Semantic Interoperability and Predictive Analytics in the Internet of Things for Smart Healthcare

Abstract

The Internet of Things (IoT), which aspires to establish a network of Internet-capable objects newlineto support a smart world, is an interesting study subject among the pool of potential research newlineareas of the contemporary technological period. To allow the intelligent enabled world, a newlinehuge number of devices are installed in all conceivable geographical locations to collect data newlinerequired for smart applications in almost all sectors. The amount and variety of data acquired newlinefrom this huge pool of devices will be tremendous. newlineKeeping in cognizance of the huge amount of data generated by the devices (sensors, newlineactuators, gateways, mobiles, smart objects etc.), deployed in our day to day lives which is newlineheterogeneous in nature in terms of formats, types and domains. This leads to an issue for newlinemachines to analyze, understand, and interpret termed semantic interoperability. To newlineovercome the issue of semantic interoperability occurred as a result of heterogeneous devices newlineand data, the possible solutions can be ontology, middleware, proxy gateways and knowledge newlinegraph. Ontology and knowledge graphs have been proven as effective solutions for semantic newlineinteroperability in the Internet of Things. newlineThe research aims at: (i) A detailed study of semantic interoperability approaches employed newlinein the Internet of Things, (ii) Dataset collection and Pre-processing, (iii) Resource newlineDescription Framework Mapping Language (RML) based approach for the development of newlineknowledge base model for smart healthcare system, (iv) Retrieval of information from the newlineknowledge graph using Competency Questions and SPARQL Queries, (v) Validation of newlineKnowledge based model, (vi) Analytics of Data using Machine Learning techniques, and (vii) newlineUtilizing the hyperparameter optimization or tuning and ensemble mechanism for newlinedevelopment of efficient prediction model and (viii) application of paired t test for prediction newlinemodels validation. newlineThis thesis has proposed a RDF Mapping Language (RML) based lightweight middleware newlinemechanism for the transformation of