Identification of medically and forensically relevant flies using a decision treelearning method.

Abstract

Blow flies, flesh flies, and house flies can provide excellent evidence for forensic entomologists and are also essential to the fields of public health, medicine, and animal health. In all questions, the correct identification of fly species is an important initial step. The usual methods based on morphology or even molecular approaches can reach their limits here, especially when dealing with larger numbers of specimens. Since machine learning already plays a major role in many areas of daily life, such as education, business, industry, science, and medicine, applications for the classification of insects have been reported. Here, we applied the decision tree method with wing morphometric data to construct a model for discriminating flies of three families [Calliphoridae, Sarcophagidae, Muscidae] and seven species [Chrysomya megacephala (Fabricius), Chrysomya rufifacies (Macquart), Chrysomya (Ceylonomyia) nigripes Aubertin, Lucilia cuprina (Wiedemann), Hemipyrellia ligurriens (Wiedemann), Musca domestica Linneaus, and Parasarcophaga (Liosarcophaga) dux Thomson]. One hundred percent overall accuracy was obtained at a family level, followed by 83.33% at a species level. The results of this study suggest that non-experts might utilize this identification tool. However, more species and also samples per specimens should be studied to create a model that can be applied to the different fly species in Thailand.