Jamovi Workflow Examples for Decision Tree and Markov Analysis

ClinicoPath Development Team

2025-06-30

Overview

This vignette provides step-by-step workflows for using the generated datasets in jamovi with the ClinicoPath decision analysis modules.

Load Data in jamovi

Step 1: Open jamovi and Import Data

1. Open jamovi
2. Import data file:
   • File → Open → Browse to:
   • appendicitis_decision_tree.csv (for decision tree)
   • heart_disease_markov.csv (for Markov chain)
   • Or use any of the test datasets in inst/extdata/

Decision Tree Analysis Workflow

Using: appendicitis_decision_tree.csv

Step 1: Navigate to Analysis

1. Navigate to: ClinicoPath → meddecide → Decision → Decision Tree Graph

Step 2: Configure Variables

Configure the following variables:

• Decision Nodes: treatment_choice
• Probability Variables: prob_surgery_success, prob_conservative_success
• Cost Variables: cost_surgery, cost_conservative, cost_complications
• Utility Variables: utility_success, utility_minor_complications
• Outcome Variables: clinical_outcome

Step 3: Set Tree Structure

• Tree Type: Cost-Effectiveness Tree
• Layout: Horizontal (Left to Right)

Step 4: Configure Display Options

Enable the following options:

• ☑ Show Node Shapes
• ☑ Show Probabilities
  
• ☑ Show Costs
• ☑ Show Utilities
• ☑ Show Node Labels
• ☑ Show Branch Labels
• Color Scheme: Medical Theme

Step 5: Configure Analysis Options

• ☑ Calculate Expected Values
• Sensitivity Analysis (optional)
• Discount Rate: 3%
• Time Horizon: 1 year

Step 6: Configure Output Options

• ☑ Summary Table
• Tornado Diagram (if sensitivity analysis enabled)

Expected Decision Tree Results

The analysis should produce:

• Decision tree visualization with nodes and branches
• Expected values table showing:
  • Strategy: Surgery vs Conservative
  • Expected Cost: ~$12,315 vs ~$7,454
  • Expected Utility: ~0.989 vs ~0.895 QALYs
  • ICER: ~$51,744 per QALY
  • Net Benefit: Varies by WTP threshold

Clinical Interpretation

• Surgery costs $4,861 more but provides 0.094 additional QALYs
• ICER of $51,744/QALY suggests surgery is marginally cost-effective
• Decision depends on patient factors and willingness-to-pay threshold

Markov Chain Analysis Workflow

Using: heart_disease_markov.csv

Step 1: Navigate to Analysis

1. Navigate to: ClinicoPath → meddecide → Decision → Decision Tree Graph

Step 2: Configure Variables

Configure the following variables:

• Decision Nodes: management_strategy
• Health States: management_strategy (or create state variable)
• Transition Probabilities: prob_asymp_to_symp, prob_symp_to_hf, prob_hf_to_death
• Cost Variables: cost_asymptomatic, cost_symptomatic, cost_heart_failure
• Utility Variables: utility_asymptomatic, utility_symptomatic, utility_heart_failure

Step 3: Set Tree Structure

• Tree Type: Markov Model Tree
• Layout: Horizontal (Left to Right)

Step 4: Configure Markov Options

• Cycle Length: 1 year
• Time Horizon: 20 years

Step 5: Configure Analysis Options

• ☑ Calculate Expected Values
• Discount Rate: 3%
• Time Horizon: 20 years

Step 6: Configure Output Options

• ☑ Summary Table
• ☑ Cohort Trace Plot
• ☑ Transition Matrix

Expected Markov Results

The analysis should produce:

