The Core of Clinical Research Design in 3 Moves: Object, Method, Analysis
- Mayta
- May 5
- 2 min read
1️⃣ Object Design: What’s the Clinical Problem?
Use the DEPTh Model to define your study’s core question:
DEPTh Type | Clinical Aim | Example Question |
Diagnosis | Identify disease | “Can symptoms predict TB?” |
Etiognosis | Find a cause | “Does drug X cause liver injury?” |
Prognosis | Predict future outcomes | “What predicts ICU stay after surgery?” |
Therapeutic | Assess treatment effect | “Does intervention A reduce readmission?” |
Methodologic | Improve measurement tools | “Is test B reproducible in clinical use?” |
2️⃣ Method Design: How Will You Study It?
Break the object into method components:
Element | What It Does |
Study Domain | Clinical population & setting |
Study Base (3x3x3) | Who, where, and when the sample is drawn |
Determinants (X) | Risk factors, exposures, or treatments |
Endpoints (Y) | What happens (outcome, diagnosis, recovery, etc) |
Clinical Parameters | Time, rates, odds, mean change |
Occurrence Equation | Links everything in one analytic model |
✳️ Occurrence Equation
Model your outcome as a function of causes or predictors:
Where:
Y = Outcome (event, diagnosis, measurement)
Xs = Main exposures, treatments, or predictors
Confounders = Other variables that distort true relationship
Bias = Selection, measurement, or analysis artifacts
Random error = Chance variation
3️⃣ Analysis Design: How Will You Analyze It?
Match the analytic method to the DEPTh type:
Logic Type | Used For | Goal |
Describe | All studies | Summarize population/data |
Explain | Etiognostic, Therapeutic | Identify causal effects |
Explore | Early phase, hypothesis-gen | Discover new relationships |
Predict | Diagnostic, Prognostic | Forecast who gets what/when |
🛠 Common tools:
Describe: Means, medians, proportions
Explain: Logistic/Cox/Linear regression
Predict: Risk scores, ML models, ROC/AUC
✅ Quick Recap
DEPTh = Your research compass 🧭
Occurrence Equation = Your modeling skeleton 💀
Design Triad:
Object = What are you solving?
Method = How will you study it?
Analysis = How will you prove or predict?
🧪 Practice Prompt
Try filling in this template for your next idea:
DEPTh Type:
Clinical Question:
Y =
f(Xs | Confounders)
Analysis Goal: Describe / Explain / Explore / Predict
Comments