Skip to content

Quickstart Guide

The RAPID Pipeline is designed as a modular and extensible structure for implementing various statistical and machine learning models for clinical research.

This guide will help you run your first analysis in minutes.

1. Core Principles and Modularity

The pipeline is built on a simple three-phase architecture, which is universal for all models:

Phase Method Role in the Pipeline
Phase 1: Initialization create() Selects the model type (e.g., Logistic, GLM, Survival) and configures parameters.
Phase 2: Training .fit() Performs the training procedure, including preprocessing and model fitting.
Phase 3: Diagnostics .summary() Generates performance metrics, assumption tests, and diagnostic plots.

This architecture ensures consistency across all models: create() → fit() → summary().

2. Prerequisites

Before you begin, make sure you have:

  • Installed RAPID
  • Python 3.10 or higher
  • A dataset ready for analysis (CSV, Excel, or DataFrame)

3. Your First Analysis (5 Minutes)

3.1 Import and Initialize

from isaric.pipelines.factory import RAPID_PipelineFactory

# Create a factory instance
factory = RAPID_PipelineFactory()

3.2 Load Your Data

import pandas as pd

# Load your dataset
df = pd.read_csv("your_data.csv")

# Example: Predict mortality using age, sex, and BMI
model = factory.create(
    "logistic",
    data=df,
    dependent_var="mortality",
    independent_vars=["age", "sex", "bmi"]
)

3.3 Train the Model

# Fit the model with optional labels for readability
model.fit(
    labels={
        "age": "Age (years)",
        "sex": "Sex",
        "bmi": "Body Mass Index"
    },
    cross_val=True,
    n_splits=5
)

3.4 View Results

# Display a comprehensive summary
model.summary(
    assumptions=["VIF", "EPV"],
    performance=["AUC", "AIC", "Accuracy"],
    plots=["forest_plot", "roc_curve"]
)

3.5 Access Results Programmatically

# Odds ratios and confidence intervals
odds_ratios = model.summary_df
print(odds_ratios)

# Performance metrics
metrics = model.performance_metrics_df
print(metrics)

# Confusion matrix
cm = model.cm
print(cm)

4. Supported Models

Model model_type Use Case
Logistic Regression "logistic" Binary outcomes (0/1)
GLM "glm" Generalized Linear Models
Survival Analysis "survival" Time-to-event data
MICE Imputation "mice" Missing data handling

Need Help?