Pii: S0895-4356(00)00344-9
نویسنده
چکیده
A critique is presented of the use of tree-based partitioning algorithms to formulate classification rules and identify subgroups from clinical and epidemiological data. It is argued that the methods have a number of limitations, despite their popularity and apparent closeness to clinical reasoning processes. The issue of redundancy in tree-derived decision rules is discussed. Simple rules may be unlikely to be “discovered” by tree growing. Subgroups identified by trees are often hard to interpret or believe and net effects are not assessed. These problems arise fundamentally because trees are hierarchical. Newer refinements of tree technology seem unlikely to be useful, wedded as they are to hierarchical structures. © 2001 Elsevier Science Inc. All rights reserved.
منابع مشابه
Pii: S0895-4356(00)00206-7
Although the placebo in a clinical trial is often considered simply a baseline against which to evaluate the efficacy of a clinical intervention, there is evidence that the magnitude of placebo effect may be a critical factor in determining the results of a trial. This article examines the question of whether devices have enhanced placebo effects and, if so, what the implications may be. While ...
متن کاملPii: S0895-4356(02)00504-8
An almost endless number of observations and experiments have effectively falsified the hypothesis that dietary cholesterol and fats, and a high cholesterol level play a role in the causation of atherosclerosis and cardiovascular disease. The hypothesis is maintained because allegedly supportive, but insignificant findings, are inflated, and because most contradictory results are misinterpreted...
متن کاملPii: S0895-4356(00)00347-4
The double-blind randomized controlled trial (RCT) is accepted by medicine as objective scientific methodology that, when ideally performed, produces knowledge untainted by bias. The validity of the RCT rests not just on theoretical arguments, but also on the discrepancy between the RCT and less rigorous evidence (the difference is sometimes considered an objective measure of bias). A brief ove...
متن کاملPii: S0895-4356(00)00232-8
Objectives: The aim was to assess the prevalence of any urinary leakage in an unselected female population in Norway, and to estimate the prevalence of significant incontinence. Methods: The EPINCONT Study is part of a large survey (HUNT 2) performed in a county in Norway during 1995–97. Everyone aged 20 years or more was invited. 27,936 (80%) of 34,755 community-dwelling women answered a quest...
متن کامل