• Markov Transition Matrix showing probabilities between states
• Markov Cohort Analysis showing population distribution over time:
  • Year 0: 100% Asymptomatic
  • Year 5: 54% Asymptomatic, 24% Symptomatic, 12% Heart Failure, 11% Dead
  • Year 20: 9% Asymptomatic, 7% Symptomatic, 16% Heart Failure, 68% Dead
• Cost-effectiveness results:
  • Total Lifetime Cost: ~$120,561
  • Total Lifetime QALYs: ~8.39
  • Cost per QALY: ~$14,370
• Markov State Transitions plot showing progression over time

Clinical Interpretation

• Standard care provides good value at $14,370/QALY
• Disease progression shows 68% mortality at 20 years
• Peak heart failure prevalence around year 15
• Results support cost-effectiveness of standard care

Comparing Strategies

For Decision Trees

• Compare expected values in the Summary Table
• Look for dominant strategies (lower cost, higher utility)
• Calculate ICERs for non-dominated strategies
• Use sensitivity analysis to test robustness

For Markov Models

• Run separate analyses for each strategy
• Compare lifetime costs and QALYs
• Calculate incremental cost-effectiveness ratios
• Examine cohort traces to understand disease progression

Sensitivity Analysis

Step 1: Enable Sensitivity Analysis

1. Enable ‘Sensitivity Analysis’ in Analysis Options
2. Enable ‘Tornado Diagram’ in Output Options

Step 2: Review Results

Results will show:

• Parameter ranges and their impact on outcomes
• Tornado diagram ranking parameters by influence
• Threshold values where conclusions change

Interpreting Results

Key Metrics to Report

• Expected costs (with confidence intervals)
• Expected utilities/QALYs
• ICERs with interpretation vs. thresholds
• Net benefit at relevant WTP thresholds
• Sensitivity analysis results

Cost-Effectiveness Thresholds

• < $50,000/QALY: Highly cost-effective
• $50,000-$100,000/QALY: Moderately cost-effective
• > $100,000/QALY: Not cost-effective (US standards)
• Thresholds vary by country and healthcare system

Reporting Results

Include in Publications

• Methods: Model structure, data sources, assumptions
• Results: Base-case cost-effectiveness results
• Sensitivity analysis: Key drivers and uncertainty
• Limitations: Model assumptions and data limitations
• Conclusions: Policy implications and recommendations

Visual Elements

• Decision tree or Markov model diagram
• Cost-effectiveness plane (cost vs. utility)
• Tornado diagram (for sensitivity analysis)
• Cohort trace plot (for Markov models)

Example Test Datasets Available

The following datasets are available for practice:

1. basic_decision_data.csv - Simple treatment comparison
2. markov_decision_data.csv - Multi-state disease progression
3. pharma_decision_data.csv - Drug comparison study
4. screening_decision_data.csv - Cancer screening programs
5. minimal_test_data.csv - Basic functionality testing
6. edge_case_data.csv - Error handling and edge cases
7. appendicitis_decision_tree.csv - Acute treatment decision
8. heart_disease_markov.csv - Chronic disease management

All datasets are located in: inst/extdata/

Load any of these files in jamovi to practice the analysis workflows.

Workflow Summary

This workflow covers:

• ✓ Data import and preparation
• ✓ Decision tree analysis configuration
• ✓ Markov chain analysis setup
• ✓ Result interpretation and reporting
• ✓ Sensitivity analysis implementation
• ✓ Clinical and policy interpretation

Additional Help

For additional assistance:

• Review the decision-tree-vs-markov-analysis.Rmd vignette
• Check the comprehensive vignettes in vignettes/
• Examine test data generation scripts in data-raw/
• Consult jamovi module documentation

Conclusion

You are now ready to perform sophisticated decision analysis and cost-effectiveness research with jamovi using the ClinicoPath module!

The workflows demonstrated in this vignette provide a systematic approach to:

• Setting up decision tree and Markov chain analyses
• Configuring appropriate variables and parameters
• Interpreting cost-effectiveness results
• Conducting sensitivity analyses
• Reporting findings for clinical and policy applications

Practice with the provided example datasets to build proficiency in these powerful analytical methods.