Publication

Abstract : In this study, we challenge disease pairs and risk strata of troponin using the state-of-the-art. Apply k-means clustering method in Machine learning (ML) to multivariate dataset that depicts Lab features and disease classes. A Monte Carlo simulation was performed to select the best number of clusters from convergence for maximizing silhouette coefficient. This stratification discloses new states and finer grained descriptions of trajectories. Troponin were also stratified by risk to evaluate cardiovascular risk more comprehensively. Some have been decoded by feature visualization, and differentiable features amount to pretty much disease classification. Their results were exemplary: they all had a zero-standard deviation with equal mean hamming distance strings, and an ANN model with a 95% accuracy rate on the testing dataset. In particular, this study exemplifies an approach by which novel analytic methods for high-dimensional representations of complex disease patterns may be leveraged to guide clinical decisionmaking.