Identification of Symptoms Based on Natural Language Processing (NLP) for Disease Diagnosis Based on International Classification of Diseases and Related Health Problems (ICD-11)
| Autor: | Sritrusta Sukaridhoto, Alviansyah Arman Yusuf, Urfiyatul Erifani, Dwi Kurnia Basuki, Fariz Bramasta Putra, Dzakiyah Salma Humairra, Yogi Putra Pratama, Rizqi Putri Nourma Budiarti, Heri Yulianus |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
030506 rehabilitation
Grammar Process (engineering) Computer science business.industry media_common.quotation_subject 030229 sport sciences Disease computer.software_genre Clinical decision support system 03 medical and health sciences Identification (information) 0302 clinical medicine Named-entity recognition Artificial intelligence 0305 other medical science business Function (engineering) computer Natural language Natural language processing media_common |
| Zdroj: | 2019 International Electronics Symposium (IES). |
| Popis: | The digestive system is a vital organ, considering its function in processing food and drinks that are consumed every day so it is very important to be maintained. But sometimes people have a lack of awareness and knowledge of the initial symptoms of digestive diseases so that being a factor causing the disease to be serious can even cause death. Identification of symptoms as early as possible is very important for the diagnosis process so that immediate control measures can be taken to prevent and overcome the spread of disease. Anamnesis process is needed to get the symptoms of the disease, question and answer process between the patient and medical personnel whose results are stored in the Electronic Medical Record (EMR) in the form of narration to assist in the process of Clinical Decision Support (CDS). EMR is often difficult to do computing processing due to inappropriate grammar. For computers to process natural languages, Natural Language Processing (NLP) is used. In this study, an NLP system was created that can identify symptoms of the digestive disease by using to optimize the CDS process. The method used to identify symptoms of the disease is Named Entity Recognition (NER), which determines which tokens are included in the symptoms of the disease. The model trained with 800 epochs produces f1-score accuracy of 0.79. Experimental results show that the NER process supported by stemming and removing stopwords in pre-processes can improve system accuracy. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|