The COVID-19 pandemic has affected almost every country causing devastating economic and social disruption and stretching healthcare systems to the limit. Furthermore, while being the current gold standard, existing test methods including NAAT (Nucleic Acid Amplification Tests), clinical analysis of chest CT (Computer Tomography) scan images, and blood test results, require in-person visits to a hospital which is not an adequate way to control such a highly contagious pandemic. Therefore, top priority must be given, among other things, to enlisting recent and adequate technologies to reduce the adverse impact of this pandemic. Modern smartphones possess a rich variety of embedded MEMS (Micro-Electro-Mechanical-Systems) sensors capable of recording movements, temperature, audio, and video of their carriers. This study leverages the smartphone sensors for the preliminary diagnosis of COVID-19. Deep learning, an important breakthrough in the domain of artificial intelligence in the past decade, has huge potential for extracting apt and appropriate features in healthcare. Motivated from these facts, this paper presents a new framework that leverages advanced machine learning and data analytics techniques for the early detection of coronavirus disease using smartphone embedded sensors. The proposal provides a simple to use and quickly deployable screening tool that can be easily configured with a smartphone. Experimental results indicate that the model can detect positive cases with an overall accuracy of 79% using only the data from the smartphone sensors. This means that the patient can either be isolated or treated immediately to prevent further spread, thereby saving more lives. The proposed approach does not involve any medical tests and is a cost-effective solution that provides robust results.

Natural Language Processing problems generally require the use of pre-trained distributed word representations to be solved with deep learning models. However, distributed representations usually rely on contextual information which prevents them from learning all the important word characteristics. The task of sentiment analysis suffers from such a problem because sentiment information is ignored during the process of learning word embeddings. The performance of sentiment analysis can be affected since two words with similar vectors may have opposite sentiment orientations. The present paper introduces a novel model called Continuous Sentiment Contextualized Vectors (CSCV) to address this problem. The proposed model can learn word sentiment embedding using its surrounding context words. It uses Continuous Bag-of-Words (CBOW) model to deal with the context and sentiment lexicons to identify sentiment. Existing pre-trained vectors are combined then with the obtained sentiment vectors using Principal component analysis (PCA) to enhance their quality. The experiments show that: (1) CSCV vectors can be used to enhance any pre-trained word vectors; (2) The result vectors strongly alleviate the problem of similar words with opposite polarities; (3) The performance of sentiment classification is improved by applying this approach.