326 - Predicting Aedes triseriatus Parity and Age with Mid-Infrared Spectroscopy

Abstract: Knowing the age of a mosquito and whether it has taken a blood meal are critical parameters in the vectorial capacity model for mosquito-borne disease transmission. Many current methods for mosquito age-grading and blood feeding/parity status determination are laborious or require specialized skills often unavailable or undesirable to mosquito control districts. Mid-infrared spectroscopy (MIRS) is a promising, low-cost, and low-labor alternative to other molecular, physiological, or anatomical age- and parity-grading methodologies. Here, we assessed the ability of MIRS to detect the chronological age (i.e., days) of Aedes triseriatus, the primary La Crosse virus vector, using a convolutional neural network. Additionally, we measured physiological age (e.g., parous/nulliparous) using a Naïve Bayes classifier. For chronological age, measured at 6 intervals (1, 2, 7, 14, 21, 28, and 35 days post-emergence), our model estimated age with a mean absolute percentage error of 34.7% and a model fit of R² = 0.92. When using this model to classify age cohorts (< 6 days, 7 to 22 days, 23 days and older) then model age cohort accuracy is 92% (CI95: 85% - 99%), 85% (CI95: 71% - 95%), and 94% (CI95: 81% - 99%), respectively. In a separate experiment, we successfully estimated parity status for Ae. triseriatus with 74% accuracy (95% CI: 63%–84%). The experiments presented here add to the growing body of evidence supporting MIRS as a valuable tool for species identification, chronological age-grading, and physiological age-grading in mosquitoes.