Kurniasari, Dian and Usman, Mustofa and Warsono, Warsono and Lumbanraja, Favorisen R (2022) Medical Report Classification Using BioWordVec Contained on Deep Learning Method. In: The 4th International Conference on Applied Sciences, Mathematics, and Informatics (ICASMI), 8 - 9 September 2022, Bandar Lampung. (Unpublished)

Buku Abstrak ICASMI 2022.pdf

Download (2MB) | Preview


A medical report classification contains a comprehensive and valuable health information on a medical diagnosis and patients’ treatment and therapy. However, text data of medical report, for example on leukaemia data are huge in volume and in unstructured format, do not follow natural language grammar. In order to do medical report classification, this study proposes a deep learning method with pre-trained word embedding BioWordVec for balanced and unbalanced data. The main purpose of this study, therefore, is to explore performances of several algorithms of deep learning. The proposed deep learning algorithms in this study are Convolution Neural Network (CNN), Long Short-Term Memory (LSTM), Hybrid CNN-LSTM and Hybrid LSTM-CNN. The results demonstrate that the medical report classification using the Hybrid LSTM_CNN method with pre-trained word embedding BioWordVec using the k-Fold cross-validation technique after resampling the data obtained the highest accuracy of 98.05%. At the same time, the average value of precision, recall, and F1-score was 97.33%. Therefore the result using the data splitting technique after resampling the data, the highest accuracy was obtained, namely, 98.83%, while the average value of precision, recall, and F1-score was 100%. Keywords: Medical Report Clasification, Deep Learning Method, k-Fold Cross Validation, Resampling, NLP.

Item Type: Conference or Workshop Item (Paper)
Subjects: Q Science > Q Science (General)
Divisions: Fakultas Matematika dan Ilmu Pengetahuan Alam (FMIPA) > Prodi Matematika
Depositing User: DIAN KURNIASARI
Date Deposited: 10 Oct 2022 02:28
Last Modified: 10 Oct 2022 02:28
URI: http://repository.lppm.unila.ac.id/id/eprint/45618

Actions (login required)

View Item View Item