H

T
L
G
A
N
E
I
H
K
A E
M TH S
N N G
O I IN
I
S LS T
I
T
C
E
E
E D S
D O E
M AR
C

Drag picture to placeholder or click icon to add

se
G
r
N cou
I
AK ng a
M cti .

N le ves
O s e a ti
I
S of rn
I
EC act alte
D he ng
- T mo
a

of

ac

on
i
t

om
r
f

E
H
T

N
O
I
S
I
EC

C
O
PR

S
S
E

1. Identification of the problem

and its nature.
2. Specification of objectives
and decision criteria.
3. Development of alternatives.
4. Analysis and comparison of
alternatives.
5. Selection of the best
alternatives.
6. Implementation of the
choice.
7. Controlling and monitoring
the results.

E
E
R
TH TU
F
O NA
N TS
O I
I
T D
A
C AN
I
F
A well-defined
I
M
T
E
N
L
problem is
E B
ID O
half-solved.
PR
Solution
must
tackle the underlying
problem,
not
the
symptoms.

N
O
I
S
I
F EC
O D
N D
O N
I
T A
A
C ES
I
F IV
I
The health care
C
T
E C IA
manager must
P
E
R
S BJ E
identify the criteria
O RIT
by which the solution
C
will be formulated.
Ex. Costs, profits, return of
investment, increased
productivity, risk, company
image

F
O

T
N
E S
PM IVE
O T
L
A
E
N
V
R
E
D LTE
A

Select the best one

after considering the
total set carefully.
-Compiling

a realistic mix of
suitable alternatives often
depends on the health care
managers
level
of
experience as well as on
the nature of the situation.

D
F
N
O
A N
S O S
I
S IS VE
Y
I
R
L
T
A
A P A
N
A OM RN
C LTE
A

A mathematical model is
abstract representation of
some real-world health care
process, system, or
subsystem.
Analyzing
and
comparing
alternative can be usually be
facilitated by computer programs
that give a skilled health care
manager the mathematical and
statistical techniques for doing
so. However, they should not be a
substitute
for
the
art
of
management.

E
TH

F
O S
N E
O IV
I
T AT
C
E N
L
R
E
E
S LT
A

T
S
BE

Which alternative best fits

my established objectives
within reasonable time and
cost constraints, and will
benefit the health care
organization as a whole?
Selecting the best alternative
depends on the objectives set by
the decision maker and the
criteria set for evaluating the
alternatives
offered
by
the
mathematical model.

N
O
I
T
A
T
EN
M
E
PL I CE
IM HO
C

F
O

E
H
T

Implementing
the chosen
alternative is
simply a matter
of putting it
into action.

S
T
L
U
S
D
E
AN E R
G H
N
T
I
L
L NG
Monitor the result
O I
R
R
T
of the decision to
N ITO
make sure that they
CO ON
occur as desired.
M
Otherwise,
the
health
manager may chose to repeat
the entire process.
Investigation may uncover an
error in the implementation
and/or
calculations,
or
perhaps a false assumption
that
affected
the
entire
process.

S
E
S
U
CA N?
T IO
A
H IS
W EC
D

R
O
PO
Bounded
Rationality
- The limits imposed on decision
making by costs, human abilities
and errors, time, technology,
and the tractability of data.

Suboptimization
In a highly competitive
environment, decisions are often
departmentalized as separate
organizational units compete for
scarce resource.

L
E
V
E
L
N EU
O LI
I
S I
I
EC N M
D IO
E
S
I
H
T EC
D

D
AN

Classifications:
1) The strategic level
of the decision
-) Vary from low to high
depending on the situation.

2) The level of
certainty
surrounding the
situation or states
of nature

L
E
V
E
L
N EU
O LI
I
S I
I
EC N M
D IO
E
S
I
H
T EC
D

D
AN

The decision milieu

maybe:
one of uncertainty or
one of risk,
each setting requires
distinctive decisionmaking tools to evaluate
the alternatives.

PAYOFF TABLE
a tool that is frequently used to select the best
alternatives given different possible outcomes.
shows expected payoffs for each alternative under
various possible conditions state of nature.
*A payoff table can be constructed using the outcome for
alternative i(Ai ), and state of nature j(Sj ) as Oij .
*Outcomes can be expressed in profits, revenue, or cost.

Insert table 3.1

PAYOFF TABLE

A major imaging center is not able to meet the

increased demand form patients for magnetic
resonance
imaging
studies
(MRIs).
The
administration is willing to explore the possibilities
by evaluating such alternatives as adding one or
two additional units or outsourcing to other
imaging centers and earning a commission of $30
per MRI.

