Fusion of Earth Observation Data and Sociodemographic census for prediction of Anxiety using Machine Learning Algorithms – A Case Study

The impact of nature on human health is of great concern in the twentieth century, especially, pre and post covid duration. Consequently, health information is important for policymakers, medical institutions, and governments for the promotion of social well-being. In this regard, the Earth Observation (EO) data, especially from Sentinel-2 plays a prominent role in various conditions of climate change, floods, and infrastructure development. This information may not represent the entire set of population. So, to maintain the correctness of the remote sensing data, based on the geographical division, various sociodemographic measures are derived from different surveys which are relevant to health, social status, and ethnicity. This work mainly focuses on the potential of the integrated dataset to improve the machine-learning (ML) algorithms in deriving the health outcomes of diverse regions such as anxiety. The various outcomes of the target variables are divided into various categories such as diabetes and hypertension, depression and anxiety, asthma, opioids, and so on. Further, with the outcomes, prescriptions can be anticipated with the combination of the Drug Bank dataset. The second objective is to understand the factors that influence the health outcomes of individuals using multiple data exploratory analysis for predicting anxiety. Finally, this study creates a dataset that addresses the prospective challenges in the field of EO and Artificial Intelligence. The results and the dataset are available in https://github.com/SudheerReddyBandi/ICIMMI.