Speech Recognition Trends In Radiology

Research and Report by

2

Healthcare Providers: Receive all peer60 exclusive reports when anyone in your organisation participates in our research (care givers, tech & support personnel,

and leadership).

Reproduction or distribution of any part of this report without authorisation is

prohibited without consent from peer60.

Contact Chris Jensen at cjensen@peer60.com for report content distribution conditions and guidelines.

AN INVITATION TO HEALTHCARE PROVIDERS

Find Out More Here

2 3

Premium Content Not Included In This Report

• Net Promoter Scores by supplier• Net Promoter Scores trended year over year• Net Promoter Scores by organization size• Supplier mindshare among first-time adopters• Supplier mindshare among replacement business• Suppliers most at risk of losing contracts• Top ways to save at risk contracts• Net growth/attrition by supplier

4

Suppliers Mentioned in this Report

4 5

Table of ContentsExecutive Summary

Our Methodology

Demographics

Preferred Method for Creating Patient Reports

Installs Among Participants

First Time Adopters

Net Promoter Scores

Replacement Market

How to Save At Risk Contracts

Potential Net Growth in Speech Recognition

Conclusion

Appendix I - Net Promoter Score Methodology

6

7

8

9

12

14

15

16

18

21

22

23

6

Executive SummaryWhat happens when a market matures around technology that fills a very real need? High adoption becomes the norm, industry leaders emerge, and inevitably some begin to dull as the shine of the latest and greatest technology wears off. In addition, there is always a small segment of the market that stubbornly holds out like a grandpa with a flip phone who makes a point to let everyone know how smartphones are overkill.

This is the speech recognition market in radiology. Other healthcare segments, such as cardiology, pathology, and speech-enabled EHR have not seen anything close to the adoption rate in radiology. This in large part is due to the fact that radiology is big money and can afford to adopt new technology at a much faster pace than other segments. In addition, radiologists have always been very mindful of efficiency. Radiology is the early adopter of technology of healthcare. Need proof? The majority (73%) of holdouts in adopting speech recognition have plans to finally take the plunge. That’s as significant as your great grandma showing up to the next family reunion with the newest phablet, enthusiastically showing everyone how it gets her on the “interwebs.”

We’re seeing an interesting story develop as we trend data from 2015’s study on the speech recognition market. While three main suppliers are performing well in the market, one is seeing significant improvements in several key performance metrics — Net Promoter Scores, mindshare, and at risk contracts.

At the end of the day, there are few audiences that technology companies love more than early adopters. But once the dust settles from the mad rush to innovate and adopt, growth opportunities scale back dramatically. Speech recognition epitomizes this trend with a 9% replacement market and Net Promoter Scores that are relatively high. This low degree of churn makes it difficult (but not impossible) for start-ups to make their way in this market while keeping the undisputed market leaders on top. The only way these kinds of markets see major shakeups is through unique new innovations or material cost reductions.

**Note: This research included participation from 375 healthcare leaders in radiology departments and diagnostic imaging centers across North America.

6 7

Our MethodologyOur data collection method is very different than market research firms...good thing we’re not a research firm. The

standard operating procedure in healthcare market research is to grind on care organization leaders in order to achieve

a minimum number of responses per supplier covered in an upcoming report...or on the opposite end of the spectrum

by offering monetary incentives to provide feedback. We’ve seen how these methodologies introduce some inherent

biases in the data collection process that is motivated primarily by a desire to sell more reports to a wider audience.

Alternatively, we believe participation in research should be quick, simple, and non-intrusive. In addition, we prefer

reaching a wide audience of all stakeholders in care organizations regardless of who they currently have as a supplier of

a particular solution, service, or piece of equipment. Simply put, we reach out to all relevant audiences in our network

of more than 500,000 healthcare providers on a given research subject and allow those who want to share their

perspective to weigh in.

If you have questions about our methodology and motivations behind making our research public, contact Chris

Jensen at cjensen@peer60.com.

