Gale B, Safriel Y, Lukban A, Kalowitz J, Fleischer J, Gordon D
State University of New York Health Sciences Center at Brooklyn, USA.
Radiol Manage. 2001 Mar-Apr;23(2):18-22.
Computer-based voice-recognition software has many potential advantages in producing reports of radiology procedures. Using voice-recognition systems, however, necessitates increased involvement of the radiologist in the process of producing the report. The radiologist, previously responsible only for recording a report onto tape and ensuring the integrity of the final report now becomes obligated to interact with the computer and to ensure the integrity of the transcription process as well as the accuracy of the final report. Two attending radiologists and one first-year radiology resident at an academic medical center timed the production of reports using both the voice recognition system and tape transcription of reports of plain films (n = 27), mammograms (n = 25), and GI/GU exams (n = 17). In addition, the taped dictations were transcribed and then corrected by the physicians. The additional correction time (determined as an average) was added to the tape times to produce a "corrected" tape time. Paired T-Test procedures were used to determine if pairs of readings (voice recognition vs. corrected tape transcription) differed in length of time. In addition, the data was stratified into three groupings--plain film, mammography, and GI/GU--in order to assess for differences between modalities. The length of time required to produce a radiology report using the commercial radiology voice recognition system employed at our center is significantly longer than that required by the traditional corrected tape transcription system. One motivation to use a voice recognition system is the cost savings achieved by eliminating transcriptionists and replacing them with the radiologist using the voice recognition system. In our institution this cost savings is estimated to be $100,000 annually. This apparent cost savings is reduced by the cost of the lost productivity of the radiologist. Compared to tape transcription, our data demonstrate a significant increase in the amount of time necessary for radiologists to produce a radiology report when using the voice recognition system currently employed in our hospital. While it is likely that future systems will require less extra time, this factor needs to be accounted for when departments consider using such systems to replace transcriptionists with radiologists.
基于计算机的语音识别软件在生成放射学检查报告方面有许多潜在优势。然而,使用语音识别系统需要放射科医生在报告生成过程中更多地参与。放射科医生以前只负责将报告录制到磁带上并确保最终报告的完整性,现在则有义务与计算机交互,并确保转录过程的完整性以及最终报告的准确性。一所学术医疗中心的两名主治放射科医生和一名第一年的放射科住院医生,对使用语音识别系统和普通X光片(n = 27)、乳房X光片(n = 25)以及胃肠道/泌尿生殖系统检查(n = 17)报告的磁带转录来生成报告的时间进行了计时。此外,磁带听写内容被转录,然后由医生进行校正。额外的校正时间(以平均值确定)被加到磁带计时中,以得出“校正后”的磁带时间。采用配对t检验程序来确定成对的读数(语音识别与校正后的磁带转录)在时间长度上是否存在差异。此外,数据被分为三个类别——普通X光片、乳房X光造影和胃肠道/泌尿生殖系统——以评估不同检查方式之间的差异。使用我们中心所采用的商业放射学语音识别系统生成一份放射学报告所需的时间,明显长于传统的校正磁带转录系统所需的时间。使用语音识别系统的一个动机是通过省去转录员并用使用语音识别系统的放射科医生取而代之来实现成本节约。在我们机构,估计每年可节约成本10万美元。但这种明显的成本节约因放射科医生生产力损失的成本而减少。与磁带转录相比,我们的数据表明,使用我们医院目前采用的语音识别系统时,放射科医生生成一份放射学报告所需的时间显著增加。虽然未来的系统可能需要更少的额外时间,但当各科室考虑使用此类系统用放射科医生取代转录员时,这一因素需要加以考虑。