A Knowledge Representation Model Based on Select and Test Algorithm for Diagnosing Breast Cancer

There exist several terminal diseases whose fatality rate escalates with time of which breast cancer is a frontline disease among such. Computer aided systems have also been well researched through the use intelligent algorithms capable of detecting, diagnosing, and proffering treatment for breast cancer. While good research breakthrough has been attained in terms of algorithmic solution towards diagnosis of breast cancer, however, not much has been done to sufficiently model knowledge frameworks for diagnostic algorithms that are knowledge-based. While Select and Test (ST) algorithm have proven relevant for implementing diagnostic systems, through support for reasoning, however the knowledge representation pattern that enables inference of missing or ambiguous data still limits the effectiveness of ST algorithm. This paper therefore proposes a knowledge representation model to systematically model knowledge to aid the performance of ST algorithm. Our proposal is specifically target ed at developing systematic knowledge representation for breast cancer. The approach uses the ontology web language (OWL) to implement the design of the knowledge model proposed.   This study aims at carefully crafting a knowledge model whose implementation seamlessly work with ST algorithm. Furthermore, this study adapted the proposed model into an implementation of ST algorithm an obtained an improved performance compared to the simple knowledge model proposed by the author of ST algorithm. Our knowledge mode resulted in an accuracy gain of 23.5% and obtained and AUC of (0.49, 1.0). This proposed model has therefore shown that combining an inference-oriented knowledge model with an inference-oriented reasoning algorithm improves the performance of computer aided diagnostic (CADx) systems. In future, we intend to enhance the proposed model to support rules. Keywords — Semantic web, ontology, OWL, breast cancer, Select and Test (ST) algorithm, knowledge representation