8

Demographics

30%

28%

18%

13%

6%5%

Respondents by Title

PACS Admin

Radiology Director

Radiology Manager

Radiologist

Radiology Chief

Other

Respondents by Study Volume

16%

12%

20%27%

24%0 - 25,000

25,001 - 50,000

50,001 - 100,000

100,001 - 250,000

250,001 +

One of the challenges in conducting meaningful research is utilizing appropriate audiences with a stake in the study topic. Many thanks to our 375 participants for their feedback regarding use of speech recognition solutions in radiology. We were fortunate to receive participation from all applicable leadership titles and sizes of organizations that have a stake in the continued success of speech in radiology.

8 9

In what is already a mature, saturated market, we continue to see speech adoption rates in radiology continue to rise by modest amounts. However, even a modest increase in adoption in a market with roughly 90% coverage is worth mentioning. In a market like this each new win is sweet indeed.

Preferred Method for Creating Patient Reports

Respondents by Title

81%

12%

6%

2%

We use a speech recognition system

We dictate a report and send it to transcription

Combination of both

Other

10

Preferred Method for Creating Patient Reports Cont.

Creating Patient Reports (by Study Volume)

63%

80% 80%

93%96%

24%

14% 13%

4% 3%0% 3%7%

4% 1%

12%

3% 0% 0% 0%

0 - 25,000 25,001 - 50,000 50,001 - 100,000 100,001 - 250,000 250,001+

We use a speech recognition system

We dictate a report and send it to transcription

Combination of both

Other

While not a huge revelation, it’s interesting to see the trend of speech usage by organization size. The larger the organization the higher the adoption, and by default a lower use of transcription services. Clearly the growth opportunities are among small to medium size care organizations. The takeaway? Suppliers that want the biggest, richest contracts are going to have to win that business in the open warfare of the replacement market. That makes the prospects a little rough, to say the least, when the undisputed market leader happens to be Nuance, a darling among all health IT companies. This market perspective of Nuance was confirmed in our previous report titled “Trends in Medical Imaging Technology”, in which providers were asked to rate their favorite imaging IT supplier. Nuance was the undisputed favorite among all imaging IT suppliers. For more on that report, download it here.

10 11

Preferred Method for Creating Patient Reports Cont.

Creating Patient Reports (Year over Year)

86%

9%

3% 2%

81%

12%6%

2%

We use a speech recognition system

We dictate a report and send it to transcription

Combination of both Other

2015

2016

Comparing adoption data from last year, speech continues to make inroads among the last remaining holdouts. Or is speech simply filling the void when a care organization decides to finally let go of transcribing reports? Whatever the reason, this trend is a very different story than what our reports show for other healthcare segments such as cardiology, pathology, and speech-enabled EHR. Look for more research from us around these other areas of speech recognition trends and adoption in coming months.

12

Installs Among Participants

Speech Recognition Installs Among Participants (Year over Year)

Speech Recognition Installs Among Participants85%

8%3% 2%

Nuance (248) M*Modal (22) Dolbey (8) Agfa (7)

2015

2016

85%

8%3% 2%

86%

8%4%

0%

Nuance (248, 153) M*Modal (22, 14) Dolbey (8, 7) Agfa (7, 0)

Let’s address the whale in the aquarium. Speech recognition in healthcare is Nuance’s world and we’re all just living in it— at least for now. When the next closest competitor has less than 10% of Nuance’s total responses, there’s no other way to skin it. Nuance is the proverbial 800-pound gorilla.

When we look at the same data point year over year (2015 to 2016), the theme is consistent.

12 13

Installs Among Participants Cont.

