Design of a FAIR digital data health infrastructure in Africa for COVID-19 reporting and research.

The limited volume of COVID-19 data from Africa raises concerns for global genome research, which requires a diversity of genotypes for accurate disease prediction, including on the provenance of the new SARS-CoV-2 mutations. The Virus Outbreak Data Network (VODAN)-Africa studied the possibility of increasing the production of clinical data, finding concerns about data ownership, and the limited use of health data for quality treatment at point of care. To address this, VODAN Africa developed an architecture to record clinical health data and research data collected on the incidence of COVID-19, producing these as human- and machine-readable data objects in a distributed architecture of locally governed, linked, human- and machine-readable data. This architecture supports analytics at the point of care and-through data visiting, across facilities-for generic analytics. An algorithm was run across FAIR Data Points to visit the distributed data and produce aggregate findings. The FAIR data architecture is deployed in Uganda, Ethiopia, Liberia, Nigeria, Kenya, Somalia, Tanzania, Zimbabwe, and Tunisia.

Optimal Length of Heart Rate Variability Data and Forecasting Time for Ventricular Fibrillation Prediction Using Machine Learning.

Ventricular fibrillation (VF) is a cardiovascular disease that is one of the major causes of mortality worldwide, according to the World Health Organization. Heart rate variability (HRV) is a biomarker that is used for detecting and predicting life-threatening arrhythmias. Predicting the occurrence of VF in advance is important for saving patients from sudden death. We extracted features from seven HRV data lengths to predict the onset of VF before nine different forecast times and observed the prediction accuracies. By using only five features, an artificial neural network classifier was trained and validated based on 10-fold cross-validation. Maximum prediction accuracies of 88.18% and 88.64% were observed at HRV data lengths of 10 and 20 s, respectively, at a forecast time of 0 s. The worst prediction accuracy was recorded at an HRV data length of 70 s and a forecast time of 80 s. Our results showed that features extracted from HRV signals near the VF onset could yield relatively high VF prediction accuracies.

Author Correction: Application of a convolutional neural network for predicting the occurrence of ventricular tachyarrhythmia using heart rate variability features.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Application of a convolutional neural network for predicting the occurrence of ventricular tachyarrhythmia using heart rate variability features.

Predicting the occurrence of ventricular tachyarrhythmia (VTA) in advance is a matter of utmost importance for saving the lives of cardiac arrhythmia patients. Machine learning algorithms have been used to predict the occurrence of imminent VTA. In this study, we used a one-dimensional convolutional neural network (1-D CNN) to extract features from heart rate variability (HRV), thereby to predict the onset of VTA. We also compared the prediction performance of our CNN with other machine leaning (ML) algorithms such as an artificial neural network (ANN), a support vector machine (SVM), and a k-nearest neighbor (KNN), which used 11 HRV features extracted using traditional methods. The proposed CNN achieved relatively higher prediction accuracy of 84.6%, while the ANN, SVM, and KNN algorithms obtained prediction accuracies of 73.5%, 67.9%, and 65.9% using 11 HRV features, respectively. Our result showed that the proposed 1-D CNN could improve VTA prediction accuracy by integrating the data cleaning, preprocessing, feature extraction, and prediction.

Machine Learning Approach to Predict Ventricular Fibrillation Based on QRS Complex Shape.

Early prediction of the occurrence of ventricular tachyarrhythmia (VTA) has a potential to save patients' lives. VTA includes ventricular tachycardia (VT) and ventricular fibrillation (VF). Several studies have achieved promising performances in predicting VT and VF using traditional heart rate variability (HRV) features. However, as VTA is a life-threatening heart condition, its prediction performance requires further improvement. To improve the performance of predicting VF, we used the QRS complex shape features, and traditional HRV features were also used for comparison. We extracted features from 120-s-long HRV and electrocardiogram (ECG) signals (QRS complex signed area and R-peak amplitude) to predict the VF onset 30 s before its occurrence. Two artificial neural network (ANN) classifiers were trained and tested with two feature sets derived from HRV and the QRS complex shape based on a 10-fold cross-validation. The prediction accuracy estimated using 11 HRV features was 72%, while that estimated using four QRS complex shape features yielded a high prediction accuracy of 98.6%. The QRS complex shape could play a significant role in performance improvement of predicting the occurrence of VF. Thus, the results of our study can be considered by the researchers who are developing an application for an implantable cardiac defibrillator (ICD) when to begin ventricular defibrillation.