BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20250704T113650EDT-4265rhurc2@132.216.98.100
DTSTAMP:20250704T153650Z
DESCRIPTION:How do we know whether a predictive model is of clinical value?
  How do we know whether a molecular marker is worth measuring? A discussio
 n of some simple decision analytic methods.\n\nhttps://www.mskcc.org/profi
 le/andrew-vickers\n\n	There is increasing interest in and use of multivaria
 ble prediction models to aid clinical management. In oncology\, it has bee
 n shown that such models are more accurate than the use of crude risk cate
 gories\, such as those based on cancer stage. Accordingly\, it has been su
 ggested that multivariable models should be used to make decisions about p
 atient care\, such as whether a patient should undergo biopsy in the light
  of a raised PSA level. Research on molecular markers has mirrored the gro
 wth of prediction models: currently an enormous volume of papers are publi
 shed examining whether a tissue or blood marker can predict the occurrence
  or course of disease. Markers and models are currently evaluated in terms
  of accuracy using metrics such as the area-under-the-curve (AUC)\, sensit
 ivity and specificity or the concordance index. A model is thought to be a
  good one if it is accurate\; a marker is claimed to be of value if it inc
 reases the accuracy of a model. But how accurate is accurate enough? For i
 nstance\, should we use a model with an AUC of 0.65\, or only those with A
 UC's above 0.75? Similarly\, if a marker improves AUC from\, say\, 0.65 to
  0.68\, is it worth using in the clinic? Or even taken the simple case of 
 two binary diagnostic tests\, with sensitivities of 91% and 51% and specif
 icities of 40% and 78%: which is better? Markers and models can also be ev
 aluated in terms of calibration. But how much miscalibration would be 'too
  much' to prevent clinical use of a model? What about the case where one m
 odel has better calibration and the other better discrimination\, which mo
 del should be used? The answers depends\, of course\, on what the model\, 
 test or marker will be used for. Evaluating models and markers in terms of
  clinical consequences is the remit of a field known as 'decision analysis
 '. The problem with traditional decision analysis is that it requires addi
 tional information\, for example\, on the benefits\, harms and costs of tr
 eatment\, or on patient preferences for different health states. Perhaps a
 s a result\, the number of papers in the literature using decision analyti
 c methods is dwarfed by those that report accuracy. In this presentation\,
  I will describe some simple decision analytic methods that can be directl
 y applied to the data set of a model or marker\, without the need for exte
 rnal information. These methods can therefore be used to tell us whether o
 r not to use a model in the clinic\, or whether a marker is a good one. I 
 will illustrate the use of the methods with some straightforward real-life
  examples. All references are available at www.decisioncurveanalysis.org\n
DTSTART:20180123T203000Z
DTEND:20180123T213000Z
LOCATION:Room 24\, Purvis Hall\, CA\, QC\, Montreal\, H3A 1A2\, 1020 avenue
  des Pins Ouest
SUMMARY:Andrew J. Vickers\, PhD\, Memorial Sloan-Kettering Cancer Center
URL:/mathstat/channels/event/andrew-j-vickers-phd-memo
 rial-sloan-kettering-cancer-center-283975
END:VEVENT
END:VCALENDAR