Title: A pattern-clustering method for longitudinal data on heroin users receiving methadone
Authors: Chien-Ju Lin - MRC Biostatistics Unit University of Cambridge (United Kingdom) [presenting]
Abstract: Methadone is used as a substitute of heroin and there may be certain groups of users according to methadone dosage. We analyze data for 314 participants over 180 days. The data consist of seven categories in which six categories have an ordinal scale for representing dosages and one for missing dosages. We develop a clustering method involving the so-called p-dissimilarity and an ordering algorithm. The p-dissimilarity is used to measure dissimilarity between the 180-day time series of the participants. It accommodates ordinal and categorical scales by using a parameter p as a switch between data being treated as categorical and ordinal. We use heatplots to evaluate the quality of clustering. A heatplot consists of horizontal lines representing data for participants by colour. The interpretability depends on the location of the participants along the vertical-axis. We propose an algorithm using a projection vector to locate participants. The heatplot can then be used for information visualisation. It displays clustering structures, relationships between participants and clusters, and the density of clusters. Despite the fact that no significant clustering structure is observed, the sequence of categories for clusters are useful for clinicians to prescribe an appropriate dosage for improving efficiency of methadone maintenance therapy.