Early Detection of Severe Flu Outbreaks using Contextual Word Embeddings

Abstract

The purpose of automated health surveillance systems is to predict the emergence of a disease. In most cases, these systems use a text categorization model to classify any clinical text into a category corresponding to an illness. The problem arises when the target classes refer to diseases sharing multiple information such as symptoms. Thus, the classifier will have difficulty discriminating the disease under surveillance from other conditions of the same family, causing an increase in misclassification rate. Clinical texts contain keywords carrying relevant information to distinguish diseases with similar symptoms. However, these specific words are rare and sparse. Therefore, they have a minor impact on machine learning models' performance. Assuming that emphasizing specific terms contributes to improving classification performance, we propose an algorithm that enriches training samples with terms semantically similar to specific terms using the deep contextualized word embeddings ELMo. Next, we devise a weighting scheme combining chi-square and semantic scores to reflect the relatedness between features and the disease under surveillance. We evaluate our model using the SVM algorithm trained on i2b2 dataset supplemented by documents collected from Ibn Sina hospital in Rabat. Experimental results show a clear improvement in classification performance than baseline methods with an F-measure reaching 86.54%.

Intelligent preliminary diagnosis system for diseases with similar clinical presentation

“…The app predicted influenza activity with a significant correlation (ρ = 0.878) compared to data from the Korea center for disease control and prevention, detecting the epidemic 10 days before the official alert. On the other hand, Karsi et al [18] proposed an approach to improve classification in automated health surveillance systems. Using deep contextualized word embeddings from embeddings from language models (ELMo) to enrich training samples with semantically similar terms and a weighting scheme, their improved model, evaluated with support vector machine (SVM) on the i2b2 dataset, showed a significant improvement in classification, achieving an F-measure of 86.54%.…”

“…This specific technique, which focuses on symptom-specific logic and data processing, can contribute to diagnostic accuracy. Meanwhile, some studies opted for the implementation of rule-based expert systems [16] and the application of machine learning models, such as the SVM [18].…”