83%

75%

84% 82%

94%

4%7%

11%7% 4%

9%

0%5% 5%

0%0%

7%

0%5%

0%

0 - 25,000 25,001 - 50,000 50,001 - 100,000 100,001 - 250,000 250,001+

Nuance

M*Modal

Dolbey

Agfa

Speech Recognition Installs (by Study Volume)

14

First Time Adopters

Expected Adoption Among First Time Adopters

73%

27%

Yes

No

Expected Adoption Among First Time Adopters (by Study Volume)

70%67%

83%

100% 100%30% 33% 17%

0 - 25,000 25,001 - 50,000 50,001 - 100,000 100,001 - 250,000 250,001+

Yes

No

This data point tells a different story than we’re used to seeing among potential first time adopters in most other segments. Typically when a new technology has seen widespread adoption the few remaining holdouts dig their heels in and almost make it a mission to resist joining the crowd. In contrast, the minority in radiology who have not adopted are planning to do so to the tune of three out of four. Those least likely to adopt are smaller organizations that in many cases are in a state of transition due to ownership changes, retiring physicians, etc.

14 15

Net Promoter Scores**See appendix 1 for an explanation of Net Promoter Score methodology.

The Net Promoter Score (NPS) data for this report puts us in a bit of an awkward position. Based on market share and providers’ willingness to discuss their speech suppliers, there’s Nuance... and then there’s everyone else. M*Modal is Nuance’s only true competition in radiology. However, with an n of 22 it’s kind of stuck in the middle. It doesn’t make sense to line it up next to Agfa or Dolbey given their even smaller pieces of the pie, but we do admit that it looks strange to compare NPS data of one supplier with 244 (Nuance) responses and another with 22 (M*Modal).

Net Promoter Scores (Group 1)

41

100

0

36

-100

Net Promoter Scores (Group 2)

38

-14

100

- 100

0

*Suppliers in this graph (Not in respective order): M*Modal, Nuance

*Suppliers in this graph (Not in respective order): Agfa, Dolbey

16

Replacement Market

Replacement Market

9%

91%

Considering switching

Not switching

As referenced earlier in the report, Net Promoter Scores that are generally high and on the rise provide no real motivation toward change. This trend will likely continue until speech technology takes a significant leap forward by one of the primary suppliers in the space.

16 17

11% 9% 10% 12%

100% 89% 91% 90% 88%

0 - 25,000 25,001 - 50,000 50,001 - 100,000 100,001 - 250,000 250,001+

Replacement Market

Considering switching

Not switching

Replacement Market (Year Over Year)

9%

91%

14%

86%

Considering switching Not switching

Cue the crickets. What was already a quiet replacement market last year is now deafening. The replacements that are happening are primarily among medium to large radiology departments and imaging centers.

Replacement Market Cont.

2015

2016

18

How to Save At Risk Contracts

How Suppliers Can Save At Risk Installs

23%

20%

17%

7% 7% 7%

Reduce Cost

20%

Expand Functionality

Nothing, Relationship

Can't Be Fixed

Improve Accuracy

Improve Customer

Service

Make Easier to Enroll

Radiologists

Other*

Cost (or the reduction in cost) is king when attempting to save an at risk speech recognition contract. The question becomes how much love...or slush...or wiggle room...or whatever you want to call it exists in these contracts. A reduction in cost will go a long way to settle shaky customers. Beyond this, expanding functionality is also high on the wish list. If a supplier has no way to resolve these one-two contract-killing punch combos, it’s out of luck on saving at-risk customers.

*Other: Improve work flow, loyalty incentives, upgrades, continuous training

18 19

How Suppliers Can Save At Risk Installs (Year Over Year)

23%

20%

17%

7% 7% 7%

20%20%

28%

8%

20%

16%

2%

6%

Reduce Cost Expand Functionality

Nothing, Relationship

Can't Be Fixed

Improve Accuracy

Improve Customer

Service

Make Easier to Enroll

Radiologists

Other*

Last year, expanding functionality was the top way to save at risk contracts. Unfortunately, the amount of R&D required to see measurable improvements in this area won’t meet most unhappy customers’ timelines for improvements before the relationship flatlines. Aside from this, reduction in costs and improvements to accuracy were also top ways to save clients with growing disinterest and wandering eyes. Yet again, reducing cost can be a short-term solution, but improving accuracy is an ever present mission of any reputable speech recognition company.

How to Save “At Risk” Contracts Cont.

