Speech Technology in Dutch Health Care: A Qualitative Study

Ellen Luchies, Marco Spruit and Marjan Askari

Department of Information and Computing Sciences, Utrecht University, Utrecht, The Netherlands

Keywords: Speech Technology, Speech Recognition, Natural Language Processing, Health Care, Dutch Health Care.

Abstract: This study investigates the opportunities of speech technology in Dutch hospitals, and to what extent speech

technology can be used for documentation. Furthermore, we clarify why speech technology is used only

marginally by Dutch hospital staff. We performed interviews where speech technology users, managers in

hospitals and software suppliers were contacted as participants. We then transcribed our interviews and

synthesized the pros and cons of speech technology as well as major barriers for the adoption. Our results

show various influencing factors that could be clarifications for the fact that only 1% of the medical staff uses

speech technology in the Netherlands. The major reasons we found are: speech technology usage at only

radiology and pathology departments, smarttexts and smartphrases of the Electronic Health Record (EHR)

compete with speech technology, caregivers have to adjust their way of working which evokes resistance,

lack of central authorization at Dutch hospitals and finally, financial barriers. Our results show that speech

technology works for radiology and pathology as a tool for documentation, but is found less useful for other

departments. For the remaining departments, different applications show potential, such as structured

reporting.

1 INTRODUCTION

Caregivers, specifically nurses and physicians,

experience the highest work load compared to many

other professionals (NOS, 2017). According to

Schumacher (2017), it is to be expected that

physicians’s workload will increase even more in the

upcoming years, among other things due to the

increasing amount of people who need care in

western countries. This increase is caused by aging of

the population (Schumacher, 2017). The groups aged

65+ and 80+ are increasing rapidly, as shown in

Figure 1. As people age, they need more care as a

result of physiological and psychological

weaknesses, as is shown in Figure 2. This

phenomenon is called multimorbidity.

A more efficient way of working is therefore

necessary to be able to cope with the increasing

workload in health care. Speech technology can offer

a solution for this problem (Ajami, 2016). This

technology has seen major improvements in the last

decade (Parente, Kock and Sonsini, 2004; Ajami,

2016).

Figure 1: The aging Dutch population (Schumacher, 2017).

According to Ajami (2016), speech technology

can contribute to a more efficient way of working.

Physicians are able to document faster, and make

reports available faster (Ajami, 2016). Nowadays,

many systems reach an accuracy up to 98% (Parente,

Kock and Sonsini, 2004; Johnson et al., 2014; Ajami,

2016). Nevertheless, speech technology is used by

only 1% of the Dutch hospital staff (Nuance, 2015).

The aim in this study was therefore to investigate the

barriers and potentials of speech technology in Dutch

health care.

Luchies, E., Spruit, M. and Askari, M.

Speech Technology in Dutch Health Care: A Qualitative Study.

DOI: 10.5220/0006550103390348

In Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2018) - Volume 5: HEALTHINF, pages 339-348

ISBN: 978-989-758-281-3

339

Figure 2: Amount of weaknesses per agegroup

(Schumacher, 2017).

2 BACKGROUND

2.1 How Does Speech Technology

Work?

Digital dictation can be seen as the predecessor of

speech technology. It worked as follows: the doctor

dictated the report, sent the dictation to the secretary

who transcribed the dictation. This document was

then sent back to the doctor for validation, and the

report was made available (S1, see Table 2 below).

However, since speech technology is available to

doctors, they are able to dictate using a computer. The

users speaks, and the system converts the speech into

words on the screen. After this phase, the doctor has

to correct the document to filter out errors.

Previously, this was done by the secretary (S1).

Speech technology uses different tools to support a

qualitative conversion from speech to text.

 Acoustic model (Koivikko, Kauppinen, and

Ahovuo, 2008): this model defines how sounds

are pronounced. (Renckens, 2009).

 Speech corpus: this corpus defines the

different ways so every sound and phoneme

can be recognized (Renckens, 2009) despite the

fact that each individual pronounces a sound

different (Ajami, 2016).