PAYOFF TABLE: EXAMPLE

UNCERTAINTY exist in any scenario when

insufficient information makes it impossible to
assess the likelihood of possible future events.
Five possible decision strategies:
1. MAXIMIN
2. MAXIMAX
3. HURWITZ
4. MINIMAX REGRET
5. LAPLACE

DECISION MAKING UNDER

UNCERTAINTY

MAXIMINidentifies worst possible scenario for each

alternative, and aims to select the alternative that will
give the largest payoff in the worst circumstances. It is
regarded as pessimistic strategy.

1: MAXIMIN CASE

MAXIMAX identifies the alternative with best payoff

(highest maximum return). It is regarded as an
extremely optimistic strategy.

2: MAXIMAX CASE

HURWITZ allows for adjusted weighting between maximin and

maximax, or allows the health care manager choose a platform on
the continuum of pessimist versus optimist.
*Hurwitz provides a measure for assigning a weight towards
optimism and the remainder of the weight to pessimism.
*Hurwitz optimism weight would vary 0 1.
-when weight is = 1, the decision becomes optimistic.
-when weight is = 0, the decision becomes pessimistic.

Formula HURWITZ VALUE (HV):

HV (Ai) = (row maximum) + (1 ) (row

minimum)
*derivation: if selection of value is other than zero, the
optimistic value also produces a weight, which can be named
the pessimistic value, denoted as 1 .

3: HURWITZ CASE

3: HURWITZ CASE

MINIMAX REGRET calculates the worst regret (or opportunity loss)

for each alternative, and chooses the one that yields the least regret,
or that the health care manager can live with best.

REGRET refer to the opportunity loss that occurs when an alternative

is chosen and a particular state of nature occurs.
-difference between the best possible outcome under a state of
nature and the actual outcome from choosing a particular alternative.
Formula REGRET

Regret (Rij ) = maximum payoff for column j

payoffij
*procedure: In order to evaluate the minimax regret decisions,
the health care manager must develop a regret table (by
using regret formula), which converts the payoff table to
opportunity losses. Once the opportunity loss table is
created, the minimax rule can be applied.

4: MINIMAX REGRET CASE

LAPLACE calculates the average payoff for each

alternatives, and selects the one with the highest average.
*know as the principle of insufficient reason which means
no reason to assign differently as no known probabilities
exist; in other words, the health care manager can assume
equally likely probabilities for each state of nature.
that is, for n state of nature, the probability for each state
would be 1/n. (uniform probability distribution)
Formula Expected Outcome

E (Ai ) = j pj Oij
*choose highest expected payoff.

5: LAPLACE CASE

RISK exist when you do not know which events

will occur, but can estimate the probability that
any one state will occur.
*Between the two extremes of certainty and
uncertainty lies the state of risk.

The health care manager may have past data

objective
probabilities

from
similar
circumstances, or subjective estimates of the
probabilities.

DECISION MAKING UNDER RISK

 Objective

probabilities

can

be

obtained

through theory or empirically.

Theoritical objective probability uses mathematical
theory and logical framework.
Conducting empirical studies to estimate the
probability of a given situation. It can be used or
converted/approximated to one of the known probability
distribution using the statistical goodness-of-the-fit test.

OBJECTIVE PROBABILITIES

 In the absence of objective probabilities that are

reliable,
subjective
probability
becomes
prominent.
*Laplace strategy, as discussed, provided that under the
principle of insufficient reason if no reliable objective
probabilities can be determined for the time being all
state of nature may be equally probable. That is, it is
better to assign equal probabilities than to have none.
When more time and information becomes available,
the health care manager can modify the probability
information subjectively.

SUBJECTIVE PROBABILITIES

SUBJECTIVE PROBABILITIES

 Once the health care manager has assessed the

rpobability distribution, computation of the
expected
values
for
each
alternative
is
straightforward, using the formula under Laplace
strategy, as follows: EMV (Ai ) = j pj Oij
If the outcomes are measured in monetary value,
the expected model (EMV) is generally named as
expected monetary value, or EMV.
If the outcomes represent regrets (opportunity
losses), then one can calculate expected
opportunity losses, or EOL.

