EMTE: An Enhanced Medical Terms Extractor using pattern matching rules - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

EMTE: An Enhanced Medical Terms Extractor using pattern matching rules

Résumé

Background: Downstream tasks, like clinical textual data classification, perform best when given good quality datasets. Most of the existing clinical textual data preparation techniques rely on two main approaches, removing irrelevant data using cleansing techniques or extracting the valuable data using feature extraction techniques. Although both approaches showed promising results on experimental datasets, they still have limitations, mainly when applied to real-world datasets. The latter datasets are challenging because they might include incomplete, inconsistent and heterogeneous records. Therefore, understanding the nature of the data is vital in order to improve its quality before its transmission to the classification model. Methods: This paper proposes a cleansing approach (called EMTE) which extracts phrases (medical terms, abbreviations, and negations) using pattern-matching rules based on the linguistic processing of the clinical textual data. Without requiring any training, EMTE extracts the valuable medical data from clinical textual records even if they have different writing styles. Furthermore, since EMTE relies on dictionaries to store abbreviations and pattern-matching rules to detect the abbreviations, negations, and medical terms, it can be easily maintained and extended for industrial use. Experiments: To evaluate the performance of our approach, it was compared to three other techniques. All four cleansing techniques were applied to a large industrial imbalanced dataset, consisting of 2.21M samples from different specialties with 1,050 ICD-10 codes. The outputs of these cleansing techniques were fed to several Deep Neural Network (DNN) algorithms to solve the ICD-10 (International Classification of Diseases, 10th edition) multi-label classification problem. In this problem, the clinical textual samples are the input data, and the ICD-10 codes are the actual labels. Results: The experimental results on several Deep Neural Network (DNN) algorithms showed that our cleansing approach significantly improves the trained models’ performance when compared to the other tested techniques and according to different metrics.
Fichier principal
Vignette du fichier
2d4c2089-5869-4a0a-ae0e-c05aa9e409ac-author.pdf (409.85 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04680298 , version 1 (28-08-2024)

Identifiants

  • HAL Id : hal-04680298 , version 1

Citer

Monah Bou Hatoum, Jean-Claude Charr, Alia Ghaddar, Christophe Guyeux, David Laiymani. EMTE: An Enhanced Medical Terms Extractor using pattern matching rules. international conference on agents and artificial intelligence, Feb 2023, Lisbon, Portugal. ⟨hal-04680298⟩
0 Consultations
0 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More