aboutusnew

2018

Annual Report 

Director's Report

A retrospective on 2018 and an outlook on 2019: What really counts

Eriksholm continues to develop as a turning axis within global research in hearing healthcare. Looking back at 2018, particularly three highlights pop to my mind.

For the first time, there is a Nature publication stemming from Eriksholm with two authors from William Demant; Thomas Lunner from Eriksholm Research Centre and James Harte from Interacoustics Research Unit. The paper was triggered by a meeting between Anders Fridberger, Thomas Lunner, and Brian Moore back in 2009.

Since then, it has developed to a paper with many prominent authors, revealing a new mechanism in understanding how the ear decodes the envelope of a signal. “A mechanoelectrical mechanism for detection of sound envelopes in the hearing organ” was published in Nature Communications on the 9th of October 2018.

This paper shows important facets of our work at Eriksholm: First of all, we are fortunate to be closely connected with the best and brightest scientists around the world. Science today is teamwork and (luckily) it does not know borders.

What is also important to note is the fact that science takes time. In this case from 2009 to 2018, that is nine years, certainly not something to measure in quarterly reporting. Many groundbreaking scientific results come from years and years of research and mutually inspirational networking. Externally and, in our case, also internally across the various organizations within William Demant.

Here at Eriksholm, we are supported by an organization in which it is acknowledged that good results come from long term collaborations. That may sound trivial, but any manager with experience in big corporate organizations knows that internal collaboration sometimes can be more difficult than external ones.

uwe_udendrs
"What we are really passionate about at Eriksholm is helping people with hearing loss and the hearing care professionals who strive to improve their lives."

 

- Senior Director Uwe Hermann

Another 2018 highlight was the invitation from the editor of the highly ranked 'Journal of the Acoustical Society of America'(JASA) to Lars Bramsløw, artificial intelligence expert and member of our Augmented Hearing research group, to become reviewer.

Eriksholm researchers are frequent reviewers of international papers. This invitation, however, was significant because it was explicitly related to topics in artificial intelligence where typically American and Chinese research institutions are leading. Sending an invitation to a European based research centre is special, and receiving it makes us really proud. 

We see this as a recognition of the ground-breaking publication from our Augmented Hearing group earlier this year. In close collaboration with our academic partners at Tampere University in Finland, we took one big step towards “cracking the hardest nut” in hearing healthcare, the so called “Cocktail Party Syndrome”, by testing the application of Deep Neural Networks to the segregation of unknown voices.

The third 2018 highlight I would like to mention is the William Demant internal transfer of former eHealth research area manager Ariane Laplante-Lévesque from Eriksholm to a new position in Oticon Medical. On the one hand a loss for Eriksholm, but on the other hand an important and good step for growing the William Demant internal collaboration and know-how transfer. Remember: the best way to transport know-how is between the ears.

The process of hiring a replacement proved that Eriksholm is an attractive workplace. We received applications from many highly skilled scientists and ended up offering the position to the distinguished research scientist Gaby Saunders from VA National Center for Rehabilitative Auditory Research (NCRAR) in Portland, Oregon. We are excited to have Gaby on board and wish her all the best in leading the way for our future eHealth research. 

So, what will come in 2019? 

Nobody knows, but I do have a hunch we will hear more about ground breaking applications of artificial intelligence in hearing healthcare.
We will see how multiple sensors become part of hearing instruments, and how they become instrumental on the way to more holistic healthcare.

eHealth will grow and become more and more important. Unfolding its potential will depend on many factors, but the driving elements are obvious: the cost pressure in all global healthcare systems to deliver more for less and the fast-growing capabilities of AI.

What really counts is the patient centered approach: “People First”. What we are really passionate about at Eriksholm is helping people with hearing loss and the hearing care professionals who strive to improve their lives.

To achieve this ambitious goal, we enjoy working with collaboration partners from all over the world. Research does not know borders. We know how little we know ourselves and how much we depend on our global academic partners as well as on our colleagues within William Demant.

