Question 1: Heart Disease Bayesian Network

Scenario:
A hospital uses a Bayesian Network to model patient risk factors and diagnose Heart
Disease (H). The network encodes probabilistic relationships between lifestyle fac-
tors, physiological conditions, heart disease, and observable clinical outcomes.

Network Variables
Node Meaning Values
S Smoking True / False
E Exercise True / False
C High Cholesterol True / False
B High Blood Pressure True / False
H Heart Disease True / False
F Fatigue True / False
D Diagnosis Test Result True / False

Network Structure
Directed edges:

S → C, S → B, E → C, E → B, C → H, B → H, H → F, H→D

S E

C B

F D

Conditional Probability Tables

S=T S=F E=T E=F
0.3 0.7 0.4 0.6
P(S) P(E)

1
S E C=T C=F S E B=T B=F
T T 0.7 0.3 T T 0.6 0.4
T F 0.8 0.2 T F 0.7 0.3
F T 0.2 0.8 F T 0.1 0.9
F F 0.4 0.6 F F 0.3 0.7
P(C | S, E) P(B | S, E)

C B H=T H=F
T T 0.9 0.1
T F 0.7 0.3
F T 0.6 0.4
F F 0.1 0.9
P(H | C, B)

H F =T F =F H D=T D=F
T 0.8 0.2 T 0.95 0.05
F 0.1 0.9 F 0.08 0.92
P(F | H) P(D | H)

2
1.1 Joint Distribution Factorization
(a) Write the full joint probability distribution
P (S, E, C, B, H, F, D)
as a product of factors using the chain rule for Bayesian Networks. Identify
the parents of each node and justify each factor.

(b) Using your factorization from (a), write the expression for
P (S=T, E=F, C=T, B=T, H=T, F =T, D=T )
and compute its numerical value step by step.

(c) A student claims the joint distribution can also be written as

P (S) · P (E) · P (C, B | S, E) · P (H | C, B) · P (F, D | H).
Is this factorization correct? Justify your answer using the conditional indepen-
dence properties of the network.

1.2 Conditional Independence and d-Separation

For each of the following statements you must:
• List all paths between the two nodes.
• Classify each intermediate node as chain, fork, or collider.
• State whether each path is blocked or active, with reasons.
• Give the final independence verdict.
(a) Is the following marginal independence true?
S⊥E

(b) Is the following conditional independence true?

S ⊥ E | {H}
Explain carefully what happens at node H and identify the well-known phenomenon
this illustrates.

(c) Is the following conditional independence true?

S ⊥ E | {C}
What is fundamentally different about this case compared to (b)?

(d) Is the following conditional independence true?

F ⊥ D | {H}
Name the structural property of the network that explains your answer.

3
1.3 Bayes Theorem
(a) A patient tests positive on the diagnosis test (D=T ). Using Bayes Theorem,
compute:
P (H=T | D=T )
Show full working. Use the law of total probability to compute P (D=T ).

(b) The same patient also reports fatigue (F =T ). Compute:

P (H=T | D=T, F =T )

Show how the two pieces of evidence combine. Explain why D and F are not
independent in general, but are conditionally independent given H.

(c) Compare the three quantities:

P (H=T ), P (H=T | D=T ), P (H=T | D=T, F =T ).

What does this progression illustrate about evidence accumulation in Bayesian

reasoning?

1.4 Variable Elimination

Query:
P (H | S=T, D=T )
Elimination ordering: E, C, B, F

(a) Write the complete set of active factors after applying the evidence S=T and
D=T .

(b) Perform each elimination step in order. For every step state:

• which factors are selected,

• the multiplication performed,
• the variable summed out, and
• the scope (variable set) of the resulting factor.

(c) Write the final unnormalized expression for P (H | S=T, D=T ) and normalize
it to obtain P (H=T | S=T, D=T ) and P (H=F | S=T, D=T ).

(d) Determine the treewidth of this network under the elimination order E, C, B, F, S.

4
Question 2: Workplace Safety Bayesian Network
Note: Questions in this section refer to the Bayesian network shown in Figure 1
below.

E W M

S
A
R

D
N

Figure 1: Bayesian Network — Workplace Safety

Node key:

Node Meaning
E Employee Background
W Workmate Influence
M Management Commitment
A Safety Attitude
S Smoking Habit
D Drinking Habit
R Risk Perception
N Number of Accidents

Directed edges:

E → A, W → A, W → S, W → D, S → D, M → R, A → N, D → N, R→N

2.1 Conditional Independence via d-Separation

For each statement below:

• List all paths between the nodes.

• Classify each intermediate node and state whether each path is blocked or active.

• Give the final independence verdict.

5
 (a) Is Number of Accidents (N ) conditionally independent of Workmate Influence
(W ) given Safety Attitude (A), Smoking Habit (S), and Drinking Habit (D)?

N ⊥ W | {A, S, D}

(b) Is the following marginal independence true?

E⊥M

2.2 Conditional Independence of Nodes

Find all nodes that are conditionally independent of M (Management Commitment)
given A (Safety Attitude), D (Drinking Habit), and N (Number of Accidents).
Justify your answer using d-separation.

2.3 Variable Elimination

(a) Show the step-by-step process to perform variable elimination to compute:

P (N | E=Long, W =Positive, D=Drinking, R=Poor)

Use the following elimination ordering:

M, S, A

For each step state the factors involved, the product formed, the variable eliminated,
and the scope of the new factor.

(b) What is the treewidth of the network given the elimination ordering M, S, A?
Show the induced graph at each step.