Speech Recognition: what’s up in Europe?

Why don’t we leave planet North America for a minute, cross the pond and take a look at what is going on in Europe?

The crystal ball predicts a “steady growth” for the European healthcare voice recognition market, with dramatic changes in the healthcare delivery system putting enormous pressures on healthcare providers to reduce costs and improve the quality of care given.” According to Frost & Sullivan, “the combined benefits of voice recognition and healthcare information systems, such as EMR/PACS/HIS, extending the productive range of healthcare organisations and individuals alike, particularly in today’s demanding healthcare environment.”

As a matter of fact, the market is definitely showing signs of major activity, not only at the local but at the regional healthcare level too. Only a week ago, Philips announced the roll-out of the first region-wide speech recognition system in the Italian healthcare sector, connecting no less than 22 hospitals. But this is only the fifth region-wide success for SpeechMagic within the last twelve months, following similar projects in France, Norway and Spain. All right, who’s next?

Clearly, the expectations are pretty much the same from one side of the ocean to the other, as Frost and Sullivan puts it: to allow healthcare organisations to meet patient quality of care expectations through accurate, accessible, shareable and real time health information.

What will be interesting to monitor over the next few months are the implementation methods within each country as well as the level of integration achieved. Especially since, in 2006, the European Commission, DG Health and Consumer Protection, observed that the health sector still lagged behind other industries in introducing systematic safety processes and recommended the introduction of the electronic medical record (EMR). An observation that led Philips and Health Language Inc. to collaborate on a prototype designed to advance system interoperability in healthcare with the help of speech recognition.

Stories / testimonials on recent European speech recognition implementations can be found on the European Hospital web site.

The excerpt from the Frost & Sullivan research service on European Healthcare Voice Recognition Systems Markets can be downloaded from the Philips Speech Recognition News Center.

Speech Recognition Expert Welcomes This Blog

No big industry news nor philosophizing on the differences between front-end and back-end speech recognition today… just a smily thread to start the week on a bright note. I am indeed thrilled that Nick van Terheyden, MD, Chief Medical Officer for Philips Speech Recognition Systems and a world-renowned expert in healthcare speech recognition, is supporting my blog intitiative. Van Terheyden said this morning, in a joint press release: “A technology that started as an administrative tool has become a critical element of care as it provides physicians with accurate patient data, enabling them to manage care effectively and draw accurate conclusions to administer the appropriate treatment. We therefore welcome this blog initiative as a way of highlighting the tremendous benefits of speech recognition to hospitals, clinicians, referring physicians and patients.”

Missing Clinical Information During Primary Care Visits

A 2005 report by the Journal of the American Medical Association (293:565-571) describes primary care clinicians’ reports of missing clinical information and their perception of patient harm resulting from these events. This survey only confirms the importance of properly capturing, tagging and centralizing medical data collected at the point of care. If you think about it, probably as much as 95% of all medical records are handled through the dictation-transcription process. As the survey outlines, it is not so much about the mere presence of an EMR, but more about the clinician’s involvement / familiarity with the EMR, which, once again, isn’t worth much without the dictated data that feeds it.

Here is an overview of the survey’s findings:

“Effectively managing clinical information (patient information such as demographics, medical history, medications, test results, and family structure) is an essential part of all medical care; it is particularly crucial for primary care to be able to fulfill what the Institute of Medicine and others consider to be its defining task of coordinating comprehensive care across the health care system. Unfortunately, multiple barriers complicate the collecting, synthesizing, recording, and sharing of clinical information, including privacy regulations, decentralized medical systems, inadequate interprofessional communication, the transfer of patients’ care within and across care settings, and the rapid turnover of patients’ insurance plans. Accordingly, physicians may not have clinical information available when it is important for a patient’s care.

Missing clinical information has been implicated in injurious adverse events. Elder et al reported that missing clinical information was associated with 15.6% of all reported errors in primary care, most of which were perceived by clinicians as likely to be harmful, and was implicated in every major category of medical error. In the only research studying missing clinical information directly, Canadian emergency department physicians reported that 15.3% of visits had important information missing at the time of the encounter that was very likely to result in patient harm. Such harm could include otherwise avoidable drug interactions or duplications, missed or delayed diagnoses, missed immunizations, unnecessary testing and procedures, and the downstream effects of such events.

…In nearly 1 in 7 visits, (clinicians) reported that clinical information important for the patient’s care was missing. Although laboratory reports and dictations or letters predominated, clinicians reported that the missing information originated from a variety of sources and often included more than 1 type. In 44% of the visits with missing information, clinicians believed the patient would be at least somewhat likely to be adversely affected. If validated by future research, these results could have serious implications for the 220 million primary care visits that occur in the United States each year.

