Explore chapters and articles related to this topic
Clinical Trial Designs
Published in Gary L. Rosner, Purushottam W. Laud, Wesley O. Johnson, Bayesian Thinking in Biostatistics, 2021
Gary L. Rosner, Purushottam W. Laud, Wesley O. Johnson
The components needed to apply Bayesian decision theory are the following. First, one needs to specify the space of possible actions one will consider. This action space may consist of nominal discrete points, such as whether or not to stop the study, or the set of actions may be essentially continuous, as when one decides on the number of future patients to enroll. Secondly, there will be a utility function. This utility function can relate each action to a loss function, a gain, or a combination of the two, such as a function that relates cost to benefit. Next, one needs to specify a sampling distribution that characterizes the stochastic nature of the data the study will collect. This distribution often includes model parameters, some of which may relate to the primary study outcome, such as the probability of response or the treatment-specific risk of an event. Finally, one needs a characterization of the uncertainty associated with the model parameters (i.e., prior distributions). One needs to account for all of this uncertainty when deciding which action to take based on the expected utility that would result from each action. Once one has the expected utility for each action, considering a discrete action space to simplify the discussion, one can choose the action that is associated with the maximal expected utility.
Characterization of Uncertainty
Published in Samuel C. Morris, Cancer Risk Assessment, 2020
Most risk analyses include many built-in assumptions that lead to overestimates of risk. This reflects the preference of regulatory agencies to err on the side of protecting public health. The degree of overestimation needed, however, can be reduced by replacing uncertainty by knowledge. Dupuis and Lipfert (1986) state: “Reliance upon modeling as the basis of regulatory policy dictates an examination of model improvement needs and the economic benefits that could result from the use of improved models.” Model improvements have the potential to eventually result in large savings in pollution control costs. A quantitative technique of decision theory, the analysis of value of future information, can help to determine how much effort could profitably be put into improving the models basic to risk assessments. Computer programs are available to carry out this type of analysis (Finkel and Evans, 1987).
Economic Evaluation and Cost-Effectiveness of Health Care Interventions
Published in Emmanuel Lesaffre, Gianluca Baio, Bruno Boulanger, Bayesian Methods in Pharmaceutical Research, 2020
Nicky J Welton, Mark Strong, Christopher Jackson, Gianluca Baio
Whether we estimate cd and ed (and therefore our utility) directly in a trial, or indirectly using a model, we are uncertain about their true values (where we take ‘true’ to mean those values we would observe in a perfect study of infinite size). Uncertainty about the true values of cd and ed can (but not always) result in decision uncertainty. Decision theory tells us that in the face of uncertainty we should choose the option that maximises our expected utility (Lindley, 1991; Bernardo and Smith, 1994; Claxton, 1999), where the expectation in this case is taken with reference to the (joint) probability distribution that represents our judgements about cd and ed based on current evidence. By placing a distribution over the unknown quantities cd and ed we are representing uncertainty about our beliefs using probability, and are therefore being Bayesian.
I would rather be vaguely right than precisely wrong
Published in Scandinavian Journal of Primary Health Care, 2022
Reidar Brumer Bratvold, Svein R. Kjosavik
The only way doctors can help patients is by making decisions, whether it is to conduct tests, give advice, prescribe drugs or treatments, or refer to another doctor. Even when a GP helps the patient by listening, understanding and being compassionate, she has made a decision to do so. Such decision-making is not easy and requires an evaluation of many complex and uncertain factors. Still, if the clinician regularly makes mediocre decisions, she may never accomplish the things that are important to the patients in her care, to herself or to the healthcare system she represents. Empirical evidence demonstrates that clinicians, as well as people in general, often make suboptimal decisions [1,2]. Even when clinicians make decisions based on good quality information, they may be inconsistent and biased. Decision theory, which has been developed over more than 300 years, provides both an overall paradigm and a set of tools to help decision-makers construct and analyze models of decision situations.
Structured Professional Judgment (SPJ) Violence Risk Case Formulation and Psychopathic Personality Disorder
Published in International Journal of Forensic Mental Health, 2022
Dylan T. Gatner, Kevin S. Douglas, Stephen D. Hart, P. Randall Kropp
Case formulation or case conceptualization is a hypothesis about the precipitants and maintaining factors of clinical problems, which serves to guide future intervention (Eells & Lombart, 2011). Although evaluations of its effectiveness are mixed (Sturmey & McMurran, 2011), case formulation is considered a core competency that facilitates evidence-based practice (Bieling & Kuyken, 2006). Violence risk case formulation is an iterative theory-driven process of explaining a person’s past and present violent-related circumstances to identify effective future risk management strategies (see Hart et al., 2011). There are various forms of forensic and violence risk case formulation (Sturmey & McMurran, 2011). One particular theory underpinning violence risk case formulation is known as action or decision theory (Hart & Logan, 2011). Action theory is best characterized as a family of theories (Aguilar & Buckareff, 2010) derived from various disciplines (e.g., philosophy, criminology, psychology). Broadly, action theory asserts that all human action is, to some degree, intentional and is influenced by mental states (e.g., beliefs, choices, or decisions; Aguilar & Buckareff, 2010; Davidson, 1980). Action theory has been applied to psychological perspectives (Eckensberger, 2001; Finkel, 2008) and criminological theories of violence and crime.
Games surgeons play
Published in British Journal of Neurosurgery, 2019
There are other possible approaches though. One of these is the study of decision theory and in particular decision quality. A decision such as whether to have a surgical operation or not can be formally modeled by estimating the likely benefits and costs of surgery and comparing them with the likely benefits and costs of conservative treatment. When this is done as objectivity and accurately as possible, given the available data, the result will fall into three broad categories: clearly surgery is a bad idea, clearly surgery is a good idea or the situation is ambiguous. Modern research methods are not great at capturing and measuring the anxiety, discomfort and inconvenience that surgery causes and so frequently in an ambiguous situation, a patient will reasonably choose not to have the operation.