 Lexicon (Koivikko, Kauppinen, and Ahovuo,

2008) (Ajami, 2016): this is the dictionary of

the computer (Renckens, 2009). It contains the

words that are recognizable by the system. If a

word does not occur in the dictionary, the

system is not aware of the existence of that

word, and is therefore not able to recognize it

(S2). In addition, a phonetical transcription is

available for each word in the dictionary

(Renckens, 2009). The quality of speech

technology strongly depends on the dictionary

that is used. When a dictionary contains many

words, the system is more likely to confuse

words with each other, which leads to more

mistakes and a lower accuracy (Ajami, 2016).

 Language model (Koivikko, Kauppinen, and

Ahovuo, 2008; Ajami, 2016): the language

model is a statistical model. It calculates the

likelihood that words are related and occur in a

certain sequence, based on previously spoken

reports in the database (Renckens, 2009). An

advantage of this is the possibility to construct

a word or sentence based on statistics when the

system is not able to do this based on speech. A

disadvantage of this is the fact that uncommon

words will not be chosen because common

words are more likely to be used according to

the databases (Vervoort, 2009).

2.1.1 The Process from Speech to Text

The process to construct a word from speech is shown

in figure 3. This figure is based on studies of Vervoort

(2017), Renckens (2009) and Geitgey (2016). It starts

by recording speech with a digital voice recorder.

These sound waves are segmented by the computer

(Renckens, 2009; Vervoort, 2017).

In the next step, the segments are converted into

numbers by sampling. This is a technique that

measures the height of the sound wave on equally

scattered points in the wave (Vervoort, 2017). After

sampling, the signal must be filtered to reduce

background noise (Vervoort, 2017). By measuring

the amount of energy in the sound waves, a

spectrogram is created (Vervoort, 2017). This is seen

as a fingerprint of the dictate (Geitgey, 2016).

Subsequently, the spectrogram is used as the input

for a neural network. The output represents the

likelihood per phoneme. To compute this likelihood,

the neural network uses the acoustic model, speech

corpus, lexicon and language model (Renckens,

2009).

For speech technology, a Recurrent Neural

Network (RNN) is used that saves previous

calculations to influence future calculations (Geitgey,

2016). This way, speech technology is a learning

system which improves itself (Geitgey, 2016).

HEALTHINF 2018 - 11th International Conference on Health Informatics

340

Figure 3: Process model from speech to text.

Speech Technology in Dutch Health Care: A Qualitative Study

341

After the calculations of the likelihood per

phoneme, a scheme representing these likelihoods is

constructed (Geitgey, 2016). This is the last step as

shown in Figure 3. In this example, the word ‘hello’

is constructed. The predictions of this scheme are

sequenced with double characters and gaps in the

word. When these are filtered, three words are still

possible: ‘hello’, ‘hullo’ and ‘aullo’. Since ‘hello’ is

more likely because it occurs more often in the

database than the other two options, ‘hello’ is chosen

(Geitgey, 2016). In case another word was meant, the

user has to correct it manually. This improvement will

then be saved for future predictions (Renckens,

2009).

2.2 Pros and Cons of Speech

Technology

In this section we state the pros and cons according to

literature enlightened by interviewees. The main

advantage of speech technology is time reduction

(Ajami, 2016; Koivikko, Kauppinen, and Ahovuo,

2008). According to a study of Nuance, people can

type 40 words per minute at best, whereas people can

speak 120 words per minute (Nuance, 2008).

Furthermore, Nuance (2015) states that doctors are

documenting 13.3 hours a week on average. For

nurses, this is 8.7 hours per week (Nuance, 2015).

This concerns an estimated 30% of the working week,

therefore speech technology could be very profitable.

Different studies show that radiology and

pathology benefit most from speech technology

(Ajami, 2016; Johnson et al., 2014). This is clarified

by the fact that radiology and pathology can cut down

on their secretaries when they start using speech

recognition, which leads to a decrease in the report

turnaround time (RTT) (Koivikko, Kauppinen, and

Ahovuo, 2008). Other departments started working

with the Electronic Health Record (EHR) before

speech technology, and already cut down on their

secretaries. Because of this, speech technology lacks

this benefit for departments other than radiology and

pathology, including the decrease in RTT and the

financial benefits of the staffing costs (M1).