…Clinicians believed that missing information would likely result in either delayed care or at least 1 duplicative medical service in 59.5% of visits with missing information…

…Clinicians reporting a full EMR in their practice were significantly less likely to report missing clinical information, but this did not eliminate the problem. Missing information was believed more likely to be outside the clinical system than within it and therefore may be beyond the reach of an EMR. The lack of impact of partial EMRs and electronic access to hospital data on adverse events has been found in other settings. We found no difference in reports of missing information when we used the concurrence among clinicians within a practice to determine the EMR variable. This difference from individual report may indicate that familiarity with or actual use of an EMR is a better predictor of effective information management than the mere presence of an EMR.”

Full report is available from the JAMA web site.

See also this thread: Speech Recognition Prototype for EMR Interoperability

Speech Recognition on Citrix

The benefits of deploying a document creation and management platform through a thin-client infrastructure is probably no news to you: no more cost-intensive, powerful workstations, substantial cost savings, increased data security, on-the-move data access for users, just to name a few.

With a growing demand from large healthcare providers for thin-client infrastructures, Citrix is forming new, strategic relationships with key players in the field of the EMR and medical records management technologies. The company has therefore launched the Citrix Ready initiative to help its customers identify recommended third party solutions that are trusted to sensibly enhance the Citrix application delivery infrastructure. The Citrix Ready status was recently awarded to the Philips speech recognition technology, SpeechMagic, after successfully completing verification testing that proved joint system compatibility.

Robert Thornton, commercial director for Philips Speech Recognition Systems, said: “The adoption of SpeechMagic in the healthcare sector has reached a new dimension, with break-through implementations on a city- and region-wide scale’. Many of these projects require the deployment of industrial grade speech recognition within a Citrix based IT infrastructure. Philips is the first to advance its speech recognition technology to fully support Citrix infrastructure as part of the Citrix Ready initiative, thus enabling our integration partners to quickly deliver on changing market needs.”

David Jones, corporate vice president, business development, for Citrix commented: “Philips has invested significant resources in adapting SpeechMagic to Citrix requirements and we are looking forward to furthering the penetration of speech recognition, digital dictation and transcription with the Citrix community.”

The Citrix Ready initiative currently supports Citrix Presentation ServerT 4 (32 and 64-bit versions), Citrix Password ManagerT 4.5 and Citrix NetScaler. Support for additional Citrix products will be added throughout 2007, says the company.

Digital Dictation or Speech Recognition?

Let’s first take a look at the terminology. As always, Wikipedia clears up any potential confusion with one of those efficient, 3-line definitions: “Digital dictation is different from Speech Recognition where audio is analyzed by a computer using speech algorithms in an attempt to automatically transcribe the document. With digital dictation the process of converting digital audio to text is done via a typist using a digital transcription software application (…)”

But this doesn’t tell us which one should be preferred to the other (Wikipedia is not that powerful…yet). The truth is, both technologies work closely together when implemented in a healthcare environment, mainly because a speech recognition engine is not worth much without the workflow automation features brought in by the digital dictation system (DDS) it typically integrates within. In a white paper dedicated to speech recognition technology for healthcare, expert Dr. Bob Yacovitch explains how the DDS is the glue that holds everything else:

The first aspect is workflow automation. “A stand-alone speech recognition solution on an individual PC does not bring the expected gains in productivity and efficiency. Speech recognition needs to be approached as part of a whole document creation platform. Real benefits only come by implementing a digital dictation workflow solution with integrated speech recognition, which takes into account the entire document creation process and not simply the transcription of a dictation. The digital dictation workflow system is the central framework that supports everything else, from voice control to workflow management and it is what the physician will be interacting with on a day-to-day basis. The difference resides in the system’s new ability to produce a “recognized text” together with the voice file. This draft report simply needs to be corrected as opposed to being fully transcribed.”

The DDS thereby seems to be the most important ingredient in the mix; giant steps can already be achieved with it, provided high-level routing management is offered. Speech recognition can turn document creation from “fast” into “light speed,” though it is not necessarily justified for all environments. Factors such as workflow complexity and the number of dictating authors play a key role in the overall ROI (return on investment), hence the need to investigate what can be achieved in terms of workflow management with a single DDS before even considering the speech recognition path.

The other keyword is integration. It is the DDS that integrates with the rest of the organization’s IT infrastructure, not the spech recognition engine, and “optimal accuracy and reliability of medical data can only be achieved in a fully integrated IT environment,” insists Yacovitch.

Download the Speech Recognition for Healthcare White Paper

Another large speech recognition installation underway

Canada-based Hôpital Enfant-Jésus and Hôpital Saint-Sacrement, merged under the name Centre hospitalier affilié universitaire de Québec (CHA), are currently rolling out a central digital dictation, speech recognition and transcription system to be fully operational in September 2007.

Over 325 physicians from 7 departments (Radiology, Pathology, Nuclear Medicine, Respiratory Therapy, Psychiatry and Hematology) will be equipped with phone dictation (28 accesses are planned) and/or PC dictation (280 licenses in total) applications.

