Author + information
- Chen Peisong,
- Feng Pinning,
- Xu Hongxu and
- Liu Min
Essential hypertension associated target organ damage (TOD) is a grading complex diseases. Distinguishing the patients with high risk of TOD is meaningful for clinical practice. In this study, we aimed to establish diagnostic models in predicting TOD based on laboratory data.
Totally 377 consecutive patients who presented to the inpatient department of the First Affiliated Hospital of Sun Yat-sen University with a primary diagnosis of essential hypertension between January 2014 and February 2016 were enrolled in our study, including 114 patients with TOD and 263 patients without TOD. 19 laboratory indexes selected from previously published literatures were evaluated in all subjects. Both Back Propagation ANNs (BP) and Logistic regression models were established for predicting the disease and their diagnostic performance were compared by receiver operating curve.
Three out of 19 indexes (blood urea nitrogen, homocysteine, alkaline phosphatase) were closely related with TOD in essential hypertension patients and finally enrolled in models. The consistent rate, sensitivity and specificity of BP ANNs were 80.30%, 74.5% and 84.21% respectively. For the Logistic regression model, the consistent rate, sensitivity and specificity was 78.2%, 75.1% and 78.3% respectively. The area under the ROC curve of the BP ANNs and Logistic regression model was 0.887 and 0.823.
Both BP ANNs and Logistic regression model were promising for the prediction of TOD by combining blood urea nitrogen, homocysteine and alkaline phosphatase. However, BP ANNs model showed better performance than Logistic regression model in predicting the disease.