Before doctors can start using speech technology,

a profile must be prepared whereby the system gets

familiar with the user’s speech and vocabulary. This

can be done by reading a text aloud (Bosch, 2005).

This is beneficial for the accuracy of the system

(Vervoort, 2017), but takes time (Ajami, 2016;

Johnson, et al., 2014). Speech technology uses a

lexicon, as described in paragraph 2.1. For medical

staff, medical terminology is added, but not

terminology that is used in daily life (S2). A

disadvantage of this dictionary holds that words that

are not included, cannot be recognized by the system

(S2). Patient friendliness increases (Ajami, 2016).

When a doctor types during a conversation, he or she

has less attention for the patient. Using speech

technology, he or she can listen to the patient without

this distraction (U1). The doctor has to dictate during

the conversation, or afterwards, since it is not (yet)

possible for software to recognize two voices at once,

i.e. Advanced Voice Technology (Tuin, 2016).

Besides, reports are available faster (Ajami, 2016)

(Johnson, et al., 2014; Koivikko, Kauppinen, and

Ahovuo, 2008), therefore patients can be cured faster,

which leads to an increased quality of patient care

(Koivikko, Kauppinen, and Ahovuo, 2008; Parente,

Kock, and Sonsini, 2004). A challenge for

implementing speech technology is the human factor

(Ajami, 2016; Dawson et al., 2014; Parente, Kock,

and Sonsini, 2004). Doctors need to adapt their way

of working and this often leads to problems (Dawson,

et al., 2014). To avoid this, intensive support is

needed (S2; Ajami, 2016). An overview of all found

pros and cons in literature is represented in Table 1.

3 METHODS

For this study we performed a literature review and a

qualitative study. We searched PubMed, Springerlink

and Elsevier for finding the relevant articles. The

following key words and/or their combinations are

used: speech recognition, health care,

spraaktechnologie, spraakherkenning, zorg, medisch,

pros, advantages, cons, werking, neural network,

acoustic model, akoestisch model and Hidden Markov

Model. While selecting articles, we focused on the

publication date and Citation index.

The data for the qualitative study were gathered

by performing ten semi-structured interviews. We

used a structured topic list and an operational model

to establish the topics of the interviews and

corresponding questions. The participants consisted

of four managers working at two different hospitals,

four suppliers of speech technology working at

different companies, and two users of speech

technology with different professions. An overview

of the participants can be found in Table 2.

HEALTHINF 2018 - 11th International Conference on Health Informatics

342

Table 1: Pros and cons of speech technology according to other studies.

Table 2: Background overview and IDs of the participants.

Managers

(of departments)

NLP

Suppliers

Users/

Doctors

M1 Martini Ziekenhuis

M2 UMC Utrecht

M3 UMC Utrecht

M4 UMC Utrecht

S1 Cedere

S2 G2Speech

S3 G2Speech

S4 Nuance

U1 Orthopaedist

U2 Radiologist

Participants were approached when they had

experience with using speech technology, the

implementation of speech technology, or facilitated

speech technology. They were approached via

LinkedIn, or participants referred to other

interviewees with experience with speech

technology, who were approached next.

The respondents had to sign an informed consent

to give permission for using the information of the

conversation. With their permission the conversation

was recorded. Next, we transcribed the interviews.

After the transcription, we validated our trans-

criptions by sending it back to the participant for their

final approval. The document was then added to the

report. Next, we performed a content analysis for

finding the potentials and barriers as mentioned by

our participants. The analysis was done using Nvivo,

version 11.

4 RESULTS

4.1 Speech Technology in Practice

The orthopaedist who was interviewed as one of the

users of speech technology explained that document-

tation became a task for the doctor, due to implement-

tation of the EHR. Besides, documenting transferred

from speaking to typing according to him, confirming

what was explained in paragraph 2.1 (U1).

After completing their profile, speech technology

is functional. However, time is still needed to

optimize the profile by correcting mistakes, as speech

technology is a learning system (S3).

Only 1% of medical staff uses speech technology.

