Etukudo, Deborah Bassey and Covenant University, Theses (2022) DEVELOPING A PHYSICIAN-PATIENT SPEECH RECOGNITION AND DOCUMENTATION MODEL FOR NIGERIAN HOSPITALS. Masters thesis, Covenant University Ota.
![[img]](http://eprints.covenantuniversity.edu.ng/style/images/fileicons/application_pdf.png) |
PDF
Download (93kB) |
Abstract
Despite the increase in the use of automated systems for physician-patient consultation in developed countries, Nigerian hospitals are still actively using the traditional approach which implies using pen and paper or typing on the computer in documenting and extracting the conversation between the physician and the patient. This has been observed to cause long consultation time, burnout during consultations, bad consultation experience for the patient, etc. Due to the Nigerian accent, the present speech recognition system when tested with a Nigerian accent could not properly detect it. This study aims to develop an automatic speech recognition system for extracting and classifying structured data from physician-patient conversations done in a Nigerian hospital. A pretrained Wav2vec model was retrained with the primary data used in this work for better detection of the Nigerian accent. The accuracy score of the SOAP classification using a deep learning approach, gave a total of 83%, compared to that of the machine learning approach which gave a total accuracy of 82%. For the speech recognition system evaluation, the Word error rate, word information lost, and match error rate gave a better accuracy of 0.889, 0.800, and 0.896 respectively as compared to the TIMIT dataset which gave 0.999, 0.999, and 0.999 respectively. The proposed system performs better in the identification of the Nigerian accent than the existing system.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Automatic Speech Recognition, Wav2vec, Electronic Health Record, SOAP Notes, Text Classification, MLP. |
| Subjects: | Q Science > QA Mathematics Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
| Divisions: | Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science |
| Depositing User: | nwokealisi |
| Date Deposited: | 23 Aug 2023 11:59 |
| Last Modified: | 23 Aug 2023 11:59 |
| URI: | http://eprints.covenantuniversity.edu.ng/id/eprint/17288 |
Actions (login required)
 |
View Item |
