Unsupervised Cardiovascular Risk Factors Clustering: Towards an Expert Recommendation System for Personalized Nutrition Therapy

High blood pressure with its high prevalence is associated with strong evidence for causing cardiovascular disease (CVD) among other risk factors. Between 1975 and 2015, The World Health Organisation recorded an increase of about 66% in the occurrence of high blood pressure, hence further threatening a reduction in the average life expectancy if left unchecked. This paper analyses cardiovascular risk prediction models and proposes an expert recommendation system for personalised nutrition therapy. The methodology of this study adopts a simulation approach and aims at establishing the feasibility of an ongoing project. Hence, data to the study was generated by imposing a reverse engineering approach on the Framingham’s criteria/model using Monte Carlo simulation. An unsupervised learning and correlation hunting of 10,000 patients’ cohorts revealed the defect of the Framingham’s model, as it failed to establish relationships between other risk factors, save the age factor. Hence, making it an inappropriate model for tackling CVD. To provide efficient prediction of CVD and advance the growing field of healthcare informatics, an expert system framework with 4 components is proposed. The implementation workflow begins with a collection of primary CVD risk factors using the WHO stepwise approach for non-communicable disease surveillance, to offer an effectual means of developing data collection instrument for patients’ cohorts data collection. The collected data is then processed into the CVD risk factors database. A fuzzy inference system is deployed to infer appropriate CVD risk scores by formulating a mapping from the input space to output crisp value. These scores are used to label the CVD dataset, for supervised learning and prediction. The predicted CVD risk level is finally matched with suitable diet plan in the nutrition database crafted by dieticians, to provide decision support on personalised nutrition, for persons living with CVD