(Vienna, 31-01-2023) A study led last year by Wolfgang Schreiner from MedUni Vienna demonstrated the advantage of decision theory as a valuable extension of conventional statistics in therapy selection in precision medicine. In a research study based thereon, a new method was investigated to render medical decisions more justifiable and independent of intuitions. This is particularly important wherever different diagnostic procedures deliver contradictory findings, each with different therapeutic consequences. The results of the study were recently published in the Journal of Personalised Medicine.
The new method, known as the Flexible Risk Evidence Combination Rules, was explored by the MedUni Vienna scientists using the example of determining the hormone receptor status of patients with breast cancer. The receptor status was determined by using immunohistochemistry (IHC) and gene expression (GE). These two diagnostic methods can sometimes yield conflicting findings. "We demonstrate how to determine the maximum acceptable risk of error for the combination of both findings," explains study leader Wolfgang Schreiner, from MedUni Vienna's Centre for Medical Data Science. Depending on the "strategy" chosen ("risky" or "safe"), fewer or more patients end up in the "undecidable" group, which should in fact mean "too risky to make a clear decision". The clinical consequence is a repetition of the measurements and a new analysis.
Formally, the adjustment of risk is achieved by calculating two essentially counteractive combination rules for each patient: Firstly, the Yager rule, which is very "conservative", reacts sensitively to contradictions in the findings and very easily yields the result "undecidable". Secondly, the Dempster rule, which is "risk-accepting" and still delivers a decision even in the case of substantial contradictions. The new instrument of "Flexible Risk Evicence Combination Rules" explored in the study allows the combination of both methods with relative weights (parameter λ, 1 - λ) and thus to choose between a 'conservative' (λ = 0) and a 'risky' strategy (λ = 1) for each specific patient group. This choice of risk to be taken naturally remains physician’s responsibility but can be defined within standard operating procedures (SOPs) and then implemented as a 'strategy'.
Considering individual factors
Choosing the right therapy in everyday clinical practice often requires complex decisions in which numerous aspects have to be considered simultaneously. Decision-making based on statistics or even intuition reaches its limits, especially in precision medicine. After all, a variety of individual factors must be taken into account when selecting personalised therapy. Correct decisions in precision medicine are formalised and reproducible by decision theory, as the study by Wolfgang Schreiner's research team revealed. This applies in particular when different diagnostic procedures render contradictory findings, each of which would have different therapeutic consequences. "As a treating physician, should one decide intuitively? Or is the contradiction so blatant that one cannot decide at all?" Schreiner describes the medical dilemma in such situations. Here, decision theory provides assistance through a quantitative evaluation of the options - including a quantifier for the significance of contradiction. "If one of the therapies is vital, one will accept a higher risk and apply it, even if other findings contradict it. Previously, strategic therapy decisions of this kind were based on clinical intuition and can now be formalised by the methodology we have presented," says Wolfgang Schreiner about the "Flexible Risk Evicence Combination Rules". This increases the quality of decisions made and renders them justifiable.
Publication: Journal of Personalized Medicine
Flexible Risk Evidence Combination Rules in Breast Cancer Precision Therapy
Michael Kenn, Rudolf Karch, Christian F. Singer, Georg Dorffner, Wolfgang Schreiner
Doi: 10.3390/jpm13010119
https://doi.org/10.3390/jpm13010119