Project Title: Artificial Intelligence and Infrared Spectroscopy to Accelerate Malaria Vector Control
Project Description: This project will develop a novel technology to quantify the efficacy of any control intervention targeted at malaria mosquitoes, by combining artificial intelligence and infrared-spectroscopy to obtain real-time information on mosquito populations and their disease transmission potential. However, genetic and ecological factors can affect the composition of mosquito cuticle in unexpected ways, and demographic predictions based on MIRS of laboratory mosquitoes might not accurately estimate species and age in wild mosquitoes. This grant will support... This project will develop a novel technology to quantify the efficacy of any control intervention targeted at malaria mosquitoes, by combining artificial intelligence and infrared-spectroscopy to obtain real-time information on mosquito populations and their disease transmission potential. However, genetic and ecological factors can affect the composition of mosquito cuticle in unexpected ways, and demographic predictions based on MIRS of laboratory mosquitoes might not accurately estimate species and age in wild mosquitoes. This grant will support all the activities and cover some personnel costs. Estimate ageing rates in wild mosquitoes by analyzing ecological and environmental determinants of age and species prediction accuracy of wild mosquitoes. 1. Collecting and increasing training dataset to 50,000 mosquitoes. 2. Using white box machine learning algorithms to select core mosquito MIRs wave numbers that predict age of known mosquitoes. 3. Using selected features to train convolutional neural meet optimize for generalizability. Develop an online platform for real-time analysis of mosquito MIRS data through machine learning. 1. We will build an online application to which users will be able to submit spectra collected in the field and obtain the output of the predictive models in real time. The choice of a web application is motivated by the high degree of accessibility that it affords and the fact that it allows the pre-trained model(s) to exist in a single place and remain up-to-date. Data submitted to the tool will be stored, along with relevant metadata (location of capture, etc.) resulting in a growing dataset that will allow the predictive models to be regularly updated.
Principal Investigator : Fredros Okumu
Department Name : EHES
Time frame: (2019-11-01) - (2021-10-31)
Bill and Melinda Gates Foundation (BMGF) (Normal)
External Collaborating Partners
The University Court of the University of Glasgow