Improved machine learning performances with transfer learning to predicting need for hospitalization in arboviral infections against the small dataset

Abstract

The prediction of hospital patients and outpatients with suspected arboviral infection individuals in research-limited settings of the urban areas is defined as a challenging process for clinicians. Dengue, Chikungunya, and Zika arboviruses have gained attention in recent years because of the high prevalence in the society and financial burden of major global health systems. In this study, we proposed a machine learning algorithm based prediction model over retrospective medical records, which are named as SISA (the Severity Index for Suspected Arbovirus) and SISAL (the Severity Index for Suspected Arbovirus with Laboratory) datasets. Therefore, we aim to inform the clinicians about the use of machine learning with transfer learning success for diagnosis and comprehensive comparison of the classification performances over the SISA/SISAL datasets in the resource-limited settings that may cause to the small datasets of arboviral infection. In this study, Convolutional Neural Network and Long Short-Term Memory have achieved 100% accuracy and 1 of area under the curve (AUC) score, Fully Connected Deep Network has provided 92.86% accuracy and 0.969 AUC score in the SISAL dataset with transfer learning. Moreover, 98.73% accuracy and 0.988 AUC score were obtained by Convolutional Neural Network and Long Short-Term Memory for the SISA dataset. Furthermore, Linear Discriminant Analysis (shallow algorithm) has provided reaching up to 96.43% accuracy. Notably, deep learning based models have achieved improved performances compared to the previously reported study.

Prognostic Value of Machine Learning in Patients with Acute Myocardial Infarction

“…AUC is an essential indicator to evaluate the performance of ML models, which could be affected by various factors. Generally, the original dataset is a decisive factor for the value of AUC, and a dataset of high quality and a larger sample size guarantees an authoritative AUC, otherwise missing values or unbalanced data may lead to a lower AUC [39,40]. Additionally, factors resulting in false negative errors also produce a lower AUC value, especially in predictive studies; for example, a shorter follow-up time means a lower probability for a model to learn positive events and thereby causes data unbalance and false negative errors [41].…”

Assessment of the Risk of Severe Dengue Using Intrahost Viral Population in Dengue Virus Serotype 2 Patients via Machine Learning

“…Several reports have demonstrated potential arbovirus hospitalization based on various risk factors by utilizing ML algorithms ( Sippy et al., 2020 ; Ozer et al., 2021 ). Our previous study developed a severe dengue prognosis model for rapid triage using demographic information and dengue antigen/antibody rapid test results from dengue patients ( Huang S. W. et al., 2020 ); however, 75% (18 in 24 false cases) of anti-dengue antibody negative cases were incorrectly discriminated.…”

CovXR: Automated Detection of COVID-19 Pneumonia in Chest X-Rays through Machine Learning

“…The datasets contain images which are usually unrelated to the transfer learning task. However, since they reuse the weights of a complex model, transfer learning models are able to achieve high accuracy with limited amounts of data [13]. This approach offers a better starting point for machine learning tasks and thus requires a shorter amount of time to train a model [14].…”