Deep Learning Evaluation of Roadside Pollution Effects on Small Mammal Health

Abstract

Small mammals are essential elements in the ecosystem of land and act as indicators of environmental wellbeing owing to their sensitivity to alterations in their habitat. With increased development and movement around roadsides, there has been an increase in pollutants such as particles, gases, and heavy metals, which might harm the health of small mammals in terms of physiological and behavioral effects. Methods used traditionally for studying wildlife include field observations and laboratory tests, and this limits their ability to establish the intricate relationship between the environment and biological impacts. In this paper, a method for the evaluation of roadside pollution impact on small mammal health using deep learning is proposed. Parameters that will be considered include PM2.5, PM10, carbon monoxide, nitrogen oxides, heavy metal concentration, traffic volume, vegetation status, body condition, activity level, and population presence.  The collected data are pre-processed and analyzed by a deep learning algorithm that forecasts health problems resulting from pollution exposure. The performance of the deep learning model is assessed through accuracy, precision, recall, and F1 score measures. It has been found that particle matter concentration, heavy metal contamination, and vehicle traffic play an important role in the health conditions of small mammals. In our research, the proposed deep learning model showed better prediction performance with an accuracy of 95.3%. It proves the effectiveness of the suggested model for forecasting relations between pollution exposure along roadsides and health reactions of wildlife. Future studies will be able to integrate real-time environmental monitoring along with biodiversity tracking.