EXPECTED VALUE MODEL

 The health care manager can assess the

expected losses and try to minimize them with
proper decision.
Formula Expected Opporturnity Loss:
EOL (Ai ) = j pj Rij

EXPECTED OPPORTUNITY LOSS

 The concept of EVPI provides an avenue for

assessing the situation and determining level of
resources the health care manager would be
willing to commit for the situation.
Information can be obtained more cheaply by
EVPI, but its quality may not be good, or not
reliable. However, the health care manager
would like to the upper limit, or maximum price,
that can be spent to obtain the information.
In order to evaluate this situation, first consider
the case in which one perfect information about
the state of nature or which event would occur.
Then choose the alternative yielding the best
outcome.

EXPECTED VALUE OF PERFECT

INFORMATION (EVPI)

 However, how the manager would have decided

if any event was certain to occur?
Formula Expected Value under Certainty
EVUC =j pj ( Best Oij given Sj )
Solution :
EVUC = (.2*15,OOO) + (.6*200,000) +
(.2*725,000)
= 268,000.

EXPECTED VALUE OF PERFECT

INFORMATION (EVPI)

 However, the health care manager is currently

operating under risk and would like to achieve
certainty conditions to make the best decision.
Fomula EVPI
EVPI = EVUC EVM
Solution:
EVPI = 268,000 177,000 = 91,000.

EXPECTED VALUE OF PERFECT

INFORMATION (EVPI)

S
F
F
O
A
P

IF
T ?
A
H TS
W OS
C

E
R
A

The techniques
discussed above can
be easily be applied
to cost off tables by
reversing the logic.

 For pessimist decision (who use maximin) : will use minimax

cost.
One would search for the maximum cost of each alternatives
and then choose the alternative with minimum costs.
Optimist decision
One would use the minimin cost by choosing the minimun
among the minimum cost.
Minimax regret decision
Works similarly to the revenue/profit situation; however, a cost
regret table has to be constructed.
In cost regret table, in each column identify the lowest cost and
subtract that from the other alternatives cost.

WHAT IF PAYOFFS ARE COSTS?

WHAT IF PAYOFFS ARE COSTS?

N
O
I
S
I
EC CH
D A
E
O
H
R
T PP
A

E
E
TR

The tree is drawn

from left to right,
with square and
circle nodes that
are connected by
lines (branches).

EXAMPLE OF DECISION TREE

 To analyze the problem using the decision tree

format, starting from the left, the expected
values are calculated for every event nodes
using EMV (Ai ) = j pj Oij
These expected values are then placed on the
event nodes to compare the alternatives. Then,
the highest expected return is the decision.
The other decision branches of the tree should
no longer be considered, thus truncated
shown by placing ll.

ANALYSIS OF THE DECISION TREE :

ROLLBACK PROCEDURE

N
O N
I
T IO
A
R IS
T
C
S
E
U D
L
IL D
E AN S
R
A FF OD
W O H
WinQSB software
T
T
F AY E
provides an easy
O
S FP M
platform for
O EE
analyzing decision
TR

problems by using
either the payoff or
the rollback
procedure for the
decision tree.

 Often, available data on various measures are in

other than monetary terms. There may be
multiple measurements on various attributes of
the problem. Under those conditions, the health
care manager must resort to other techniques to
evaluate or assess outcomes.
Multiple-attribute decisions can use procedures
to simplify the process. They are
Dominance
Minimum attribute satisfaction
Most important attribute

DECISION ANALYSIS WITH

NONMONETARY VALUES AND MULTIPLE
ATTRIBUTES

 If an alternative (X) is at least as good as

another alternative (Y) on all attributes and
strongly the choice at least one attribute, then
alternative X dominates alternative Y.
Evaluation of the alternatives using dominance
procedure is conducted by considering a pair of
alternatives at a time.
If there are many alternatives, there might be
many pair-wise comparisons that have to be
completed.
Dominance process is best for reducing a
number
of
inferior
alternatives
from
consideration, but may not find a unique
solution for the decision where two
DOMINANCE
alternatives survived
the process. PROCEDURE

 When evaluating alternatives, especially in

contract of proposals, minimum acceptable
standards may be considered. When request for
proposals are developed, they often specify the
acceptable standards, or minimum attributes.
All alternatives are considered simultaneously
for each minimum attribute. If any alternative is
not satisfactory for a given minimum attribute,
that alternative is eliminated.

MINIMUM ATTRIBUTE SATISFACTION

PROCEDURE

 If neither the above procedures yielded a

solution, this procedure in most instances will.
Applying that ranking is done as in minimum
attribute
satisfaction,
by
simultaneously
considering all alternatives, first for highestranking attribute and then if no solution is
obtain, for the next attribute.
Although
the
most
important
attribute
procedure can find unique solutions, it can
often do so without evaluating all attributes.

MOST IMPORTANT ATTRIBUTE

PROCEDURE