Cluster Validity Index for Uncertain Data Based on a Probabilistic Distance Measure in Feature Space.

Cluster validity indices (CVIs) for evaluating the result of the optimal number of clusters are critical measures in clustering problems. Most CVIs are designed for typical data-type objects called certain data objects. Certain data objects only have a singular value and include no uncertainty, so they are assumed to be information-abundant in the real world. In this study, new CVIs for uncertain data, based on kernel probabilistic distance measures to calculate the distance between two distributions in feature space, are proposed for uncertain clusters with arbitrary shapes, sub-clusters, and noise in objects. By transforming original uncertain data into kernel spaces, the proposed CVI accurately measures the compactness and separability of a cluster for arbitrary cluster shapes and is robust to noise and outliers in a cluster. The proposed CVI was evaluated for diverse types of simulated and real-life uncertain objects, confirming that the proposed validity indexes in feature space outperform the pre-existing ones in the original space.

Availability of patient classification using clinical data.

OBJECTIVES: The aim of this study is examine the reliability and validity of the patient classification which is based on clinical data with comparing to nurse's check. METHOD: Nurse Experts estimated the content validity of extracting KPCS-1(Korea patient classification system for nurses Version 1) activities score from clinical data in storage of AMIS (Asan Medical Center information system). After verifying the content validity of extraction method from clinical data, two methods extracting KPCS-1 score (from clinical data vs. nurses' recording) were compared for 348 patients. RESULTS: This study demonstrated that extracting patient classification from clinical data is high value of validity (except 4 items excluded from this study), reliability between two methods extracting KPCS-1(from clinical data and nurses' recording) is high value (ICC=0.96, p<.001) and construct validity of two methods has similarity. CONCLUSIONS: It is believed that the patient classification system which is made from only clinical data without nurse's work burden is available. And 4 items which was excluded on KPCS-1 and examine area which had low level of reliability are needed to be amended.