WebGIS-Based Real-Time Surveillance and Response System for Vector-Borne Infectious Diseases

Abstract

The diseases transmitted through vectors such as mosquitoes are named vector-borne diseases (VBDs), such as malaria, dengue, and leishmaniasis. Malaria spreads by a vector named Anopheles mosquitos. Dengue is transmitted through the bite of the female vector Aedes aegypti or Aedes albopictus mosquito. The female Phlebotomine sandfly is the vector that transmits leishmaniasis. The best way to control VBDs is to identify breeding sites for their vectors. This can be efficiently accomplished by the Geographical Information System (GIS). The objective was to find the relation between climatic factors (temperature, humidity, and precipitation) to identify breeding sites for these vectors. Our data contained imbalance classes, so data oversampling of different sizes was created. The machine learning models used were Light Gradient Boosting Machine, Random Forest, Decision Tree, Support Vector Machine, and Multi-Layer Perceptron for model training. Their results were compared and analyzed to select the best model for disease prediction in Punjab, Pakistan. Random Forest was the selected model with 93.97% accuracy. Accuracy was measured using an F score, precision, or recall. Temperature, precipitation, and specific humidity significantly affect the spread of dengue, malaria, and leishmaniasis. A user-friendly web-based GIS platform was also developed for concerned citizens and policymakers.

Advancing vector-borne disease prediction through functional classifier integration: A novel approach for enhanced modeling

“…In the context of vector-borne diseases, the MLP model processes input data related to environmental conditions, demographic factors, and historical disease occurrences through the input layer. The hidden layers, characterized by nodes employing weighted connections and activation functions, enable the network to discern intricate patterns and non-linear relationships within the data (Javaid et al 2023;Kofidou et al 2021). The output layer provides predictions or classifications, such as the likelihood of vectorborne disease occurrence in a specific region or population.…”

“…Activation functions in the output layer depend on the nature of the prediction task, employing sigmoid functions for binary classification and softmax functions for multiclass classification. Training an MLP involves adjusting weights and biases to minimize the difference between predicted outputs and actual outcomes, typically utilizing optimization algorithms like gradient descent (Javaid et al 2023). MLP models excel in capturing complex patterns and relationships within datasets, especially in scenarios where traditional linear models may be inadequate (Tito et al 2023).…”

Automatic mapping of high-risk urban areas for Aedes aegypti infestation based on building facade image analysis

“…aegypti infestation levels [34][35][36][37], it is common to use field survey data related to socioeconomic status [38,39], such as income, education, and crowding. There are also instances of leveraging environmental information, such as temperature [40,41], humidity, or precipitation [42]. We are especially interested, however, in the domain of images, which has received growing attention from the research community.…”

Automatic mapping of high-risk urban areas forAedes aegyptiinfestation based on building facade image analysis

“…aegypti infestation levels [34][35][36][37], it is common to use field survey data related to socioeconomic status [38,39], such as income, education, and crowding. There are also instances of leveraging environmental information, such as temperature [40,41], humidity, or precipitation [42]. We are especially interested, however, in the domain of images, which has received growing attention from the research community.…”