Benefits of Speech Recognition

Document Sample
Benefits of Speech Recognition Powered By Docstoc
					                                                                                              Benefits of Speech Recognition
                                                                                      John Dorsky, Mellar Davis, Eduardo Oliveira, Lewis Rubin
                                                                                                       Leading in Healthcare
                 Speech Recognition Systems                           Speech Recognition Systems: Real Time                 Transcription Services Savings after EMR Implementation                                                   Rationale for Speech Recognition Systems
                                                                                                                                                CCF Weston, FL                                                                   •    Hard copy record is gradually transforming into an
    Fusion Text – 2 transcription options                       Ancillary Features
                                                                                                                                                                                                                                      electronic record
     • Digital dictation (sound)                                • Auto text design for routine report                       • Number Physicians : 100                                                                            •    Transcription needs will increase
     • Simultaneous speech recognition                          • Intelligent speech interpretation technology eliminates   • Year of EMR implementation: 2006                                                                   •    Inefficiencies will increase since the “typed” word is
                                                                  background noise, redundancy                              • Comparison of Transcription costs between 2005 and 2006                                                 slower than the “spoken” word
                                                                Advantages                                                                                                                                                       •    Overhead costs will increase with diminished revenues
                                                                                                                                        Transcription Costs Before EMR                                                           •    Patient care may become less efficient
           Speech Recognition Fusion Systems                    • Generates text in 1/3 of the time needed for typing                               2005
                                                                • Face to face time with patient                                 Month total                104,570.93    105,433.59    117,312.79     97,513.10    98,602.87

    Run-in                                                      • Improves efficiency 50% over transcription services                                                                                                                 Weakness of Speech Recognition Systems
                                                                                                                                                                Savings with EMR
    • All physicians digital dictate                            • Shortens front-end, back-end processing                       Sum of Jan-May of 2005                                  $   523,433.28
                                                                                                                                                                                                                                 •     Technology adapted to CCHS
    • Dictation is recorded as a sound file                     • Adapted to different physician preference – editing,           Sum of Jan-May of 2006                                       53,306.37
                                                                                                                                                                                                                                 •     Physician buy in
                                                                  correcting, hard copy                                                                                                                                          •     Receiver operator dependency
    • Generation of voice recognition file by transcriptions                                                                    5 months of Savings                                     $   470,126.91
                                                                                                                                                                                                                                 •     Upfront costs during transition
    • Evaluation mode between digital and voice recognition     • Attach reports and images in real-time                                                                                                                               (transcription plus speech recognition systems)
                                                                                                                                Approximate annual savings using a full year is $1.3 million.
    • Speech recognition adapts to speaking habit, profile of   • Reduce telephone traffic
      each physician                                            • Improves physician efficiency by reducing                                           Current Transcription Costs                                                                          Speech Magic
                                                                                                                                                     2007       Jan          Feb            March        April        May
    • Switch to speech recognition                                interruptions                                                   Month total                 14,299.60      9,539.77       9,958.10    10,354.38     9,154.52   •     Discussions between Philips and EPIC
Advantages of Fusion Systems                                                                                                                                                                                                     •     License to distribute both together
• Turnaround time from 24-96 hours to 4-72 hours                     Consult note to                Improves patient care                                                                                                        •     Support for Speech Magic in Smart Text
                                                                     physician same day                                                                                                                                                –   Progress notes, in-basket ED
• Annual transcription cost in outsourcing reduced by 40%                                                                   Software License (Dragon Naturally Speaking) $800-1000                                                     –   Speech to text
• Transcription production improves 50%                                                                                                                                                                                                –   Use of Smart Tools to create pre-defined text
                                                                                                                                                                                                                                       –   Basic navigation available; workflow specific n/a
• Little change in physician work method                                                                                                        $200 Set-up

