Abstract
Decision analytical methods have been utilized and demonstrated to be of use for a broad range of applications in medical contexts, from regular diagnostic strategies and treatment to the evaluation of diagnostic tests and prediction models and benefit-risk assessments. However, a number of issues still remain to be clarified, for instance ease of use, realism of the input data, long-term outcomes and integration into routine clinical work. In particular, many people are unaccustomed or unwilling to express input information with the preciseness and correctness most methods require, i.e., the values need to be “true” in some sense. The common lack of complete information naturally increases this problem significantly and several attempts have been made to resolve this issue. This is not least the case within psychiatric emergency care where the information available often is of a highly qualitative nature. In this article we suggest the use of so called surrogate numbers that have proliferated for a while in the form of ordinal ranking methods for multi-criteria and show how they can be adapted for use in probability elicitation.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
- 2.
We will, unless otherwise stated, presume that decision problems are modelled as simplexes S x generated by x 1 > x 2 > … > x N , Σx i = 1, and 0 ≤ x i .
- 3.
We assume a standard one-level probability tree with a decision node followed by alternatives of action. For each alternative, there is a set of exhaustive and mutually exclusive events which the decision-maker is asked to rank with respect to the probabilities of occurrence. The results are easy to generalise to multi-level trees.
- 4.
Note that these are only suggestions for illustrative purposes. This paper does not intend to discuss problems with eliciting verbal probability statements and their conversion to numerical data. The cardinal information is rather in an actual implementation of the method considered to be input using graphical sliders in a software tool.
- 5.
In order to simplify the presentation, we use p i to represent the probability p(c i ).
- 6.
For simplicity in generation procedures, but without loss of generality, assume that the events in all alternatives have the same number of consequences, N.
- 7.
A second success measure we used is the matching of the three highest ranked alternatives (“podium”), the number of times the three highest evaluated alternatives using a particular method all coincide with the true three highest alternatives. A third set generated is the matching of all ranked alternatives (“overall”), the number of times all evaluated alternatives using a particular method coincide with the true ranking of the alternatives. The two latter sets correlated strongly with the first and are not shown in this paper.
- 8.
The standard deviations between sets of 10 runs were around 0.2–0.3 per cent.
References
Sygel, K., Danielson, M., Ekenberg, L., Fors, U.: Handling imprecise information in emergency psychiatric care. In: Proceedings of ICSSE IEEE International Conference on System Science and Engineering (2015)
Barron, F.H.: Selecting a best multiattribute alternative with partial information about attribute weights. Acta Psychol. 80(1–3), 91–103 (1992)
Corner, J.L., Corner, P.D.: Characteristics of decisions in decision analysis practice. J. Oper. Res. Soc. 46(3), 304–314 (1995)
Danielson, M., Ekenberg, L.: Rank ordering methods for multi-criteria decisions. In: Zaraté, P., Kersten, G.E., Hernández, J.E. (eds.) GDN 2014. LNBIP, vol. 180, pp. 128–135. Springer, Heidelberg (2014)
Danielson, M., Ekenberg, L., Larsson, A.: Distribution of belief in decision trees. Int. J. Approximate Reasoning 46(2), 387–407 (2007)
Danielson, M., Ekenberg, L., Larsson, A., Riabacke, M.: Weighting under ambiguous preferences and imprecise differences in a cardinal rank ordering process. Int. J. Comput. Intell. Syst. 7, 105–112 (2013)
Danielson, M., Ekenberg, L., He, Y.: Augmenting ordinal methods of attribute weight approximation. Decis. Anal. 11(1), 21–26 (2014)
Delavande, A., Gine, X., McKenzie, D.: Measuring subjective expectations in developing countries: a critical review and new evidence. J. Dev. Econ. 94, 151–163 (2011)
Hogarth, R.M.: Cognitive processes and the assessment of subjective probability distributions. J. Am. Stat. Assoc. 70, 271–289 (1975)
Johnson, E.M., Huber, G.P.: The technology of Utility Assessment. IEEE Trans. Syst. Man Cybern. 7(5), 311–325 (1977)
Lenert, L.A., Treadwell, J.R.: Effects on preferences of violations of procedural invariance. Med. Decis. Making 19(4), 473–481 (1999)
Lichtenstein, S., Slovic, P. (eds.): The Construction of Preference. Cambridge University Press, New York (2006)
Morgan, M.G., Henrion, M.: Uncertainty—A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis. Cambridge University Press, New York (1990)
Rey-Biel, P.: Equilibrium play and best response to (stated) beliefs in normal form games. Games Econ. Behav. 65, 572–585 (2009)
Stillwell, W., Seaver, D., Edwards, W.: A comparison of weight approximation techniques in multiattribute utility decision making. Organ. Behav. Hum. Perform. 28(1), 62–77 (1981)
Tversky, A., Kahneman, D.: The framing of decisions and the psychology of choice. Science 211(4481), 453–458 (1981)
Wallsten, T.S., Budescu, D.V.: Encoding subjective probabilities: a psychological and psychometric review. Manage. Sci. 29(2), 151–173 (1983)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Rights and permissions
Copyright information
© 2015 Springer International Publishing Switzerland
About this paper
Cite this paper
Danielson, M., Ekenberg, L., Sygel, K. (2015). Robust Psychiatric Decision Support Using Surrogate Numbers. In: Fujita, H., Guizzi, G. (eds) Intelligent Software Methodologies, Tools and Techniques. SoMeT 2015. Communications in Computer and Information Science, vol 532. Springer, Cham. https://doi.org/10.1007/978-3-319-22689-7_44
Download citation
DOI: https://doi.org/10.1007/978-3-319-22689-7_44
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-22688-0
Online ISBN: 978-3-319-22689-7
eBook Packages: Computer ScienceComputer Science (R0)