EHR usability is lacking. EHR documentation is slow and inflexible. As a result, EHR documentation quality suffers. Our NIH-funded study hypothesizes that a new method of EHR data capture using Natural Language Processing (“NLP”) would improve the quality of EHR documentation. “NLP-Based EHR Data Capture” is a method by which a physician dictates, the dictation is transcribed, NLP generates structured data from the dictation, and the structured data is inserted in the EHR. With test subjects from various specialties, we compared the NLP-Based EHR Data Capture method against the standard EHR data capture method by measuring the amount of time each method requires, how satisfied physicians are with them, and the documentation quality they generate.
The rate of meaningful adoption of EHR systems among U.S. health care providers remains below expectations, while many physicians who are using EHRS report dissatisfaction with their systems. Physician time and computer skill requirements, and psychological factors related to ease of use, have been cited among the barriers to EHR adoption and meaningful use. These usability issues, which impact clinician and staff workflow, reduce the quality of documentation found within EHRs, and the extent to which their potential for systematic health care improvement is realized.
Natural Language Processing (NLP)-Based EHR Data Capture may solve these problems by using NLP to capture and structure patient record detail from unstructured narrative documentation or other sources. NLP-Based EHR Data Capture is a process by which the physician dictates, the dictation is either speech-recognized or transcribed, the dictated text is processed by NLP, and the structured codes generated by NLP are inserted in the EHR.
Our NIH-funded study is titled “Applying NLP to Free Text as an EHR Data Capture Method to Improve EHR Usability.” The study assesses the effectiveness of NLP-Based EHR Data Capture as a solution for fostering adoption and better documentation by improving EHR usability measures and solving problems of: 1) inefficiency, or time required for data capture; 2) effectiveness, including documentation completeness and quality; and 3) clinician dissatisfaction with EHR use.
For the study, neurologists, cardiologists, endocrinologists, and nephrologists were recruited. Each subject is given four admitting note history and physical (“H&P”) sections representing common scenarios for his or her specialty. Each admitting note is documented with a different combination of standard EHR data capture (which requires using keyboard and mouse to fill EHR form fields) and NLP-Based EHR Data Capture.
1. The subject is provided a case. The subject will document the case using standard EHR data capture and/or NLP-Based EHR Data Capture.
2. The NLP-Based EHR Data Capture method requires that the subject dictates her note. The subject’s dictations are transcribed and run through the NLP engine. A third party places the transcription and structured data in the EHR sections indicated by the NLP output.
3. The following day, the subject reviews each admitting note generated using both methods. Afterward, the subject completes a Standard Usability Scale questionnaire for each EHR data capture method.
4. Next, a documentation expert, another physician with expertise in the subject’s field, reviews the documentation generated and completes a questionnaire to assess each note’s completeness, accuracy, and documentation quality.
Conclusion: By the above procedure, we compare the time required for data capture, physician satisfaction, and documentation quality resulting from NLP-Based EHR Data Capture and standard EHR data capture. Results and analysis will be presented.
ZyDoc Chairman James M. Maisel, M.D. has unique medical informatics experience as CEO of ZyDoc and Retina Group of NY. Since training as the Esther Dyson Foundation vitreoretinal fellow at NY Presbyterian Cornell, he has served thousands of ophthalmology patients, and diabetic patients and their health care providers. He is an educator for patients and for diabetic educators and professionals, and a pharmaceutical medical advisor.
Formerly CEO of HOST, Dr. Maisel led ZyDoc in developing informatics technologies including a prototype EMR purchased by the DOD, speech recognition, e-transcription, and NLP. He is currently Principal Investigator for an NIH grant-funded study, “Applying NLP to Free Text as an EHR Data Capture Method to Improve EHR Usability,” involving ZyDoc’s NLP-powered MediSapien™ knowledge management platform. The grant funding was awarded to ZyDoc through a Phase I SBIR grant of the National Library of Medicine of the National Institutes of Health. In mid-2012, in competition against MModal, Nuance, and others, ZyDoc was awarded a contract by the State of Alaska Department of Health and Social Services Division of Behavioral Health / Alaska Psychiatric Institute (API) to provide an enterprise-wide voice recognition, transcription and dictation system including MediSapien reporting. Under Dr. Maisel’s leadership, ZyDoc has also won numerous awards from the medical informatics, speech recognition and computing industries, including: LISA Awards for Business Productivity, Transcription Workflow, EMR and Transcription; Tablet PC Healthcare Productivity Award; SpeechTECHNOLOGY Innovators Award; AMA Technology Summit Nominee; and Medical Records Institute 3rd National Medical Transcription Business Recognition Program.
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
For exhibition and sponsorship opportunities at Strata Rx conference, contact Sharon Pierce at (203) 304-9476 or email@example.com
For information on trade opportunities with O'Reilly conferences email mediapartners
View a complete list of Strata Rx 2013 contacts