A big “thank you” to all of you!

 

- Uwe Hermann


Augmented Hearing

The vision for Augmented Hearing goes beyond full compensation for lost fidelity and muddy contrasts in impaired hearing.

Why should people with hearing impairment not appreciate all details of music and be able to process independent sound streams while easily switching attention between them? Why not surpass normal hearing?

Augmentation aims at enhanced speech perception, enhanced music fidelity, enhanced perception of environmental sounds, and the enhanced ability to perceive individual sounds as individual sounds. The field of hearing research delivers many significant insights on how hearing impairment changes the perception of multiple sounds and subsequently affects the neural encoding of sounds in the brain.

These insights provide our research with specific goals, e.g. finding ways to improve the ability to segregate competing voices or ways to improve the perception of melodic cues in music. Augmented hearing advances from integrating progress in machine learning and artificial intelligence. This rapidly expands the possibilities to fulfil our dreams and visions.

Along with the new possibilities comes a need for even more personalization. Once again, data and machine learning provide interfaces that help us understand individual coping strategies in order to make hearing devices adopt to your life and hearing.

Collaboration is essential for Augmented Hearing. We are grateful for our achievements with partners in the EVOTION project, at Tampere University, at Technical University of Denmark, and at University of Southern Denmark.

Augmented Hearing is headed by Niels Henrik Pontoppidan. Read more about the projects of the group here. 


lars_ihcon3

Senior Scientist Lars Bramsløw presenting at IHCON 2018

Enhanced segregation of voices

While people with normal hearing take segregation for granted, people with hearing impairment struggle with competing voices. Research with EEG has shown that people with hearing impairment lose the ability to filter out unwanted voices and noises and thus the ability to tune in and tune out of voices and sounds.

Our collaboration with Tampere University has led to a method where Deep Neural Networks (DNN) are trained to separate two known voices in order to present the voices separately in each ear. The listening tests with people who have hearing loss showed that the DNN separation enhances the segregation of voices by raising the speech recognition from 52 % to 64 % when listeners have to pay attention to both voices. It improves the speech recognition from 58 % to 95 % when the listener pays attention to only one voice.

dnn_medteskst

Figure 1: Word recognition scores as interaction effect of processing mode and test mode, including the three DNN modes. Vertical bars denote 95 % confidence intervals.

In our quest to bring solutions to family dinners and other social situations, we looked further into the brain’s encoding of dichotic separated voices in collaboration with the colleagues in Cognitive Hearing Science in 2018. 

Starting with the same DNN separation as in the 2018 JASA paper, the outcome was extended with scalp EEG recordings to correlate the speech envelopes and the EEG brain waves.

The preliminary findings indicate that the dichotic presentation of DNN separated voices increases the suppression of the unwanted voice as the corresponding correlation between EEG and unwanted speech envelope decreases when voices are separated with DNN. The correlation between the EEG and the wanted speech envelope remains the same.

The results of the competing voices test are very promising as they indicate that separation with DNN improves the brain’s ability to process the two voices as two separate objects leading to improved recognition.

Digitized hearing care

Many press releases from CES (Consumer Electronics Show) in Las Vegas announced products interfacing the virtual assistants from Amazon and Google. Interfacing a product with those virtual assistants enables control via conversational interfaces, thus people can interact with their devices by uttering commands which the virtual assistants transform into actions for the device.

We now ask: can virtual assistants allow hearing device users to interact through voice, and especially how can the interaction take place in natural language not limited by the wording chosen by the device but customized to the words which a hearing device user uses naturally?

In our vision the conversational interface allows the user to express wishes rather than being limited to keywords like program names and volume adjustments. As a fully digital solution, the dialogue is joined with data from the device describing the sound environment and the current settings.

This is an assistant that helps the user navigate the settings space and keeps a diary. And by transforming the logged data and interactions into compelling dashboards, it empowers the next dialogue with the hearing care professional so the two can reach individually optimized settings faster.


evotion-fig

Figure 2: Aggregated usage patterns per hour across the day for the four EVOTION settings.