* Other 2016: Improve workflow, loyalty incentives, upgrades, continuous training Other 2015: Improve workflow, automatic software upgrades, continuous training

2015

2016

20

How Suppliers Can Save At Risk Installs (by Study Volume)

0% 0%

33%29%

57%

0% 0%

67%

14%

29%

0% 0% 0%

29%

14%

0% 0% 0% 0%

29%

0%

33% 33%

0% 0%0% 0% 0% 0%

29%

0%

67%

0%

29% 29%

0 - 25,000 25,001 - 50,000 50,001 - 100,000 100,001 - 250,000 250,001+

Reduce Cost

Expand Functionality

Nothing, Relationship Can’t Be Fixed

Improve Accuracy

Improve Customer Serviice

Make Easier to Enroll Radiologists

Other*

How to Save “At Risk” Contracts Cont.

* Other: Improve work flow, loyalty incentives, upgrades, continuous training

20 21

Potential Net Growth in Speech Recognition

64%

38%

6%

-29%

Potential Net Growth

In order to take a closer look at net growth (or net attrition), we examined the relationship between installs among participants, number of times participants are being considered for new or replacement contracts, and number of at risk contracts. We broke this data set out by supplier. In theory, this gives us a reading of the growth potential each supplier might look forward to over the next couple of years. Since this data includes responses where multiple suppliers are vying for the same contract of a new or replacement install, we look at these statistics as best case scenario growth potential. Despite this, the value in this information is the relationship between each supplier—Who’s poised for significant growth, stalling out, or seeing contraction in the market relative to the number of current installs among participants.

Potential Net Growth (Year over Year)

64%

38%

6%

21%

43%

9%

2015

2016

*Suppliers in this graph (Not in respective order): Agfa, Dolbey, M*Modal, Nuance

*Suppliers in this graph (Not in respective order): Dolbey, M*Modal, Nuance

22

ConclusionSpeech recognition in healthcare is a well-contained market with few competitors. Healthcare providers are beginning to accept that while current accuracy levels will continue to improve, there is no reason to believe that monumental improvements are on the horizon. So where do they expect these solutions to enhance their value propositions? Cost. This is the trend of technology adoption regardless of industry. When there are no leaps forward in other benefits, the expectation is that the price tag will steadily decline.

One storyline this year in speech is not significant growth or a high churn replacement market, but rather the emergence of a well-known supplier that could shake up the market. When key metrics are all pointing in the right direction, the future looks awfully bright.

Finally, a very telling piece of data in this report is the disparity between radiology providers with the M.D. after their names and those without. The non-physician radiology leaders are much more inclined to give positive feedback about their speech supplier than are radiologists including chiefs of radiology. Given the choice of whose good graces a supplier wants to be under, the physician leaders appear to be a more coveted group. Unfortunately, all the primary speech recognition suppliers are viewed on a range of ever so slightly positive to highly negative. Finding ways to please radiologists should be the top priority among these companies.

22 23

Appendix I - Net Promoter Score MethodologyThe suppliers outlined in this report are rated based on the Net Promoter Score (NPS) methodology, which measures customer satisfaction by asking one simple question: “How likely are you to recommend your supplier to a peer or colleague?” and yields results that major academic research suggests is the most accurate measure of overall satisfaction. Participants are given a scale of 0-10 to define their overall willingness to refer a supplier to a colleague. Those who choose between 0-6 are tagged as “detractors”. 7-8 respondents are “passives” and those who select a 9-10 are the company’s “promoters”.

A Net Promoter Score is displayed as a single number, ranging from 100 (a perfect score) to -100 (say goodbye to the business), by subtracting the percent of detractors (those most likely to speak out negatively about the business) from the percent of promoters (those who sing praises). Passives are tossed out of the equation as they contribute no detriment or benefit to the overall equation. This simple formula looks like this:

% Promoters - % Detractors = Net Promoter Score (NPS)

Scores of 30 and above are considered good to great. Score ranging from 0-29 should be a cause for alarm. Negative scores are often a sign of a sinking ship— and we all know what happens to captains in those cases. The exception to this is when all suppliers in a space have scores that range from anemic to worse.