Effects of sample preservation methods and duration of storage on the performance of mid-infrared spectroscopy for predicting the age of malaria vectors

Abstract

Background: Monitoring the biological attributes of mosquitoes is critical for understanding pathogen transmission and estimating the impacts of vector control interventions on the survival of vector species. Infrared spectroscopy and machine learning techniques are increasingly being tested for this purpose and have been proven to accurately predict the age, species, blood-meal sources, and pathogen infections in Anopheles and Aedes mosquitoes. However, as these techniques are still in early-stage implementation, there are no standardized procedures for handling samples prior to the infrared scanning. This study investigated the effects of different preservation methods and storage duration on the performance of mid-infrared spectroscopy for age-grading females of the malaria vector, Anopheles arabiensis.

Methods: Laboratory-reared An. arabiensis (N = 3681) were collected at 5 and 17 days post-emergence, killed with ethanol, and then preserved using silica desiccant at 5 °C, freezing at - 20 °C, or absolute ethanol at room temperature. For each preservation method, the mosquitoes were divided into three groups, stored for 1, 4, or 8 weeks, and then scanned using a mid-infrared spectrometer. Supervised machine learning classifiers were trained with the infrared spectra, and the support vector machine (SVM) emerged as the best model for predicting the mosquito ages.

Results: The model trained using silica-preserved mosquitoes achieved 95% accuracy when predicting the ages of other silica-preserved mosquitoes, but declined to 72% and 66% when age-classifying mosquitoes preserved using ethanol and freezing, respectively. Prediction accuracies of models trained on samples preserved in ethanol and freezing also reduced when these models were applied to samples preserved by other methods. Similarly, models trained on 1-week stored samples had declining accuracies of 97%, 83%, and 72% when predicting the ages of mosquitoes stored for 1, 4, or 8 weeks respectively.

Conclusions: When using mid-infrared spectroscopy and supervised machine learning to age-grade mosquitoes, the highest accuracies are achieved when the training and test samples are preserved in the same way and stored for similar durations. However, when the test and training samples were handled differently, the classification accuracies declined significantly. Protocols for infrared-based entomological studies should therefore emphasize standardized sample-handling procedures and possibly additional statistical procedures such as transfer learning for greater accuracy.

Keywords: Age-grading; An.arabiensis; Machine learning and infrared spectroscopy; Malaria; Sample handling; Vector control.

Fig. 1

A Mosquitoes collected in disposable cups ready to be killed. B Mosquitoes anesthetized and killed with ethanol. C Mosquito samples being packed in 2-ml Eppendorf tubes ready to be stored for different durations. D Mosquitoes placed on paper towels to allow total evaporation of liquid before scanning

Fig. 2

A Evaluation of different machine learning classifiers for predicting age for mosquito samples preserved in silica gel. The other three panels show confusion matrices with mosquito age predictions from an SVM classifier trained with silica-preserved mosquitoes and used to evaluate samples preserved in B silica gel, C ethanol, and D freezing

Fig. 3

Bar plots showing the declines in classification accuracies when test and training datasets are handled similarly or differently. Here, the SVM models are trained with mid-infrared spectra of mosquitoes preserved using silica (A), ethanol (B), or freezing (C) and then used to predict age classes of samples preserved by one of the three methods. The figure also shows results of the SVM models trained with mid-infrared spectra of mosquitoes stored for 1 week (D), 4 weeks (E), or 8 weeks (F) and then used to predict ages of samples stored for either of the three durations. Reference samples are marked with asterisks. In all cases, the classification accuracy was highest when the training and test samples were handled the same way

Fig. 4

A Evaluation of different machine learning classifiers for predicting age of mosquito samples stored for 1 week. The other three panels show confusion matrices with prediction of mosquito ages from an SVM classifier trained with 1-week samples and used to evaluate samples stored for 1 week (B), 4 weeks (C), and 8 weeks (D)

Fig. 5

Confusion matrices showing prediction accuracies of mosquito ages from a standard SVM classifier trained with samples preserved in silica gel, stored for 1 week, and then used to predict age classes of test samples handled the same way or differently. Silica-preserved samples are shown in panels A, B, C; ethanol-preserved samples on panels D, E, F and frozen samples on panels D, H, I