According to one of the suppliers (S4), this is

relatively low compared to other countries in the

Benelux. In the U.S., this technology is extensively

used (Shagoury, 2010). In the next paragraph we

explain our findings about the limitations of this

technology in the Netherlands.

4.2 Limited Use of Speech Technology

An overview of our results is represented in Table 3.

It shows which aspect is named by which group. N

states how many participants mentioned a particular

aspect. NU stands for the group of users, NS for the

group of suppliers and NM for the group of managers.

The last column contains the totals.

Speech Technology in Dutch Health Care: A Qualitative Study

343

Our results show that, in almost every hospital in

the Netherlands, pathology and radiology are using

speech technology, as explained by the participating

managers. However, these are only two departments

which use speech technology often. The remaining

departments do not use it, or it is occassionally used

by a few doctors from the department, as is the case

with the orthopaedist we interviewed. This is one of

the clarifications for the limited use of this technology

in the Netherlands. One of the suppliers stated that

radiologists and pathologists do not see any patients,

but only investigate the radiological photos or

bodyparts of the patient for diagnosis. Furthermore,

these specialisms document the most by far. This is

confirmed by a manager at the radiology department

of UMC Utrecht. One of the users stated that the EHR

is a barrier to use speech technology, since the EHR

is not adapted for the use of speech recognition.

Another reason of the limited use of speech

technology could be false expectations that people

have of speech technology, mentioned by two

suppliers. Since this technology was not very

functional at its introduction, many users kept this

opinion and therefore are not willing to use or obtain

it now. Users expect the technology to work

immediately, however, this is not the case in practice,

since usage of this technology demands a developed

profile. In addition, it was mentioned that accuracy

actually decreases after using speech technology for

a long period.

Besides the fact that the EHR is not adapted to

speech technology, it has smarttexts and

smartphrases. Managers tell that these are shortcuts

in the EHR. Doctors just have to type the abbreviation

of a commonly used word and the complete word

shows up on the screen. When doctors are used to

working with these shortcuts, speech technology is

less beneficial for them. The fields of the EHR do not

require much information. Polyclinic letters are

therefore set up easily. When the doctor has filled in

all information, he just has to click on the information

he wants to state in the letter and an automatic text

with previously filled in values shows up on the

screen. This auto-complete technology competes with

speech technology, which could be one of the reasons

for the limited use of speech technology.

To use speech technology, various adaptations

need to be done. First of all, doctors need to adapt

their speech. Users have to speak calmly and

articulate well. Besides, the work environment has to

be adapted. A place needs to be created where the

hardware and software is available, as well as a silent

environment.

The human factor is the most important problem

when implementing speech technology. Using speech

technology, physicians need to do secretarial work

(documenting) which evokes resistance, according to

one of the managers.

Hospitals have many projects that are legitimately

obliged to or that need to be done in terms of patient

safety. These projects often have priority over speech

technology. Furthermore, hospitals have limited staff

and financial capacity for speech technology, and

therefore are not able to support these projects. This

support is one of the most important aspects when

implementing speech recognition software.

The authorities in hospitals and the management

structures are slightly different in the Netherlands

compared to some other counteries. There is no

central authority that makes decisions like working

with speech technology or not. This is done

departmental based or even individually, making the

adoption of speech technology slow.

The extensive use of speech technology in the

U.S. can be clarified by the fact that it was initially

available for the English language only. English

speech technology works better because it exists

longer and had more time to mature. This is different

for the Dutch recognition rate, due to the relatively

small number of Dutch speekers in the world.

To sum up, there are plenty of factors influencing

the implementation of speech technology. They are

summarized in table 3.

4.3 Potential of Speech Technology

We asked participants their opinion about the

potential of speech technology as a tool for

documentation. An overview of the answers is

showed in table 4. Indecisive answers were excluded

from the table.

Participants who thought that speech technology

as a tool for documentation has high potential, do not

understand why the use is limited to 1%. One of the

users compared speech technology to automatically

driving cars. “It is already possible, but just a few

people bought it. It will become cheaper, easier and

people will get used to it, and next, adopt it. This is

the same for speech technology” (U1). One of the

people who did not expect a large potential worked at

the central ICT department of UMC Utrecht, and

experienced a booming period around 2005, but

stated that the hype of speech technology is over

nowadays. He explained that doctors who can work

with smarttexts and smartphrases in the EHR do not

benefit enough from speech technology.

