ICD-10 Coding of Spanish Electronic Discharge Summaries: An Extreme Classification Problem

Abstract

Objective: Medical coding is used to identify and standardize clinical concepts in the records collected from healthcare services. The tenth revision of the International Classification of Diseases (ICD-10) is the most widely-used coding with more than 11,000 different diagnoses, affecting research, reporting, and funding. Unfortunately, ICD-10 code sets tend to follow biased, unbalanced, and scattered distributions. These distribution attributes, along with high lexical variability, severely restrict performance when coded clinical records are used to infer code sets in uncoded records. To improve that inference, we explore a combination of example-based methods optimized to capture codes with different appearance frequencies in data sets. Materials and Methods: The proposed exploration has been carried out on Spanish hospital discharge reports coded by experts, excluding all sentences without any biomedical concept. Representations based on semantic and lexical features are explored, using both global and labelspecific attributes. In turn, algorithms based on binary outputs, groups of subsets and extreme classification are compared. Lists of codes together with their confidence values (certainty probabilities) are suggested by each method. Results: Diverse spectral behaviors are shown for each method. Binary classifiers seem to maximize the capture of more popular codes, while extreme classifiers promote infrequent ones. In order to exploit such differences, ensemble approaches are proposed by weighting every output code according to the method, confidence value and appearance frequency. The rule-based combination reaches a 46% Precision at 10 (P@10), which means a 15% improvement over the best individual proposal. Conclusion: Assembling methods based on weighting each code according to training frequency and performance can achieve better overall Precision scores on extreme distributions, such as ICD-10 coding. INDEX TERMS Extreme classification, XMTC, ICD-10 coding, text mining.

Automatic ICD-10 codes association to diagnosis: Bulgarian case

“…International Statistical Classification of Diseases and Related Health Problems is a classification system for medical diagnoses which is used for health management and clinical purposes 1 . This classification is used world-wide and has translations in multiple languages.…”

“…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]. Some preliminary work was done for Bulgarian using SVM and small training corpora [8], with achieved F-score 0.84.…”

“…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].…”

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

“…It is important to note that in the process of conversion some paragraphs can be damaged, thus some minor manual cleaning and formatting of the result file is needed. For the rest of the (automatic) part of the processing 7 , the structure of the index is very important for the rule based information extraction. A screenshot of the document can be seen at Figure 2.…”

“…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%. Almagro, Unanue et al [7] have been carried out an exploration on 7254 Spanish hospital discharge reports for a period of 3 years resulting in total 76,525 identified codes with approximately 7,000 unique ones. Bagheri, Sammani et.…”