Modeling Method Decision Tree

Preview:

Citation preview

Modeling MethodDecision Tree

Decision tree: 4 basic elements

2

http://www.excelquant.com/decision-tree-analysis-in-excel/

https://msdn.microsoft.com/en-us/library/ms175312.aspx

3

4

5

The process of the analysis is:

1. Structure the problem: identify alternatives, list possible outcomes, represent sequence of events

2. Assign probabilities to all chance events

3. Assign utility, or other values, to all outcomes

4. Evaluate the expected utility, or other values, and cost of each choice

6

7

Parameters

8

Name Description Root Definition Low HighpEradicateRad Prob tumor is eradicated by radiation 0.6 0 1pEradicateRadSurg Prob tumor is eradicated by radiation and surgery 0.8 0 1cRadiation Cost of radiation 30K 25000 35000cSurgery Cost of surgery 50K 40000 60000cFollowupAnnual Cost of followup per year 2K 1800 2200effEradicated Life expectancy if tumor eradicated 10 8 15effNotEradicated Life expectancy if tumor is not eradicated 3 0 3

Step 1: structure the problem

9

Step 2: assign probabilities to all chance events

Note: sum of probabilities at each chance node is equal to 1

10

Step 3: assign utility, or other values, to all outcomes

11

Step 4: evaluate the expected utility, or other values, and cost of each choice

12

(0.6 x 10) + (0.4 x 3) LYs

(0.8 x 10) + (0.2 x 3) LYs

(0.6 x 50K) + (0.4 x 36K) THB

(0.6 x 100K) + (0.4 x 96K) THB

ICER Calculation

(0.6 x 10) + (0.4 x 3) LYs= 6 + 1.2 = 7.2

(0.8 x 10) + (0.2 x 3) LYs= 8 + 0.6 = 8.6

(0.6 x 50K) + (0.4 x 36K) THB= 30,000 + 14,400 = 44,400

(0.8 x 100K) + (0.2 x 96K) THB= 80,000 + 19,200 = 99,200

Cost = 99,200 – 44,400 = 54,800  LY = 8.6 – 7.2 = 1.4 

= 39,142.86 THB/LY

Limitations of decision tree are:

1. Conclusion depends heavily on the assumptions, so sensitivity analysis is important

2. Force movement to occur in one direction from left to right

3. Hard to do the model with long cycle time

14

TreeAge Pro®Decision Tree

Exercise: Decision tree by TreeAge Pro®- install program - create new decision tree (Ctrl + N)- run sensitivity analysis

16

17

18

File Projects New project

File Open

File New decision tree or Ctrl + N

19

1. Add branch by double click on node

2. Define name

20

3. Add node

21

Tree Tree Preferences

4. Select tree preferences

22

5. Add probabilities

23

6. Change “Terminal node” and edit payoffs

24

25

8. Roll back

Analysis Roll Back

26

Define variables (parameters)

Window views variable properties

Add variable

Edit variable

Delete variable

27

Define variables (parameters)

28

Edit probability

Add “#” to

sum prob. = 1

29

TreeAge Pro®Markov model

Exercise: develop Markov model by TreeAge Pro®- install program - create new decision tree (Ctrl + N)- run sensitivity analysis

31

32

Convert to STD (State Transition Diagram)

33

1. Add branch by double click on node

2. Add Markov node by drag and drop from Palette

34

3. Define name and insert node for Markov

35

4. Complete chance nodes as shown in State Transition Diagram

36

5. Assign terminal nodes for each branch

37

6. Add variable properties

38

7. Add variables in the model

39

8. At Markov nodeAnalysis Expected value

40

Collapse subtree : Ctrl + J

41

TreeAge Pro®Tornado Diagram

Exercise: Decision tree by TreeAge Pro®- install program - create new decision tree (Ctrl + N)- run sensitivity analysis

43

44

Tornado diagramAnalysis Sensitivity analysis Tornado diagram

45

TreeAge Pro®Monte Carlo Simulation

1. Define definitions

47

48

49

2. Define variable properties

3. Run: Analysis Monte Carlo Simulation Sampling

50

ICER distribution (Charts -> scatter plots -> ICE scatter)

51

Acceptability curve (Charts -> CE acceptability)

52

Recommended