Rapid age-grading and species identification of natural mosquitoes for malaria surveillance

Abstract

The malaria parasite, which is transmitted by several Anopheles mosquito species, requires more time to reach its human-transmissible stage than the average lifespan of mosquito vectors. Monitoring the species-specific age structure of mosquito populations is critical to evaluating the impact of vector control interventions on malaria risk. We present a rapid, cost-effective surveillance method based on deep learning of mid-infrared spectra of mosquito cuticle that simultaneously identifies the species and age class of three main malaria vectors in natural populations. Using spectra from over 40, 000 ecologically and genetically diverse An. gambiae, An. arabiensis, and An. coluzzii females, we develop a deep transfer learning model that learns and predicts the age of new wild populations in Tanzania and Burkina Faso with minimal sampling effort. Additionally, the model is able to detect the impact of simulated control interventions on mosquito populations, measured as a shift in their age structures. In the future, we anticipate our method can be applied to other arthropod vector-borne diseases.

Mosquito metabolism shapes life-history strategies ofPlasmodiumparasites

“…The cycle of seeking, digesting, and egg-laying is continued during their entire life span, and mosquito death is modeled as an age-dependent function. With these parameters we could model stable mosquito population dynamics (Figure 1A) that matches the age-distribution of mosquitoes observed in field settings 25 (Figure S1).…”

“…However, the age distribution of our simulated mosquito populations is remarkably similar to wild An . coluzzi mosquitoes, as recently measured with novel age-grading methods 25 . Like in our simulations, these natural mosquito populations have a very small proportion of old individuals (>17 days) that feed frequently (> 4 blood meals).…”

Using transfer learning and dimensionality reduction techniques to improve generalisability of machine-learning predictions of mosquito ages from mid-infrared spectra

“…This result is consistent with Siria et al . [10], where CNN underperformed as a result of the difference in data distribution between the training and evaluation data driven by non-genetic factors such as ecological variation. When near-infrared spectroscopy was used to predict the age of Anopheles mosquitoes reared from wild populations, a similar limitation was reported [8,29].…”

“…We and others have demonstrated that spectroscopic analysis of mosquitoes using near infrared (12500 – 4000 cm -1 ) or mid-infrared (MIR) (4000 – 400 cm -1 ) frequencies can identify key biochemical signals that vary with age [8,9]. These methods, when combined with specific machine learning (ML) techniques, allow for rapid estimation of mosquito ages [9,10].…”

“…These methods are known to be computationally efficient and to improve generalisability [11]. Deep learning (DL) techniques such as convolutional neural networks (CNN) have recently been used on large spectra data [10], improving generalisability through transfer learning (i.e., updating a pre-trained model with a small amount of new data from a different target population). However, when trained on large datasets, such techniques remain computationally expensive and may necessitate repeated sampling of hundreds of mosquitoes from different populations and environments to allow successful generalisability.…”