Artificial Intelligence-Based Computer-Aided Diagnosis of Neonatal Sepsis: CurrentAdvances, Challenges and Clinical Opportunities in Nigeria

The mortality rate of neonatal sepsis in Nigeria ranges from 26.7% to 33.3% because of the low accuracy of the available methods of diagnosis, owing to its nonspecific symptoms. Also, the long time required (48 to 72 hours) to complete laboratory tests using common diagnostic tools, such as blood cultures and biomarkers. These challenges often lead to delayed treatment of infected babies and, eventually, high mortality rates. This article presents a review focusing on the potential of Artificial Intelligence-Based Computer-Aided Diagnosis (AI-CAD) systems for early diagnosis and treatment of neonatal sepsis. Specifically, the paper discusses traditional diagnostic methods for neonatal sepsis and their limitations. In addition, artificial intelligence methods such as machine learning, artificial neural networks, Random Forest, XGBoost, Support Vector Machine, and Long Short-Term Memory were studied as potential diagnostic tools for neonatal sepsis prediction. The paper presents a conceptual proposal for an AI-CAD system and the evaluation metrics for it. Such metrics include the F1-Score, area under the precision-recall curve (AUPRC), and the area under the receiver operating characteristic curve (AUROC), which could distinguish true positives from false positives at a baseline of 0.5. The review highlights the capabilities of AI-CAD systems to revolutionize neonatal healthcare in developing countries, improving diagnostics and reducing infant mortality rates. Nonetheless, the successful incorporation of such technologies requires addressing several important considerations, including the development of an electronic medical database, ethical issues, and Medical-Tech (MediTech) professionals’ collaborations.