HEALTHINF 2018 - 11th International Conference on Health Informatics

344

Table 3: Overview of the different aspects mentioned by participants for the limited use.

Table 4: Expected potential of speech technology as tool for

documentation consistent with participants.

Speech technology as

tool for documentation

has potential

Speech technology as

tool for documentation

has no potential

1 supplier

2 users

3 managers

1 manager

1 supplier

80% of all participants (all users, half of the

suppliers, and all managers) think high potential can

be expected in other applications than documentation,

such as structured reporting. When data is entered

fragmentedly, one can do analyses on these data. This

way, more information is obtained from the enormous

amount of data. This information can be used

internationally by using codes such as ICD10 and

ATC. Languages are not understandable by everyone,

but these codes are the same for every language, a

supplier explained.

In addition to structured reporting, speech

commands are mentioned. Computers can be

commanded by speech. This is useful when human

hands and eyes are busy (Ajami, 2016), for example

when operating.

Furthermore, decision support is mentioned as a

potential application field by various participants.

This works as follows: the computer suggests a

possible diagnosis based on the information entered

by the physician. An overview of the particpants’

answers on the pros and cons of speech technology

can be found in Table 5. Column N states how many

participants mentioned that particular aspect in

percentage of all participants. The major advantages

are the shorter RTT mentioned by 40% of the

participants, and the decrease in time needed for

administration, mentioned by 50% of the participants.

Finally, the major disadvantage is the financial

aspect, mentioned by 40% of the participants.

5 DISCUSSION

5.1 Conformity Literature and Results

In this study we investigated the reasons for the

limited use of speech technology in Dutch health care.

Our main findings to clarify the limited use were:

speech technology is only implemented in radiology

and pathology departments, doctors need to adapt

their way of working, no central authority for Dutch

hospitals, and finally the financial barrier.

Our main findings concerning the potentials and

barriers of speech technology were the decreased

RTT and the decreased time needed for administra-

Speech Technology in Dutch Health Care: A Qualitative Study

345

Table 5: The pros and cons according to participants in percentage of all participants.

tion as advantages, and the financial aspect as major

disadvantage. We compared the mentioned pros and

cons by participants with the pros and cons found in

literature, by comparing Table 1 and Table 5.

In our study, patient safety was not mentioned by

the participants as an advantage, but the factors that

lead to an increased patient safety were mentioned.

These factors are an increased quality of

documentation and a shorter RTT. Ajami (2016)

states that the duration of patients’ stay is reduced by

speech technology. This is not mentioned by our

participants. Moreover, Ajami (2016) states that

different accents are a disadvantage of speech

technology because the technology cannot cope with

this. In contrast, our participants explained that the

technology can handle different accents because of

the profile that needs to be made to get used to

different accents and a users’ vocabulary.

Furthermore, the study of Parente et al (2004)

found that users can speak to the computer like they

normally do to other people. This is contradicted with

our findings. Our findings show that users need to

speak slowly and articulate well. The recognition rate

was found to actually decrease after a while. This is

not found in literature, but can be explained by a

finding of Ajami (2016). Since speech technology is

a learning system, the system saves new words and

new pronunciations per word, as an individual does

not have the same pronunciation every time. The

vocabulary in the dictionary increases, and therefore

the system will confuse words more often since

dictionaries consisting of a lot of words tend to

confuse words with each other more often (Ajami,

2016). Our comparison showed that the remaining

aspects from Tables 1 and 5 correspond to each other.

