ABSTRACT
We present Language-Interfaced Fine-Tuning (LIFT) in application to COVID-19 patient survival classification. LIFT describes translating tabular Electronic Health Records (EHRs) into text inputs for transformer neural networks. We study LIFT with a dataset of 5,371 COVID-19 patients. We focus on the predictive task of survival classification utilizing demographic and medical history features. We begin by presenting information about our dataset. We preface our investigation in text-based transformers by reporting the performances of conventional machine learning models such as Logistic Regression and Random Forest classifiers. We also present the results of a few configurations of tabular input-based Deep Multilayer Perceptron (MLP) networks. 86% of the patients in our database survived in the measured time window. Thus, predictive models are heavily biased to predict that a patient will survive. We emphasize that this problem of Class Imbalance was a major challenge in developing these models. Our balanced sampling strategy from examples in the majority and minority classes is crucial to achieving even reasonable predictive performance. For this reason, we also report performance based on Precision, Recall, and F-score metrics, in addition to Accuracy. Having established baselines with tabular inputs, we then shift our focus to the prompts for translating from tabular to text inputs. We report the performance of 5 prompts. The LIFT model achieves an F-score on the held-out test set of 0.21, slightly behind the Deep MLP with Tabular Features score of 0.23. Both models outperform the Random Forest with Tabular Features at 0.15. We believe that LIFT is a very exciting direction for machine learning in healthcare applications because text-based inputs enables us to take advantage of recent advances in Transfer Learning and Retrieval-Augmented Learning. This study illustrates the effectiveness of converting tabular EHRs to text inputs and utilizing transformer neural networks for prediction. © 2022 IEEE.
ABSTRACT
Objective: To evaluate the behavior of the viruses responsible for acute respiratory infections before (2016-2019) and after (2020-2021) the start of the circulation of the SARS-CoV-2 virus in pediatric patients treated at a reference center from Barranquilla, Colombia. Materials: and methods: An observational descriptive study was carried out, data were obtained reviewing the influenza-like illness and severe acute respiratory infection database in the pediatric population of the sentinel surveillance reference center in the district of Barranquilla during the years 2016 - 2021, applying inclusion and exclusion criteria. Results: During 2016-2019, the average age of individuals was 1.3 (±1.7) years, during 2021 it was 2.3 (±3.5) years. The distribution by sex was similar, predominantly male. August and February were the months with the highest record of symptoms for 2016-2019 and 2021, respectively, the most frequent being cough, fever, shortness of breath, and diarrhea. By 2021 there was higher use of antibiotics and antivirals reported than in 2016-2019. Most patients tested negative for viral detection. When comparing the percentage of viruses detected by age group and years of detection, positivity was lower in 2021 by every age group, and respiratory Syncytial Virus (RSV) was the most frequently detected. Conclusions: : There was less virus positivity in viral detection tests in the pediatric population during 2021. RSV persists as the main etiology affecting this population, especially infants. The use of antibiotic therapy in viral infections continues to be a problematic practice in their management. Sentinel surveillance can be strengthened throughout the country.