Power BI Tutorial: Analyzing Survival Patterns in Heart Disease Patients

Categories: Power BI

Tags:

Power BI Tutorial: Analyzing Survival Patterns in Heart Disease Patients

By Ankit Srivastava – Data Analyst & BI Consultant

Data visualization can turn complex medical data into life-saving insights. In this Power BI project, I’ve analyzed heart failure clinical records from a Kaggle dataset to identify the key factors influencing patient survival.

This dashboard, titled “Survival Patterns in Heart Disease Patients,” visually explores relationships between multiple clinical indicators such as age, anaemia, diabetes, blood pressure, serum creatinine, and more — all affecting patient outcomes.

🎯 Objective

The primary goal of this dashboard is to analyze and visualize how different health parameters affect survival among patients diagnosed with heart disease.

Instead of just looking at individual metrics, we used Power BI’s interactive visuals and DAX measures to calculate survival-related KPIs and highlight patterns between variables.

The target field is DEATH_EVENT, a Boolean indicator where:

1 = Patient Died
0 = Patient Survived

📘 Dataset Overview

We’re using the Heart Failure Clinical Records Dataset, available on Kaggle:
🔗 Predict Survival of Patients with Heart Failure

It contains 12 clinical attributes and one target field (DEATH_EVENT).

Column	Description	Type / Unit
age	Patient’s age	Years
anaemia	Decrease of red blood cells (1 = Yes, 0 = No)	Boolean
creatinine_phosphokinase (CPK)	CPK enzyme level	mcg/L
diabetes	If patient has diabetes	Boolean
ejection_fraction	Blood percentage pumped with each heartbeat	%
high_blood_pressure	If patient has hypertension	Boolean
platelets	Platelets in the blood	kiloplatelets/mL
sex	0 = Female, 1 = Male	Binary
serum_creatinine	Level of serum creatinine	mg/dL
serum_sodium	Level of serum sodium	mEq/L
smoking	If patient smokes	Boolean
time	Follow-up period	Days
DEATH_EVENT	Target variable (1 = Died, 0 = Survived)	Boolean

🧮 Step 1: Importing and Cleaning Data

Open Power BI Desktop.
Click Get Data → CSV and load the dataset.
Rename the table to heart_failure_clinical_records_dataset.
Verify that all Boolean columns (anaemia, diabetes, etc.) are correctly recognized as Whole Numbers.

To enhance readability:

Replace 0 and 1 in sex with “Female” and “Male.”
Create an Age Group column for better visualization.

👉 Calculated Column (Age Group):

Age_group = 
SWITCH(
    TRUE(),
    heart_failure_clinical_records_dataset[age] < 40, "Below 40",
    heart_failure_clinical_records_dataset[age] < 50, "40-50",
    heart_failure_clinical_records_dataset[age] < 60, "51-60",
    heart_failure_clinical_records_dataset[age] < 70, "61-70",
    "71+"
)

📊 Step 2: Creating DAX Measures

We’ll create a few key DAX measures to summarize survival statistics.

1️⃣ Survival Rate (%)

Survival Rate = 
DIVIDE(
    CALCULATE(
        COUNTROWS('heart_failure_clinical_records_dataset'),
        'heart_failure_clinical_records_dataset'[DEATH_EVENT] = 0
    ),
    COUNTROWS('heart_failure_clinical_records_dataset'),
    0
)

2️⃣ Average Age of Survivors

Average_Age_Survival = 
CALCULATE(
    AVERAGE('heart_failure_clinical_records_dataset'[age]),
    'heart_failure_clinical_records_dataset'[DEATH_EVENT] = 0
)

3️⃣ Total Survived

Survived = 
CALCULATE(
    COUNTROWS('heart_failure_clinical_records_dataset'),
    'heart_failure_clinical_records_dataset'[DEATH_EVENT] = 0
)

4️⃣ Total Died

Died = 
CALCULATE(
    COUNTROWS('heart_failure_clinical_records_dataset'),
    'heart_failure_clinical_records_dataset'[DEATH_EVENT] = 1
)

5️⃣ Gender Label

Gender = IF('heart_failure_clinical_records_dataset'[sex] = 0, "Female", "Male")

These measures will power our KPI cards and comparison visuals.

