8/20 reading & discussion

In the first article, how CAT combined IRT method to improve the functional outcome become more precision and effecience!
First step,we need to choose a specified item pool(specified outcome domain).
2nd step , item selection: use IRT to select an initial item to provide good discrimination over a wide range that all p't answer as the first question.
3rd step, patient respone:according to p't respone, we can have an initial score estimate and the score of confidence interval.
-IRT have 2 methods (Rasch modeling & CI).
These two methods can help the CAT to improve the functional outcome become more precision and effecience!
-->got three parameter:1,2,3-parameter;1-parameter is for difficult,2-parameter is for discrimination,3-parameter is for guessing.Now,it has assumption that item have equal dicrimination parameter and that guessing not a factor in the date.
-->Rash modeling :it can provide an index of each item's difficult-easy level, that will let us to determine if this items of the measurement need to modify or make shorten to become more comprehensive measurement or not.Besides,it also can make the item and p't's ability on the same metric. It can provide information about p't's abilities are fit with the specific component outcome domain.

-->confidence interval: it relative to the probabilities of patients, it can determine the suitable place of individual along functional dimension.
IRT can develop measure scale that are comprised of items with a known relationshipbetween item responses and position on an underlying functional domain.,called item characteristic curve. Using this approach,probabilities of patients scoring a particular response on an item at various funtional ability levels can be modeled.These probability can used to determine the individual's most like ly position along the functional dimension.

Both methods will develop the measurement become more precise and faster with effeicient!Because they make the possible area/range/distribution of p't abilities will become more smaller and that will faster let we determine where the p't abilities is!

4th step, score calculation: after the 3rd steo-the basis of the patient's response,then can estimate initial score and confidence interval.
5th step, stop rules or recontinue to step 2: that are based either the size of the pre-programmed confidence Interval are satisfied or the maximum number of items that used to test.
6th step,functional outcome: If there are not satisfied,a new item from the item pool is continued to administer, untilthe stop rule is satisfied.


Question:future challenges
1.why difficult to construct item bank? is the items too much? or other else?
2.why so few expert use CAT instrument
3.advertisement?!