A study of entity-linking methods for normalizing Chinese diagnosis and procedure terms to ICD codes

Wang, Qiong; Ji, Zongcheng; Wang, Jingqi; Wu, Stephen; Lin, Wei; Li, Wenzhen; Li, Ke; Xiao, Guohui; Jiang, Qing; Xu, Hua; Zhou, Yi

doi:10.1016/j.jbi.2020.103418

Cited by 34 publications

(18 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The problem of ICD-10 code association has been explored in multiple works. One of the primary issues is the lexical variability of the diagnoses which has been addressed by incorporating knowledge bases, such as dictionaries, expanding the dataset through machine translation and working directly with meanings [1], synonyms and ontologies [19]. Automatic association of ICD-10 codes task was addressed for several languages other than English [16] like Spanish [1], Chinese [17], Dutch [7], and Turkish [11] [4].…”

Section: Related Workmentioning

confidence: 99%

“…A few works are tackling the classification task using information retrieval techniques , statistical classifiers and document similarity [1], as well as machine learning techniques [6]. Another approach is to view the problem as a multi-label classification task and use neural networks like CNN, LSTM/BiLSTM and HA-GRU [19], or applying BERT which has shown good results on this task in German [3]. Due to the high number of classes, combining different models into ensemble has shown slight improvement in the results [3].…”

Section: Related Workmentioning

confidence: 99%

“…The text classification corpus consists of 345,591 diagnoses and their corresponding ICD-10 codes. We apply data pre-processing on the diagnosis text by stemming using bulstempy 18 and removing the stop words using the BTB list 19 . We split the dataset using stratification and assign 80% for training, 10% for hyper-parameter tuning and 10% for testing using scikit-learn library 20 .…”

Section: Fine-tuning Bert For Icd-10 Classification Taskmentioning

confidence: 99%

“…As input for the fine-tuning procedure we tokenize the input diagnosis and prepend and append the special tokens [CLS] and 17 https://towardsdatascience.com/perplexity-intuition-and-derivation-105dd481c8f3 18 https://github.com/mhardalov/bulstem-py 19 http://bultreebank.org/wp-content/uploads/2017/04/BTB-StopWordList.zip 20 https://scikit-learn.org/stable/ [SEP] around it, as described in the original BERT paper [13].…”

Section: Fine-tuning Bert For Icd-10 Classification Taskmentioning

confidence: 99%

See 3 more Smart Citations

Automatic ICD-10 codes association to diagnosis: Bulgarian case

Velichkov

Gerginov

Panayotov

et al. 2020

CSBio '20: Proceedings of the Eleventh International Conference on Computational Systems-Biology and Bioinformatics

View full text Add to dashboard Cite

This paper presents an approach for the automatic association of diagnoses in Bulgarian language to ICD-10 codes. Since this task is currently performed manually by medical professionals, the ability to automate it would save time and allow doctors to focus more on patient care. The presented approach employs a fine-tuned language model (i.e. BERT) as a multi-class classification model. As there are several different types of BERT models, we conduct experiments to assess the applicability of domain and language specific model adaptation. To train our models we use a big corpora of about 350,000 textual descriptions of diagnosis in Bulgarian language annotated with ICD-10 codes. We conduct experiments comparing the accuracy of ICD-10 code prediction using different types of BERT language models. The results show that the MultilingualBERT model (Accuracy Top 1-81%; Macro F1-86%, MRR Top 5-88%) outperforms other models. However, all models seem to suffer from the class imbalance in the training dataset. The achieved accuracy of prediction in the experiments can be evaluated as very high, given the huge amount of classes and noisiness of the data. The result also provides evidence that the collected dataset and the proposed approach can be useful in building an application to help medical practitioners with this task and encourages further research to improve the prediction

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Fine-tuning Bert For Icd-10 Classification Taskmentioning

confidence: 99%

Section: Fine-tuning Bert For Icd-10 Classification Taskmentioning

confidence: 99%

See 2 more Smart Citations

Automatic ICD-10 codes association to diagnosis: Bulgarian case

Velichkov

Gerginov

Panayotov

et al. 2020

CSBio '20: Proceedings of the Eleventh International Conference on Computational Systems-Biology and Bioinformatics

View full text Add to dashboard Cite

show abstract

“…Already there are various research for other languages. Wang, Qiong, et al [5] present a study that aims to develop and evaluate effective methods that can normalize diagnosis and procedure terms written by physicians to standard concepts in ICD in Chinese using an entity-linking framework and two manually annotated datasets (8,547 diagnoses and 8,282 procedures). Marovac, Avdić et al [6] present in their paper the process of creating medical lexical resources for the Serbian language and they achieve mapping to certain ICD-10 codes with precision over 80%.…”

Section: Introductionmentioning

confidence: 99%

Automatic Generation of Annotated Corpora of Diagnoses with ICD-10 codes based on Open Data and Linked Open Data

Boytcheva

Velichkov

Velchev

et al. 2020

Proceedings of the 2020 Federated Conference on Computer Science and Information Systems

View full text Add to dashboard Cite

We propose methods for automatic generation of corpora that contains descriptions of diagnoses in Bulgarian and their associated codes in ICD-10-CM (International Classification of Diseases, 10th revision, Clinical Modification). The proposed approach is based on the available open data and Linked Open Data and can be easily adapted for other languages. The resulted corpora generated for the Bulgarian clinical texts consists of about 370,000 pairs of diagnoses and corresponding ICD-10 codes and is beyond the usual size that can be generated manually, moreover it was created from scratch and for a relatively short time. Further updates of the corpora are also possible whenever new open resources are available or the current ones are updated.This research is partially funded by the Bulgarian Ministry of Education and Science, grant DO1-200/2018 'Electronic health care in Bulgaria' (e-Zdrave) and the Bulgarian National Science Fund, grant DN-02/4-2016 'Specialized Data Mining Methods Based on Semantic Attributes' (IZIDA). We are grateful to anonymous reviewers for useful comments and suggestions.

show abstract