Knowledge Graphs Using Ontological Enrichment for Understanding Contributing Factors of Sleep Quality

In this research, Latent Dirichlet Allocation models for summarizing text information have been used to describe the factors contributing to Sleep Quality as psychological construct, in form of ontologically enriched knowledge-graphs. Data-mining on text from research articles from PubMed, internet articles, sleep questionnaires, and curated text passages from sleep encyclopedia as corpus for the text-models was performed. Natural Language Processing Toolkit with SciPy in Python were employed to perform the analyses, and Latent Dirichlet Allocation models were trained on the obtained datasets. The obtained word relationships were further visualized using network approaches, where the probabilities of the enriched terms were taken, and ontological enrichment was performed using six ontological datasets from the National Center for Biotechnology Information ontologies. The complete enriched and combined Latent Dirichlet Allocation models were used to make inferences about potential contributing factors. Furthermore, an experimental investigation of sleep and microsleep was performed. Microsleep is often remained from investigations for factors affecting sleep quality. The research results elaborate that microsleep frequency range is higher than sleep frequency in alpha and theta bands. This work elaborates the importance of ontologies in embeddings during text analyses, and explores the potential of target information in psychometrics. The obtained results can be employed to develop psychometric instruments measuring Sleep Quality, including microsleep as a cognitive factor.