Skip to content

Quick Start: Your First Publication

Get from raw data to a draft manuscript in under 30 minutes.

Prerequisites

  • A MedTWIN account (sign up here)
  • Your clinical data in Excel (.xlsx) or CSV format

Step 1: Create a Project

  1. Click New Project from your dashboard
  2. Enter a project name (e.g., "Diabetes Outcomes Study")
  3. Select a template:
  4. Retrospective Cohort - Most common
  5. Randomized Controlled Trial
  6. Case-Control Study
  7. Blank - Start from scratch

Create Project

Step 2: Upload Your Data

Drag and drop your Excel file, or click to browse.

Supported formats:

Format Extensions
Excel .xlsx, .xls
CSV .csv
JSON .json

MedTWIN automatically:

  • Detects all sheets in your workbook
  • Profiles each column (type, missing values, unique counts)
  • Identifies potential patient IDs and date fields
📁 Your Upload
├── Sheet 1: "Demographics" (1,247 rows × 15 columns)
├── Sheet 2: "Lab Results" (8,432 rows × 12 columns)
└── Sheet 3: "Outcomes" (1,247 rows × 8 columns)

Step 3: Map Your Columns

This is where the magic happens. MedTWIN's AI suggests how to map your columns to standardized variables.

Review AI Suggestions

Each column shows:

Indicator Meaning
🟢 High Confidence AI is confident. Usually correct.
🟡 Medium Confidence Review recommended
🔴 Low Confidence Manual mapping needed

Example Mapping

Your Column → Standard Variable Confidence
pt_id → patient_id 🟢 High
dob → date_of_birth 🟢 High
hba1c_baseline → hba1c 🟢 High
outcome_30d → mortality_30day 🟡 Medium
misc_notes → unmapped —

Edit Mappings

Click any row to:

  • Change the target variable
  • Set as "Do not map"
  • Add transformation (e.g., convert units)

Validate & Commit

Click Validate to check:

  • ✅ Required fields are mapped
  • ✅ Data types are correct
  • ✅ No duplicate patient IDs
  • ✅ Date formats are parseable

Then click Commit Mapping to create your master dataset.

Step 4: Configure Your Analysis

Select Analysis Type

Type When to Use
Logistic Regression Binary outcome (yes/no)
Cox Regression Time-to-event with censoring
Linear Regression Continuous outcome

Define Variables

  1. Outcome Variable: What you're predicting (e.g., mortality_30day)
  2. Predictor Variables: What might predict it (e.g., age, hba1c, egfr)
  3. Covariates: Confounders to adjust for (e.g., sex, smoking_status)

Run Analysis

Click Run Analysis. MedTWIN will:

  1. Clean and validate your data
  2. Run the statistical model
  3. Generate results tables
  4. Calculate confidence intervals

Typical runtime: 10-60 seconds depending on data size.

Step 5: Generate Your Paper

Select Sections

Choose which sections to generate:

  • Abstract
  • Introduction
  • Methods
  • Results
  • Discussion
  • References

Review & Edit

Each section opens in the Paper Editor:

  • Block-based editing (like Notion)
  • Inline statistics that link to source data
  • Drag-and-drop tables and figures

Traceability

Hover over any statistic to see:

HR = 1.42 (95% CI: 1.18-1.71, p < 0.001)
├── Source: Cox regression, Run #3
├── Outcome: mortality_30day
├── Predictor: hba1c_baseline
└── Click to view full computation

Step 6: Export

Export Formats

Format Best For
Word (.docx) Journal submission
PDF Sharing, review
LaTeX Journals requiring LaTeX
Markdown Version control

Export Bundles

For reproducibility, export:

  • Analysis Bundle: Code + data + results
  • ETL Scripts: Data transformation code
  • Audit Trail: Complete action log

What's Next?

Need Help?