Clinical coding is carried out in hospitals to support statistical analysis of clinical data that leads to funding, insurance claims processing and research. Ever expanding and changing clinical classification systems such as ICD-10-AM and ACHI, challenges in the healthcare industry are increased due to increasing set of codes, the complexity of manual code assignment, and extensive training and recruitment costs. The use of Natural Language Processing (NLP) and Machine learning (ML) techniques for computer-assisted coding or auto-coding is considered as a possible solution to overcome the problems of manual coding. This perception is questioned in this work, by carrying out experimental tests on a selected set of NLP and ML techniques, using 190 discharge summaries related to diseases of respiratory and gastrointestinal systems. The results indicate that accuracy of auto-coding ranges between 40% to 79% depending on the computational techniques used. The paper concludes that without human involvement, auto-coding would not be a reality in the current healthcare data environment.
| Authors : | Rajvir Kaur and Jeewani Anupama Ginige |
| Publication Year : | 2018 |
| Publication Type : | 1 |
| Conference Name : | Health Information Management Association of Australia (HIMAA)- National Centre for Classification in Health (NCCH) National Conference |
| Conference Location : | Hobart, Tasmania |
| Pages : | 25-33 |
| ISSN : | 978-0-9946206-5-1 |
| Publisher : | Health Information Management Association of Australia Ltd |
| Place of Publication : | Australia |
| Link to article : | https://www.accd.net.au/Downloads/Current/Conferences/HIMAA_NCCH%202018%20Proceedings.pdf |
| Conference Website : | https://himaa.eventsair.com/2018-hobart-himaa-ncch-conference/ |