Neuronal nets and logistic regression analysis provide improved prediction of infective endocarditis compared to the modified Duke Score: a post-hoc analysis of the prospective PRO-ENDOCARDITIS study

Abstract

Abstract Introduction The modified Duke score is the currently recommended diagnostic algorithm in suspected infective endocarditis (IE). The categorization in major and minor criteria enables an easy clinical application, but may not optimally utilize individual patient's information. In contrast, detailed statistical evaluation of multiple characteristics using artificial intelligence and logistic regression report improved prediction of various cardiovascular diseases over conventional clinical strategies. We tested the hypothesis that neuronal nets and logistic regression analysis would provide improved prediction of IE as compared to the modified Duke score. Methods This post-hoc evaluation of the prospective observational PRO-ENDOCARDITIS study was conducted at the West German Heart and Vascular center between December 2017 and May 2019 and includes 261 patients. Duke criteria and clinical characteristics were prospectively collected. Transesophageal echocardiography (TEE) imaging was evaluated by a blinded cardiologist at a central core-lab. IE as primary endpoint was adjudicated by an independent clinical endpoint committee. The database was divided into a training (70%) and validation cohort (30%). We compared the value of the Duke score, neuronal nets and logistic regression analysis for prediction of the primary endpoint. Results The mean age was 60.1±16.1 years, 37.2% were female. In 47 cases, IE was present. The modified Duke score achieved an AUC of 0.863 in the training cohort and 0.913 within the validation cohort. The logistic regression and the neural net exceeded the predictive value in both cohorts (training cohort: 0.992 and 0.986; validation cohort: 0.964, 0.957; for logistic regression and neuronal nets, respectively, Figure 1). Without the use of TEE, the remaining Duke criteria only poorly predicted IE (training cohort: 0.771, 0.951 and 0.938; validation cohort: 0.835, 0.862 and 0.780, for the Duke score, logistic regression and neuronal nets, respectively). Discussion Logistic regression analysis and neuronal nets provide improved prediction of IE as compared to the clinically established modified Duke score. Further studies on larger databases are needed to confirm our results and provide algorithms for clinical routine. Funding Acknowledgement Type of funding sources: None.