                                                                                                                                  Hardware                       $100 - 400
                                                                                                                                                                                                                                     Opportunities of Speech Recognition Systems
                                                                                                                                  Upgrade                         Headset,
          Speech Recognition Systems: Real Time                                                                                                                 microphones                 Per physician $2400                  •     “Complete” transition to fully electronic record
                                                                                                            Reduces LOS
                                                                                                                                                                                                                                 •     Reduce overhead, freeing funds for innovative
 • Link speech recognition to word processing in real-time                                                                                                                                                                             projects or direct patient care
                                                                       Redirect FTEs                                                    Update every 4 years $400
 Requirements                                                          to patient care
 • Open document                                                                                                                                                                                                                        Threats of Speech Recognition Systems
 • Switch on and off italics                                    Acuracy                                                                                       Intangible Costs                                                   •     Technology may not be advanced enough for rapidly
 • Go back and forth for corrections                            • 90-95%- operator dependent                                                                                                                                           changing medical environments (limited usefulness to
                                                                                                                            •      Personnel for maintenance                                                                           non-specialized or non-complex practice
 • Allow for spelling                                           • PC compatible headsets wired, wireless, handheld
                                                                                                                            •      Time to full implementation by physician and system                                                 environments)
 • Insert blocks of text (“smart phrases”) (“speech               with barcode scanners, digital recorders                                                                                                                       •     Lack of integration: systems not fully integrated
                                                                                                                            •      Downtime
    macros”)                                                                                                                                                                                                                           resulting in “piecemeal” add on technologies that are
                                                                                                                            •      Integration into Epic                                                                               cumbersome
 • Navigate forms without keyboard
                                                                       Transcription Costs per Month: CCHS                                                                                                                       •     Start-up costs
 • Recognize acronyms which are turned to full phrases
 • Differentiate “command” from “text”                                                                                           Speech Recognition Systems: Stakeholders                                                                                Recommendation
Pitfalls                                                        •   Private practice $1000/physician/year
• Receiver operator dependent                                   •   General Internal Medicine: $79,000/year                  •      Transcriptionist – Fusion Technology                                                         • Fusion technology for efficiency before advancing to
• Upstart physician time                                        •   Orthopedic Surgery: $168,000/year                        •      Physicians- Real Time Technology                                                               Real Time
• Acoustic background interference                              •   Urology: $39,000/year                                                                                                                                        • Once efficiency is tested, roll out Real Time to target
• Dialect differences                                           •   Taussig Cancer Center $24,000/year                                                                                                                             services for feasibility. Target services would have
• Pronunciation
                                                                                                                                       Speech Recognition Systems Address                                                          been involved with Fusion testing
                                                                    Transcription Costs by Transcription Service                                                                                                                 • First step is to investigate, interview and bid multiple
• Integration into existing database systems                                                                                •      Front-end, back-end report processing                                                           companies (Dragon, Philips Speech Recognition)
                                                                                    Per Month
• Context vocabulary                                                                                                        •      Delays in care                                                                                Action Items
                                                                        Company                     Costs per Year
• Spontaneous dictation often leads to:                                 CYMED of Ohio                         906,147       •      Inefficiencies due to:                                                                        • Clarify indirect costs
      –   Missing articles                                                                                                          – Telephone traffic                                                                          • Invite companies to display , present and bid
                                                                        Medical Transcription              47,615
      –   Missing verbs                                                                                                             – Time delays between studies/consultation and reports                                       • Reassess based on technology-go or no go
                                                                        Marquis Inc                         1,943,855
      –   Lack of or incorrect punctuation                                                                                                                                                                                       • Trial Fusion Speech Recognition-3-6 months
                                                                        Premier Office Tech                142,509          •      Length of stay
      –   Redundancies and repeats                                                                                                                                                                                               • Visit facilities using Speech Recognition Systems
                                                                        Spheris Operations               2,608,531          •      Communication between consultant/referring physician
•     Integration into other systems                                                                                                                                                                                             • Assess efficiency, accuracy, turn around time,
                                                                                                                            •      Overhead costs
•     Electronic failure (power outage “downtime”)                      Totals                             $5,648,657                                                                                                              satisfaction, cost , work accomplished per unit

Shared By:
Description: Speech recognition is a cross-disciplinary. Past two decades, speech recognition technology has made significant progress, starting from the lab to the marketplace. It is expected that the next 10 years, speech recognition technology into the industrial, household appliances, communications, automotive electronics, medical, family services, consumer electronics and other fields. Voice recognition dictation machine in some areas by the U.S. media as the application of computer development in 1997 one of 10 events. Many experts believe that voice recognition technology in 2000 to 2010 the top ten key areas of information technology, technology development technologies. Voice recognition technology in the areas of: signal processing, pattern recognition, probability theory and information theory, mechanism of sound and hearing mechanism, artificial intelligence and so on.