DADilek Aydogan-Kilic

Dilek Aydogan-Kilic

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Dilek Aydogan-Kilic has a total of 3 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Comparative study of feature extraction methods for automated ICD code classification using MIMIC-III medical notes and deep learning models

    by Dilek Aydoğan Kılıç, Deniz Kenan Kılıç, Izabela Ewa Nielsen

    ICD standardizes diagnosis codes globally, aiding payments, research, planning, and quality management. Its complexity leads to longer exams, higher training costs, increased workforce needs, coding errors, and unreliable data. Automated ICD systems using ML address these issues. Long medical notes complicate ML, making feature extraction crucial for efficient ICD classification. Despite numerous studies, no systematic analysis of feature extraction methods, especially in deep learning (DL), exists. The MIMIC-III dataset is used with two preprocessing combinations, fundamental and advanced. TF-IDF, word2vec, GloVe, fastText, and BERT feature extraction methods are compared using DL models such as NN, CNN, and BiLSTM. For word2vec and fastText, CBOW and skip-gram architectures are compared. ROC-AUC, F1-score, precision, and recall metrics are calculated for DL performances. Advanced preprocessing improves performance for all feature extraction and DL methods. The best results for advanced preprocessing are micro ROC-AUC of 91.74\% (BiLSTM+fastText (skip-gram)), macro ROC-AUC of 88.58\% (BiLSTM+word2vec (CBOW)), micro F1/precision of 64.84\%/62.34\% (BiLSTM+word2vec (CBOW)), micro recall of 68.16\% (BiLSTM+fastText (skip-gram)), macro F1/precision of 59.67\%/57.71\% (BiLSTM+word2vec (CBOW)), and macro recall of 63.38\% (BiLSTM+fastText (skip-gram)). FastText is the most successful feature extraction method in DL models with fundamental preprocessing. However, models using well-implemented preprocessing highlight other feature extraction methods that perform better and operate more quickly. As DL model performance improves, differences between feature extraction performances diminish. Though not focused on the best results, CNN and BiLSTM with word2vec, GloVe, and fastText are competitive with current studies. Lastly, if computing power is limited, CNN may be preferable over BiLSTM with these feature extraction methods.

  • Open Access

    EXAMINATION OF SUMMARIZED MEDICAL RECORDS FOR ICD CODE CLASSIFICATION VIA BERT

    by Dilek Aydogan-Kilic, Deniz Kenan Kilic, Izabela Ewa Nielsen

    The International Classification of Diseases (ICD) is utilized by member countries of the World Health Organization (WHO). It is a critical system to ensure worldwide standardization of diagnosis codes, which enables data comparison and analysis across various nations. The ICD system is essential in supporting payment systems, healthcare research, service planning, and quality and safety management. However, the sophisticated and intricate structure of the ICD system can sometimes cause issues such as longer examination times, increased training expenses, a greater need for human resources, problems with payment systems due to inaccurate coding, and unreliable data in health research. Additionally, machine learning models that use automated ICD systems face difficulties with lengthy medical notes. To tackle this challenge, the present study aims to utilize Medical Information Mart for Intensive Care (MIMIC-III) medical notes that have been summarized using the term frequency-inverse document frequency (TF-IDF) method. These notes are further analyzed using deep learning, specifically bidirectional encoder representations from transformers (BERT), to classify disease diagnoses based on ICD codes. Even though the proposed methodology using summarized data provides lower accuracy performance than state-of-the-art methods, the performance results obtained are promising in terms of continuing the study of extracting summary input and more important features, as it provides real-time ICD code classification and more explainable inputs.