JMIR Publications

ABSTRACT

Background:

Leptospirosis, a zoonotic disease caused by Leptospira bacteria, continues to pose significant public health risks, particularly in tropical and subtropical regions.

Objective:

This systematic review aimed to evaluate the application of Machine Learning (ML) and Deep Learning (DL) techniques in predicting and diagnosing leptospirosis, focusing on the most commonly used algorithms, validation methods, data types, and performance metrics.

Methods:

Following PRISMA guidelines, we conducted a comprehensive review of studies applying ML and DL models for leptospirosis detection and prediction, examining algorithm performance, data sources, and validation approaches.

Results:

The review found that algorithms such as Support Vector Machines (SVM), Artificial Neural Networks (ANN), Decision Trees (DT), and Convolutional Neural Networks (CNN) were frequently employed, with some models achieving accuracy rates of up to 98%. Despite these advancements, public datasets were underutilized, with private datasets from hospitals and medical records serving as the primary sources of data.

Conclusions:

Key challenges identified in this review include the limited application of transfer learning, small dataset sizes, and inconsistent use of advanced performance metrics like Area Under the Curve (AUC). Additionally, the review highlighted the absence of ensemble methods and data augmentation techniques that could enhance the robustness and generalizability of models. While machine learning (ML) and deep learning (DL) techniques show promise for improving leptospirosis prediction and diagnosis, future research should prioritize the use of larger, more diverse datasets, the adoption of transfer learning strategies, and the integration of advanced ensemble and validation techniques to further strengthen model accuracy and generalization. Clinical Trial: None!