Does Bayes have it? Decision Support Systems in diagnostic medicine

Lindgaard, Gitte; Pyper, Catherine; Frize, Monique; Walker, Robin

doi:10.1016/j.ergon.2008.10.011

Cited by 20 publications

(10 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some studies [23,17] found that users did not demonstrate much confirmation bias, even though the studies were designed to elicit this bias. Others [15,13] have observed confirmation bias among DSS users. These studies frame a DSS and its recommendations as a source of potentially confirmatory or disconfirmatory information to be considered among other information in the decision task.…”

Section: Discussionmentioning

confidence: 96%

Customization bias in decision support systems

Solomon

2014

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Many Decision Support Systems (DSS) afford customization of inputs or algorithms before generating recommendations to a decision maker. This paper describes an experiment in which users make decisions assisted by recommendations of a DSS in a fantasy baseball game. This experiment shows that the act of customizing a DSS can lead to biased decision making. I show that users who believe they have customized a DSS's recommendation algorithm are more likely to follow the recommendations regardless of their accuracy. I also show that this customization bias is the result of using a DSS to seek confirmatory information in a recommendation.

show abstract

Section: Discussionmentioning

confidence: 96%

Customization bias in decision support systems

Solomon

2014

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

show abstract

“…In the last few decades, nonlinear classifiers like BNs have been in very high demand for an accurate and reliable analysis of such medical data sets, which may contain a considerable amount of uncertainty (e.g., uncertainty in clinical observations or test results). In recent years, a number of research projects have been carried out to exploit the BNs for various types of data analysis and decision making [18][19][20][21][22][23][24][25]. Even though the previous works investigated and developed the Bayesian inference models theoretically, a few of them actually exploited the semantic characteristics of the BNs to derive highly specific information from the links between the attributes.…”

Section: Discussionmentioning

confidence: 99%

Discovery of the connection among age-related macular degeneration, MTHFR C677T and PAI 1 4G/5G gene polymorphisms, and body mass index by means of Bayesian inference methods

Çelebiler¹,

Şeker²,

YÜKSEL³

et al. 2013

Turk J Elec Eng & Comp Sci

View full text Add to dashboard Cite

Age-related macular degeneration (AMD) is the leading cause of irreversible blindness in the elderly. The aim of this study was therefore to explore the relationship between the presence of multiple gene polymorphisms and 2 distinct advanced 'dry and wet' AMD phenotypes, and to assess gene interactions with the influence of personal factors in a Turkish population as a pilot study.For the analysis, the data were collected from 73 unrelated participants, grouped as 29 wet and 26 dry AMD patients, and 18 healthy controls. They were all genotyped for the multiple gene polymorphisms in 12 different genes. The data set collected was then analyzed using the Bayesian inference methods and visualized by means of the Bayesian networks.The results suggest that: 1) the PAI-1 4G/5G and FV G1691A genes have joint roles in the separation of the 3 groups; 2) both wet and dry AMD can be separated from the control group using the genes PAI-1 4G/5G, FV G1691A, FXII V34L, and PT G20210A; 3) although the wet AMD and control groups can be separated by the combination of the ACE I/D and B-fibrinogen-455 G-A gene polymorphisms, there seems to be no significant effect of the genes on the separation between the dry AMD and control groups; 4) the wet AMD and control groups can be distinguished by the combination of body mass index and the MTHFR-C677T and PAI-1 genes; and 5) there is a correlation between wet AMD and a high body mass index ( > 30 kg/m 2 ) . It was also found that the impact of body mass index on the disease development seems only in question with the connective availability of the genes MTHFR C677T and PAI-1. It can be concluded that the combination of the MTHFR C677T and PAI-1 4G/5G gene polymorphisms in the presence of obesity may increase the risk of wet AMD. In addition, the results further support a complex interplay among genetic and environmental factors in the development of different phenotypes.

show abstract

“…Bisantz (2006) provided the comprehensive discussion on the analysis and design of human-technology systems in cognitive engineering. Recently, Lindgaard et al (2009) discussed how psychological research can contribute to the design of medical diagnostic decision support systems.…”

Section: Introductionmentioning

confidence: 99%