According to the manager from the central ICT

department at UMC Utrecht, the hype of speech

technology happened around 2005. This statement is

in accordance with the predictions of the Gartner

hype cycle. This cycle characterizes a typical

progression of a new technology (Linden and Fenn,

2003). In 2014, speech technology was already placed

at the end of the cycle (Gartner, 2014). In 2015 and

2016 (the most recent one) the technology is not

included anymore in the models (Gartner, 2015-

2016). This suggests that the adoption of speech

technology is already over. However, one of our

interviewed users of speech technology refuted this

trend. He stated that people will get used to the

technology, and the technology will become easier

and cheaper. Speech technology will reach the

majority of the medical staff after this phase. This is

consistent with Rogers’s theory of innovations. First,

the most progressive 2,5% (the innovators) will adopt

HEALTHINF 2018 - 11th International Conference on Health Informatics

346

the innovation, and after this, the remaining four

groups will follow (Rogers, 1995). Accordingly,

speech technology is now only adopted by a part of

the innovators which would indicate that the adoption

of speech technology has yet to start.

The majority of the participants recognizes the

potential for speech technology as a tool for

documentation, but most potential is expected in

other applications of the technology. The study of

Parente, Kock and Sonsini (2004) expects a lot of

potential for speech technology as a tool for

documentation. The study of Johnson et al. (2014) is

more cautious. They state that speech technology can

have benefits, but there are many factors that need to

be taken into account, such as financial problems and

resistance of doctors (Johnson et al, 2014). The study

of Ajami (2016) is less positive and states that the use

of speech recognition is time-consuming, awkward

and not userfriendly. However, they state that the

technology will become reality in the end.

Nevertheless, the more recent the studies, the more

negative they tend to report on speech technology.

Finally, the human factor was mentioned by all

groups of participants. Moreover, this is mentioned in

many previous studies, and the study of Dawson

(2014) is fully committed to this factor. This

highlights the human factor as a major problem when

implementing speech technology. However, it is

possible that this is only the case in the beginning of

the implementation process, because of habituation

(Groves and Thompson, 1970).

5.2 Strengths and Limitations

To the best of our knowledge, this was the first study

to explore the advantages and disadvantages of

speech technology and to find the limitations of this

technology for Dutch health care. Another strength of

our study is the selection procedure of the

participants. All participants had relevant experience

in using, facilitating and/or implementing speech

technology, and all different types of stakeholders

were taken into account.

Nevertheless, our study had some limitations. Our

findings are based on only ten participants. We would

have wanted to increase the group of users, since we

interviewed only two and the remaining groups all

had four participants. Unfortunately, in the context of

this research, all users were doctors, and no nurses

were included. Because doctors are busy and hard to

reach, we did not succeed in expanding the group of

users.

5.3 Recommendations and Future

Studies

Our findings have implications for different parties.

Our results showed that managers, directors, ICT

department staff and other people need to know better

what to expect with the implementation of speech

technology to get familiar with all the different

factors that influence this process and its

implementations. This way a well-considered

decision can be made. Besides, users should know

better what to expect of the technology to be better

prepared to possible problems or obstacles. Also,

when the decision to implement speech technology is

made, we recommend intensive user support. This

was rated as very important according to our findings.

More research is needed, preferably with a more

extensive study design, to further confirm our

findings. For future research, new target groups

should be investigated, such as nurses, who document

on average 8.7 hours per week (Nuance, 2015).

According to Bosch (2005), speech technology could

be very useful for the General Practice (GP) as well.

It should be investigated how feasible implementing

speech technology is for the GP. Furthermore, the

other application potentials of speech technology as

named in paragraph 4.3 could be investigated. Future

studies are needed to investigate structured reporting,

speech commands and decision support in practice.

Finally, we propose to design and evaluate such

analytical applications of speech technology to

improve the daily practices of domain experts from

an Applied Data Science context (Spruit & Jagesar,

2016).

6 CONCLUSIONS

We discovered various barriers influencing the

adoption of speech technology. However, the

majority of participants in our study still thought there

is high potential for this technology. They

acknowledged that other applications of this

technology may be more beneficial than

documentation. Our results showed that speech

technology is useful as tool for documentation at the

radiology and pathology departments, but is less

useful as tool for documentation at other departments.

For those other departments, higher potential of

speech technology is expected in other applications

such structured reporting, speech commands and

decision support.

Speech Technology in Dutch Health Care: A Qualitative Study

347

REFERENCES

Advanced Voice Technology. (sd). Is de transcriptie van

een interview of vergadering middels spraakherkenning

mogelijk? Retrieved from Advanced Voice

Technology: https://spraakherkenning.nl/over-spraak

herkenning/interviews-en-vergarderingen-uitwerken/

Ajami, S. (2016). Use of speech-to-text technology for

documentation by healthcare providers. The National

Medical Journal of India, 148-152.

