measure, which adjusts the hearing aid functionality, could comprise one or more of the following adjustments of the hearing aid functionality: reinstalling software on the hearing aid, updating software on the hearing aid, changing one or more algorithm parameters, performing remote automatic fine-tuning of the hearing aid, and/or updating one or more pre-sets/programs on the hearing aid.

[0098] Alternatively, the data processing system 9 may be comprising partly within the hearing aid 3 and partly outside the hearing aid, for example within a remote server 5, such that one or more of the method steps are performed by circuitry within the hearing aid 3 and the rest on circuitry comprised outside the hearing aid.

[0099] If the data processing system 9, or part of it, is comprised in a remote server 5, it may execute one or more adjustments of the hearing aid functionality by pushing them to the hearing aid 3 or it may await a request. For example, the hearing aid 3 may periodically request updates and/or fine-tunings.

[0100] If the response measure is arranging for human support, it could comprise, for example, notifying the hearing aid user, notifying a hearing care professional, and/or notifying a customer service employee. Notifying the user 1 of the hearing aid 3 could be achieved, for example, via an app or via a communication means comprised in the hearing aid 3.

[0101] Although particular features have been shown and described, it will be understood that they are not intended to limit the claimed invention, and it will be made obvious to those skilled in the art that various changes and modifications may be made without departing from the scope of the claimed invention. The specification and drawings are, accordingly to be regarded in an illustrative rather than restrictive sense. The claimed invention is intended to cover all alternatives, modifications and equivalents.

LIST OF REFERENCES

[0102] 1 User [0103] 3 Hearing aid [0104] 5 Remote server [0105] 7 Data transmission [0106] 9 Data processing system [0107] 11 Database [0108] 13 Wired or wireless communication [0109] 15 External device.