EVOTION

The Horizon 2020 project EVOTION enters it’s final year, and this is the time to bring the different achievements together to one big achievement. The EVOTION project aims to facilitate public health policies based on evidence stemming from data collected from EVOTION participants wearing hearing devices and their corresponding clinical health records from audiology departments in the consortium.

While the consortium is busy transforming the collected data into demonstrators of public health policy, the data also reveal interesting patterns of usage. Aggregating the usage of the different settings over the hours of the day as showed in Figure 2 reveals three distinct patterns:

• The default settings are generally used the most, and the usage is almost constant throughout the day.

• The settings with increased noise reduction seem to be used more as the day progresses.

• Finally, a setting that just amplifies without noise reduction seems to be used more around lunch and late dinner time. 

EVOTION shows the very short distance between research and products, as HearingFitness™, launched by Oticon in 2018, reused the data-logging interface of EVOTION with few updates. While HearingFitness™ is not a copy of EVOTION it re-uses the underlying technology to help people with hearing aids benefit more from using them.

Moreover, EVOTION is the first project paving the way for future projects where Eriksholm shares data from hearing devices with project partners to develop new and improved services for people who wear hearing devices.

Cognitive Hearing Science

Within the Cognitive Hearing Science area our focus is "Hearing & Cognition". This includes research in various aspects of cognitive processing assessed by pupillometry, eye gaze estimation, brain wave analysis, and behavioral performance during listening tasks. This knowledge is used for various purposes such as developing new ways to steer future hearing aids and evaluating hearing aid outcome.

Our team has already demonstrated remarkable results in utilizing the brain waves and especially eye gaze estimation to provide amplification of the attended speaker in situations reflecting a restaurant conversation. As these results can be obtained by sensors placed in the ear canal, it will be possible to embed the technologies in future generations of hearing aids. A prototype of such a hearing aid has been developed and tested this year on people with hearing impairment, who are very positive towards the user benefit.

Our work has also provided evidence of user benefit in hearing aids under more ecological, realistic, daily-life conditions. This evidence has mainly been based on pupillometry and working memory scores, although new cognitive hearing assessment methods are under development.

Eriksholm’s strategic research area Cognitive Hearing Science is headed by Thomas Lunner. Learn more about Cognitive Hearing Science here.


cocoha_illustration_960x640

Illustration of the aim of COCOHA

Cognitive Control of a Hearing Aid

One of our main achievements in 2018 has been a very successful finalization of the Cognitive Control of a Hearing aid (COCOHA) project. The COCOHA project received funding from the European Union’s Horizon 2020 research and innovation programme and was a collaboration between École Normale Supérieure in Paris, UCL in London, UZH in Zürich, DTU in Copenhagen, and Eriksholm Research Centre. 

The scientific output of the project was a research platform of a hearing aid with integrated sensors to assess head movements and electrophysiology from the ear canal. This platform was used to gain knowledge of improvements in speech intelligibility by cognitive control of a hearing aid obtained by both brain wave and eye gaze estimation.

As a result, we tested the developed platform on several hearing impaired persons, and some of the impressions from the eye gaze steering are provided in the video below.

In the other video, you can learn more about the COCOHA project. You can also take a look at the official COCOHA website.

New project about identifying listening effort in individual listeners

In 2018, the PUPILS project got funded by the Oticon Foundation. PUPILS is a cooperation with DTU Hearing systems group (Professor Torsten Dau) and DTU Compute (Per Bækgaard), and the project aims to examine the feasibility of the pupillometry method for identifying listening effort in individual listeners.

Recent research from Eriksholm and partners gathered evidence that listening effort is affected by both hearing impairment and the applied signal processing in a hearing aid (Wendt et al. 2017; Ohlenforst et al. 2017). We demonstrated a benefit of the noise reduction scheme OpenSound Navigator on listening effort for people with hearing impairment using pupillometry (Wendt et al. 2017).

