Dear DCEA Students,
In general the HW is looking good. Many of you exceeded the requirements of this first assignment by including elaboration beyond the basic depiction of the decision, major health consequences, and plausible probabilities and utilities for the outcomes.
As mentioned in the HW instructions, you can make your life easier by depicting multiple years of testing on one branch. You don't need to depict the consequences of each year of testing separately. If you added test performance, (false negatives etc., great!) – and not necessary at this stage.
One problem I saw in many of the HWs: An attempt to assess the outcome of screening before putting in the true prevalence of the condition (e.g. breast cancer). Look at the “Diagnostics test” sheet in the “Tree for diagnostic tests” file posted at the top of the CLE under “Tools and Resources” for an example of a good way to proceed.
Confine each chance node to two branches. The is technically nothing wrong with having 3+ branches from a chance node, but it is sometimes harder to keep track of the logic of the model when you do that. There is no 3+ branch version that can’t be converted to a 2-branch version. For example, in the tree below, the elaboration for Adverse Events for TxA is what I do not recommend. The version for TxB is what I do recommend. It adds a chance node, but, unless your tree is very simple, will save you aggravation over the long run.
A few students did not yet submit their homework. Please do so in the next 24 hours.
See you Thursday!
- Elliot
