Predicting dengue transmission rates by comparing different machine learning models with vector indices and meteorological data.

Machine learning algorithms (ML) are receiving a lot of attention in the development of predictive models for monitoring dengue transmission rates. Previous work has focused only on specific weather variables and algorithms, and there is still a need for a model that uses more variables and algorithms that have higher performance. In this study, we use vector indices and meteorological data as predictors to develop the ML models. We trained and validated seven ML algorithms, including an ensemble ML method, and compared their performance using the receiver operating characteristic (ROC) with the area under the curve (AUC), accuracy and F1 score. Our results show that an ensemble ML such as XG Boost, AdaBoost and Random Forest perform better than the logistics regression, Naïve Bayens, decision tree, and support vector machine (SVM), with XGBoost having the highest AUC, accuracy and F1 score. Analysis of the importance of the variables showed that the container index was the least important. By removing this variable, the ML models improved their performance by at least 6% in AUC and F1 score. Our result provides a framework for future studies on the use of predictive models in the development of an early warning system.

Community-based mosquito surveillance: an automatic mosquito-on-human-skin recognition system with a deep learning algorithm.

BACKGROUND: Public community engagement is crucial for mosquito surveillance programs. To support community participation, one of the approaches is assisting the public in recognizing the mosquitoes that carry pathogens. Therefore, this study aims to build an automatic recognition system to identify mosquitos at the public community level. We construct a customized image dataset consisting of three mosquito species in either damaged or un-damaged body conditions. To distinguish the mosquito in harsh conditions, we explore two state-of-the-art deep learning (DL) architectures: (i) a freezing convolutional base, with partial trainable weights, and (ii) training the entire model with most of the trainable weights. We project a weighted feature map on different layers of the model to visualize the morphological region used by the model in classification and compared it with the morphological key used by the expert. RESULT: It was found that the model with architecture two and the Adam optimizer achieves at least 98% accuracy in mosquito and conditions identification and when implemented on an independent dataset, the Xception model generalizes the best result with an accuracy of 0.7775 and 0.795 precision. Moreover, most of the morphological regions used by the model are able to match those of the human expert. CONCLUSION: We report a customized DL model for performing pest mosquito taxonomy identification, and through visualization, some regions using computers to discriminate mosquito species could be adopted later in systematic identification. © 2022 Society of Chemical Industry.

Implementation of a deep learning model for automated classification of Aedes aegypti (Linnaeus) and Aedes albopictus (Skuse) in real time.

Classification of Aedes aegypti (Linnaeus) and Aedes albopictus (Skuse) by humans remains challenging. We proposed a highly accessible method to develop a deep learning (DL) model and implement the model for mosquito image classification by using hardware that could regulate the development process. In particular, we constructed a dataset with 4120 images of Aedes mosquitoes that were older than 12 days old and had common morphological features that disappeared, and we illustrated how to set up supervised deep convolutional neural networks (DCNNs) with hyperparameter adjustment. The model application was first conducted by deploying the model externally in real time on three different generations of mosquitoes, and the accuracy was compared with human expert performance. Our results showed that both the learning rate and epochs significantly affected the accuracy, and the best-performing hyperparameters achieved an accuracy of more than 98% at classifying mosquitoes, which showed no significant difference from human-level performance. We demonstrated the feasibility of the method to construct a model with the DCNN when deployed externally on mosquitoes in real time.