Those findings support the assumption that a more complete characterization of difficulties in speech perception arising as a consequence of hearing impairment can be gained when testing listening effort using pupillometry. However, the pupillometry method has only been considered on a group level and no attempts have been made to evaluate the method in individual listeners.

PUPILS aims to make this transition from a group level towards an individual listener level. This is challenging since the underlying components affecting the pupillary response are not fully understood yet.

The overall goal is to study whether a reliable estimate of the individual listening effort can be assessed by pupillometry. If the reliability of this method can be proven, it facilitates the development of the procedure and the test protocol towards a standardized test feasible for clinical use.

hear-eco-kickoff_960x640

The six HEAR-ECO early stage researchers and their supervisors at the HEAR-ECO kick-off meeting

HEAR-ECO

Six highly motivated early stage researchers have been recruited within the Horizon2020 EC Marie-Curie ITN project “Innovative hearing aid research – ecological conditions and outcome measures” (HEAR-ECO) in 2018.

Coming from different disciplines including audiology, (neuro-) psychology, mechanical & biomedical engineering, and acoustics, all six researcher started their PhD projects sucessfully. The six PhD projects will explore and develop new tools and methods for examinating the impact of hearing-impairment and hearing-aid signal procesing on listening effort in everyday communication.

The students are enrolled either at the VU University medical center in Amsterdam or at the University of Nottingham (Hearing Sciences – Scottish Section), but since Oticon is their industrial partner, they will spend half of their time at Eriksholm Research Centre. 

Find more about the individual projects here.

CHS publications about pupillometry

In 2018, two publications came out that demonstrated signs of disengagement when listening to speech in very noisy environments.

By the help of pupillometry, it was shown that effort investment is high in listening situations with around 50 % speech intelligibility. However, a drop in effort mobilization can be observed with decreasing intelligibility. This drop in effort mobilization is interpreted as a sign of disengagement for noisy environments where speech intelligibility becomes very hard and almost impossible. This disengagement was found for people with and without hearing-impairment. Learn more about those studies here: Wendt et al. 2018 and Ohlenforst et al. 2018.

More publications dealing with pupillometry can be found in a special issue in Trends in Hearing dedicated to a workshop called “The Second Workshop on the Application of Pupillometry in Hearing Science". The workshop assessed listening effort and was held in Amsterdam in 2017. Eriksholm contributed with three articles. 


annual-report-2019---facilitating-the-development-of-young-researchers

Facilitating the development of young researchers

The Cognitive Hearing Science group is a great scaffold for the development of student researchers. We support the academic growth of students through training, supervision and guidance, aiming to help them reach the peak of their potential and develop good research practices for the future. In 2018, ten students from various universities completed their bachelor’s and master’s projects in the field of cognitive hearing science.

Bachelor's projects:

• Johan Nyberg and Elias Nyhlen from University of Gothenburg : ‘Pupillometry and perceived effort in difficult listening situations’

• Lotte Simone Emilie Petersen from University of Southern Denmark: ‘Cleaned vs. uncleaned ears: A comparative in-ear dry-electrodes performance study’

Master's projects:

• Sanne Mehrfeld Møller from University of Southern Denmark: ‘Listening effort by combining pupillometry and the Danish SWIR’.

• Martin Sollenberg and Johan Lindqvist from Lund University ‘Real-Time Multiple Audio Beamforming System’.

• Andrew Segar and Carlos Francisco Mendoza Lara from Lund University: ‘Auditory attention modulation and understanding the relation between sound and the ‘listening brain’’.

• Murron Harrison from University of Glasgow: ‘Evaluating the Use of Steering a Hearing Aid in A Dynamic Multi-Talker Environment Using Body Signals’.

• Golam Reza Sadeghnia from Technical University of Denmark: ‘Development and Evaluation of an External, Steerable Microphone Array for Improving Spatial Selectivity of Hearing Aids’.

eHealth

In 2018 the eHealth team has undergone a series of transitions. The most important one being Gaby Saunders joining the team in August as new eHealth research area manager.