The speech recognition module will be activated for four of the seven departments (Radiology, Respiratory Therapy, Psychiatry and Nuclear Medicine). In back-end mode, the process will be fully transparent to physicians as their dictations will automatically be processed by the speech recognition engine before they are sent to a transcriptionist for correction. The front-end mode will also be available to allow physicians to play a more active role in the report creation process, by allowing them to correct their reports, onscreen and in real-time, as they are being dictated.

The speech recognition engine will provide a French Canadian medical speech recognition ConText (specialized recognition vocabulary), which was developed exclusively for French Canadian users.

Fully integrated with the hospitals’ existing computer system, the new system will simplify the documentation workflow and facilitate the access to medical reports for all relevant medical personnel. Bidirectional HL7 interfaces will indeed be implemented between the dictation/transcription platform and existing ADT, RIS and Pathology systems, to ensure that patient information can always be accessed by and shared among the appropriate medical personnel.

CIOs to invest more in technology as a way to improve physician access to information and reduce medical errors

In 2005, the Journal of the Institute of Medicine attributed 98,000 patient deaths every year in the US due to medical errors, many of them – such as poor documentation – preventable. The good news is that healthcare Chief Information Officers (CIOs) seem more than ever committed to address this critical issue. The latest Chief Information Officers (CIO) Survey conducted by Health Data Management and The Quammen Group indeed shows that an overall growth in IT budgets is underway, with respondents clearly focused on improving access to information for clinicians and reducing medical errors. Of those respondents who expect their I.T. budgets to increase in fiscal year 2007, 51% said the primary factor for those budget increases is to improve clinician access to information. Another 27% cited reducing medical errors/improving quality as the primary factor for spending hikes:

When it comes to investing in emerging technologies, speech recognition comes first:

Full Survey results can be found on the Health Data Management web site.

Vendor Comparison Chart

Here is a rather helpful online tool provided by Outpatient Care Technology Magazine. Updated every year, the Comparison Charts section of their web site allows you to put different speech recognition systems side-by-side in order to compare the various features offered, such as:

• Speech engine used
• Remote access support
• Specialty vocabulary
• Voice commands
• Batch processing
• Macros/templates
• Memory
• Number of ports/users
• Redundancy
• Interfaces
• Sampling rate
• Input type
• Etc…

You can compare up to 5 vendors using the online tool or order a print copy that lists all of them.

Speech Recognition Prototype for EMR Interoperability

Physicians know it best: entering and processing EMR (Electronic Medical Record) data is still very much time-consuming and unreliable due a lack of integration between systems and a large volume of manual entries involved. Royal Philips Electronics and Health Language have launched a prototype called interop 6.1 designed to advance system interoperability in healthcare with the help of speech recognition. The idea is to automate the conversion of free text into consistent medical terminology to enable direct storage in the (EMR), thereby providing physicians with a central, reliable source of medical information.

Interop 6.1 extracts findings, diagnoses, drugs, allergies and other relevant information from dictated reports. Subsequently, the information is formatted for upload to the EMR. The system generates codes that support the classification of medical conditions and the structuring of clinical data, such as the World Health Organization’s International Classification of Diseases (ICD) codes and the Systematized Nomenclature of Medicine (SNOMED) codes.

Interop 6.1 also identifies medical terms from the history of present illnesses (HPI) to be indexed n the EMR. The indexed terms can be used for statistical evaluation, thus helping to improve hospital procedures related to patient safety, medication and continuity of care.

“Healthcare communities need to standardize existing systems and augment them with new integrated solutions focused on the patient” said Marcel Wassink, managing director for Philips speech Recognition Systems. “The ability to exchange data accurately, effectively and consistently is key to eliminating cross-organizational boundaries and ensure the uniform and secure availability of critical information. This will eliminate documentation errors and improve the quality of care.”

A prototype system which extracts information from freely-dictated text and allocates it to the corresponding fields in the EMR is available.

Healthcare Documentation: A Report on Information Capture and Report Generation

This report by the Medical Records Institute addresses the fundamental question of what information capture method(s) should be used to improve healthcare documentation, and provides a number of recommendations to the healthcare community. Extracts:

“Accurate, accessible, and shareable health information is a well-accepted prerequisite of good health care. Yet, the healthcare system in the United States continues to accept illegible handwriting and other documentation practices that diminish the quality of healthcare documentation through reduced accuracy, accessibility, and shareability…

…This report focuses on six documentation methods, i.e., handwriting, speech, direct computer input, document imaging, device capture, and clinical imaging, as well as their hybrids. The evolution from handwriting to electronic healthcare documentation is concurrent with a transition from free (i.e., unstructured and typically unsearchable) text to structured and interactive text. To better understand this transition, this Report addresses the characteristics of unstructured, structured, and interactive data capture styles.”

> Read full report abstracts