🎨 Step 3: Designing the Dashboard Layout

To make the report visually appealing, I designed the dashboard background in Canva.
It includes:

A heart illustration on the right
Icons for gender representation
A professional layout with space for 4 KPI cards and 8 visuals

Import the background:

Go to Insert → Image → Set as Canvas Background

🧠 Step 4: Adding Key Metrics

At the top of the dashboard, place KPI cards for:

Survival Rate
Average Age (Survived)
Total Survived
Total Died

Each KPI uses the measures created above.
You can customize card visuals with teal or clinical red colors for contrast.

Example:

67.89% Survival Rate
58.76 Years Avg Age of Survivors
203 Survived
96 Died

📈 Step 5: Visuals and Insights

Now we’ll add visualizations to analyze how survival changes with different parameters.

🔹 1. Survival Rate by Age Group

Visual Type: Clustered Column Chart

Axis: Age_group
Values: Survival Rate
Insight: Patients between 40–60 years show higher survival (≈75%), while rates drop sharply after 70.

🔹 2. Survival Rate by Anaemia

Visual Type: Donut Chart

Legend: anaemia
Value: Survival Rate
Insight: Anaemia reduces survival likelihood, as oxygen transport efficiency decreases.

🔹 3. Survival vs. CPK Levels

Visual Type: Combo Chart (Bar + Line)

X-axis: creatinine_phosphokinase
Left Y-axis: Survival Rate
Right Y-axis: Survived
Insight: Higher CPK spikes (above 5000 mcg/L) correspond to poor survival, indicating cardiac muscle stress.

🔹 4. Survival Rate by Platelets

Visual Type: Bar Chart

Axis: platelets
Value: Survival Rate
Insight: Both very high and very low platelet counts are linked to lower survival probability.

🔹 5. Survival by Diabetes

Visual Type: Donut Chart

Legend: diabetes
Values: Survival Rate
Insight: Diabetic patients show slightly reduced survival, reflecting cardiovascular strain.

🔹 6. Survival by Ejection Fraction

Visual Type: Scatter Plot with Trendline

X-axis: ejection_fraction
Y-axis: Survival Rate
Insight: A clear positive correlation — patients with ejection fraction > 50% have the highest survival rate.

🔹 7. Survival by High Blood Pressure

Visual Type: Donut Chart

Legend: high_blood_pressure
Values: Survival Rate
Insight: Hypertensive patients have lower survival rates, emphasizing the importance of BP control.

🔹 8. Survival by Serum Creatinine

Visual Type: Column Chart

X-axis: serum_creatinine (binned)
Y-axis: Survival Rate
Insight: As serum creatinine increases, survival decreases — indicating renal dysfunction as a major risk factor.

🧩 Step 6: Gender-Based Visual Segmentation

Create slicers for:

Gender
Diabetes
High Blood Pressure

This allows viewers to filter patterns for male/female patients or comorbidities.

⚙️ Step 7: Interactivity and Design Polish

Align visuals symmetrically for readability.
Use consistent color palettes (dark maroon for low survival, pink for higher rates).
Add subtle drop shadows for modern look.
Enable cross-filtering so that clicking on any visual updates others dynamically.

📚 Step 8: Interpretation of Results

After building this dashboard, we can infer that:

Patients aged below 60 have significantly better outcomes.
Anaemia, diabetes, and high blood pressure directly correlate with higher mortality.
Ejection fraction is the strongest positive predictor of survival.
High serum creatinine and CPK indicate poor survival prognosis due to heart and kidney stress.
Gender differences are minimal but present in older cohorts.

🔍 Step 9: Export and Publish

Save your Power BI file.
Publish to Power BI Service for sharing.
Optionally, export it as a PDF or image to share on LinkedIn or your portfolio.

💬 Step 10: Learning Outcome

This project beautifully combines data analytics with clinical understanding.
Using Power BI and DAX, we turned raw medical data into a decision-support dashboard that can help researchers and healthcare professionals identify key survival factors.

It demonstrates:
✅ How to preprocess and model health data
✅ How to calculate DAX-based KPIs
✅ How to design clear, professional dashboards
✅ How to interpret visual patterns

🧠 Bonus Tip

You can extend this dashboard in Part 2 by:

Using Power BI Q&A for natural language insights
Predicting survival using Machine Learning models integrated with Python visuals
Adding trend analysis by follow-up time

🎨 Final Thoughts

Data visualization is not just about numbers — it’s about saving lives through insights.
This Power BI dashboard is a perfect example of how analytics bridges the gap between data science and medical research.

By using publicly available data, anyone can learn to tell impactful stories with Power BI.

🧩 Created by: Ankit Srivastava
💼 Data Analyst | Power BI | Python | Tableau
🌐 Learn more tutorials at SlideScope.com