Despite the changes, the eHealth team has been very productive. We have spent the year finalizing older projects, planning future projects, and defining the strategic direction of the research group.

Eriksholm's strategic research area eHealth is headed by Gaby Saunders. Learn more about eHealth here.

ehealth_group

The eHealth team

The new eHealth strategy

The new eHealth research strategy will focus on applying eHealth at every stage of the hearing healthcare journey. Our research will support hearing healthcare access for every individual with hearing loss by examining how best to increase access and motivation to individuals with hearing loss who have not yet sought help for their difficulties.

We will also work to support individuals with hearing aids by examining how best to achieve improved quality of life through person-centered care in combination with personalized hearing devices that adapt to the user’s listening conditions and needs.

Finally, our research will aim to broaden the scope of hearing healthcare to include early identification and intervention for conditions associated with hearing loss, such as vestibular problems and cognitive decline.


Projects finalized in 2018

The projects finalized this year resulted in seven publications in peer-reviewed journals. These papers reported on topics as diverse as:

• exploring hearing aid problems from the perspectives of hearing aid owners and clinicians

• eHealth and the hearing aid adult patient journey

• user experiences with an internet-based aural rehabilitation program

• audiologic considerations for people with normal hearing sensitivity yet hearing difficulty and/or speech-in-noise problems

• using big data for evidence-informed hearing health policies

• a review of the Third Internet and Audiology International meeting that took place in 2017 in Louisville, USA.


evotion_film

One of the EVOTION educational videos

EVOTION

This year, the eHealth team was very much involved in the ongoing EU Horizon2020 funded EVOTION Research and Innovation Project that is being conducted in collaboration with 12 partners across seven countries in the EU.

The primary goal of EVOTION is to build an evidence base and platform for formulating public health policy pertaining to prevention, early diagnosis, and treatment/rehabilitation for hearing loss.

This past year, Johanna Gutenberg, Lukas Tietz, and Gaby Saunders participated in the project by contributing to planning and development of the public health policy decision-making platform that will be one of the tangible end products of the work, developing educational user videos and website content, and conducting a business case analysis for the policy making platform. For more information see http://h2020evotion.eu/ and the Augmented hearing section of this annual report. 

internet-and-audiology_960x500

The Internet & Audiology meeting in 2017

Internet & Audiology meeting

Planning for the fourth Internet and Audiology meeting is well underway, under the guidance of Gaby Saunders from Eriksholm Research Centre, Jill Preminger from University of Louisville, and Helen Cullington from University of Southampton. These meetings provide a forum for scientists and clinicians to share and learn about innovative audiological research and clinical practices that have arisen from the field of eHealth.

The 2019 conference will take place at the University of Southampton, UK, on 17-18 June 2019. For more information and to registration, submit an abstract for presentation and/or submit an application for a travel award go to: http://www.internetaudiology.com/2019/

If you are interested in learning more about the 2017 meeting, see the special issue of the American Journal of Audiology 2018 Volume 27 issue 3s in which 14 selected articles arising from presentations at the conference are featured. Note that two of these articles describe eHealth research projects that took place at Eriksholm.

2018 in numbers

Find more details at our Publications page and our Academic Relations page.

24 papers published in peer reviewed journals
6 papers published in conference proceedings
1 book chapter published
14 conference posters
21 conference presentations
Participation in 41 conferences
33 manuscripts reviewed for peer reviewed journals
Supervision of 9 students at Eriksholm Research Centre
Supervision of 26 students external to Eriksholm Research Centre
7 external examinings


According to Google Scholar, researchers working at Eriksholm Research Centre in 2018 were cited 1,462 times throughout 2018. According to Elsevier Scopus®, researchers working at Eriksholm Research Centre in 2018 were cited 966 times including citations from authors and 766 times excluding citations from authors throughout 2018. All Eriksholm Research Centre publications in the Elsevier Scopus® database was cited 707 times in 2018.