“…(1) The classic, close-to-data approach is to state specific patterns of interest (e.g., in-hospital functional decline) and find persons with them, or to search broadly for common (high-prevalence) patterns. Whether stated a priori or not, patterns are identified by simple computer algorithms, visual data review, or variability statistics (e.g., standard deviation of repeated measures; Beckett et al, 1996; Bowling, Farquhar, & Grundy, 1994; Carlson et al, 1998; Charlton, 1989; Ferrucci et al, 1996; Fortinsky, Covinsky, Palmer, & Landefeld, 1999; Gill, Gahbauer, Han, & Allore, 2009; Jette, Harris, Cleary, & Campion, 1987; Latham, 2012; Maddox & Clark, 1992; Romoren & Blekeseaune, 2003; Verbrugge & Balaban, 1989; Verbrugge, Reoma, & Gruber-Baldini, 1994; Verbrugge, Yang, & Juarez, 2004). (2) As sophisticated quantitative techniques advanced, an alternative approach was possible.…”