Our knowledge of modeling and our mathematical literacy allow us to use a variety of methods. We can explore new ways of thinking to solve specific problems. Here are some of the approaches we can use:

Value of information

To make a decision, we need information and we all want to make informed decisions. But generating or gathering information has a cost. Time, money, resources, competitive edge – all can be exhausted by the quest for more information. A wise approach is first to query the value of the information we intend to generate.

Bayesian updating

We all do Bayesian updating every day, innately. In an uncertain world, we assign probability of occurrence to different outcomes of interest. Naturally, as time goes by and information accrues, we update our estimates of probability. We continuously update our knowledge. This update is generally an improvement as it is based on more information, evidence and experience. Bayes’s theorem allows us to analyze and quantify this updating.  Bayesian updating is very useful for risk-sharing modeling.

Optimization

We use mathematical programming to select the best option from a list of possibilities, subject to a number of constraints.  For example, a healthcare purchaser wants to maximize the benefit to a particular group of people, subject to some budgetary and logistical constraints.

Multi-criteria decision analysis (MCDA)

Decision-making is rarely based on a single criterion. It is in fact a multi-faceted process. So we need to account for multiple criteria, to which different stakeholders may attach different weight. The term MCDA encompasses a set of tools that allow us to do so in a transparent and consistent manner. MCDA can also be combined with optimization techniques or fuzzy logic.

Fuzzy logic

The world isn’t black or white, but grey! In addition to uncertainty, there is a lot of imprecision or ambiguity in our world. A response rate in a specific group can be uncertain, but what is a response? A 20% decrease in a disease activity score? So a patient with a 19.9% decrease is not a responder while one with a 20.1% decrease is?

Fuzzy logic contrasts with Boolean logic that assigns the logical value TRUE or FALSE (1 or 0). Responder = TRUE (1) if decrease ≥ 20% or Responder = FALSE (0) if decrease < 20%. With fuzzy logic, things can be TRUE or FALSE to some degree. Fuzzy logic formalizes imprecision using natural language and can be perceived as a qualitative computational method. The terms “No-Response”, “Low Response”, “Mild Response” or “High Response” are used instead of a Boolean variable such as Responder = TRUE/FALSE.  Fuzzy logic can be useful for defining perceived value or quality of life.

Real option

In an uncertain and constantly changing environment, decision-making cannot be a one-off process. It is wiser to proceed in stages and allow some flexibility in decision-making.  This is what the real-option approach does.  For example, progressive investments give you the right but not the obligation to invest more later on. Therefore, real-option valuation can help you allocate budgets and take investment decisions.