Diagnosis of diabetes and hypertension: A performance comparison between transcriptome data and clinical data

Abstract

Diabetes and hypertension are two related diseases common among most of the people all over the word. The impact of these diseases can elevate the risk of developing additional health issues, including cardiovascular disease, kidney disease, and other related conditions. Numerous research initiatives are underway to uncover the underlying mechanisms of these diseases. This study compares the roles of clinical data and transcriptome data in the diagnosis of patients with diabetes or hypertension. This study utilizes two distinct datasets, each containing unique clinical features along with a third dataset that includes transcriptome characteristics, all analyzed through machine learning algorithms. In both diseases, there is a marked difference in the accuracies of the models based on various clinical features. This highlights the importance of selecting appropriate clinical features to develop an optimal diagnostic model for these conditions. In comparing the best clinical model with the transcriptome model for diagnosing diabetic patients, it has been found that the transcriptome data yields superior results. Conversely, for diagnosing hypertension patients, the clinical data proves to be more effective. Hence, identifying the appropriate set of clinical features, clinical data could become more effective for diagnosing both diabetes and hypertension.