Bosch, B. (2005). Brian Bosch over spraakherkenning op

de computer: Voor de huisarts met twee vingers ...

Huisarts en Wetenschap, 891-892.

Dawson, L., Johnson, M., Suominen, H., Basilakis, J.,

Sanchez, P., Estival, D., Hanlen, L. (2014). A usability

framework for speech recognition technologies in

clinical handover: A pre-implementation study. Journal

of Medical Systems.

Gartner (2014, 2015, 2016). Gartner Hype Cycle.

Geitgey, A. (2016, December 23). Machine Learning is Fun

Part 6: How to do Speech Recognition with Deep

Learning. Retrieved May 2, 2017, from Medium.

Groves, P. M., and Thompson, R. F. (1970). Habituation: A

dual-process theory. Psychological Review, 419-450.

Johnson, M., Lapkin, S., Long, V., Sanchez, P., Suominen,

H., Basilakis, J., and Dawson, L. (2014). A systematic

review of speech recognition technology in healthcare.

BMC Medical Informatics and Decision Making, 14.

Koivikko, M. P., Kauppinen, T., and Ahovuo, J. (2008).

Improvement of Report Workflow and Productivity

Using Speech Recognition—A Follow-up Study.

Journal of Digital Imaging, 378-382.

Linden, A., and Fenn, J. (2003, May 30). Understanding

Gartner's Hype Cycles.

NOS. (2017, April 3). Koks, onderwijzers en artsen ervaren

hoogste werkdruk. Retrieved June 3, 2017, from NOS:

http://nos.nl/artikel/2166294-koks-onderwijzers-en-

artsen-ervaren-hoogste-werkdruk.html

Nuance. (2008). Dragon NaturallySpeaking 10 Preferred. :

Nuance Communications.

Nuance. (2015, June 24). Nuance Healthcare Partner event

2015. Retrieved April 12, 2017, from Cedere:

https://www.cedere.nl/2015/06/24/nuance-healthcare-

partner-event-2015/

Parente, R., Kock, N., and Sonsini, J. (2004). An Analysis

of the Implementation and Impact of Speech-

Recognition Technology in the Healtchare Sector.

Perspectives in Health Information Management.

Renckens, E. (2009, May 7). Het luisterend oor van de

computer. Retrieved April 28, 2017, from NEMO

Kennislink: https://www.nemokennislink.nl/publica

ties/het-luisterend-oor-van-de-computer

Rogers, E. M. (1995). Diffusion of Innovations. New York:

The Free Press.

Schumacher, J. (2017, May 22). Cijfers: vergrijzing en

toenemende zorg. Retrieved June 8, 2017, from

Zorgvoorbeter: http://www.zorgvoorbeter.nl/

ouderenzorg/hervorming-zorg-cijfers-vergrijzing.html

Shagoury, J. (2010). Chapter 11 Dr. “Multi-Task”: Using

Speech to Build. In A. Neustein, Advances in Speech

Recognition (pp. 247-273). New York: Springer US.

Spruit, M., & Jagesar, R. (2016). Power to the People!

Meta-algorithmic modelling in applied data science. In

Fred,A. et al. (Ed.), Proceedings of the 8th International

Joint Conference on Knowledge Discovery,

Knowledge Engineering and Knowledge Management

(pp. 400–406). KDIR 2016, November 11-13, 2016,

Porto, Portugal: ScitePress.

Tuin, C. (2016, October 20). Microsoft claimt doorbraak

spraakherkenning. Retrieved from Computable:

https://www.computable.be/artikel/nieuws/developme

nt/5860772/5440850/microsoft-claimt-doorbraak-

spraakherkenning.html

Vervoort, D. (2017, April 21). Zo werkt spraaktechnologie.

Retrieved April 26, 2017, from Clickx:

http://www.clickx.be/achtergrond/169324/zo-werkt-

spraaktechnologie/

HEALTHINF 2018 - 11th International Conference on Health Informatics

348