Age Determination Of Anopheles Mosquitoes Using Infrared Spectroscopy & Artificial Neural Networks

Here we collect the spectral data of Anopheles Gambiae mosquitos in the age range 1 - 17 days, with 2 day intervals. The data is collected using a Fourier Transform Infrared Spectrometer (FTIR) in the mid infrared region fitted with an Attenuated Total Reflectance (ATR) unit Peak heights from the spectra are analysed using an artificial neural network algorithm which learns to predict the age of the mosquitos. The network was first trained using a set of peak height data using the batch backpropogation algorithm and then validated using a new set of data. The validation involves calculating the difference between predicted and actual mosquito age. The Mean Squared Error was calculated to be 2%, which corresponds to a standard deviation of +/- 2 days. The results mean a big step forward in the speed at which mosquito age can be calculated in the lab, and the field; the spectrometer reading takes around two minutes to complete and the neural network calculation time is negligible. The equipment is modestly priced, but can be run on a table top with mains power and the sample preparation takes around one minute. Currently the best method of age determination of mosquitos involves ovary dissection which is time consuming and laboreous work, so this research has the potential for impact in the field, and further research into other species and types of data.