Project Motivation

The Motivation for AdmitGuard

AdmitGuard Intelligence is a full-stack data science platform designed to demonstrate modern machine learning integration in a secure, healthcare-style environment. It bridges the gap between raw clinical data and actionable administrative insights.

The Clinical Conundrum

Problem Overview

Reactive vs. Proactive Care

Hospital systems often rely on lagging indicators to identify at-risk populations. By the time a 30-day readmission occurs, the opportunity for preventative intervention has passed, resulting in poorer patient outcomes and significant financial penalties under the HRRP framework.

Data Necessity

Integrating Disparate Signals

Effective risk stratification requires the synthesis of demographics, historical diagnoses, laboratory results, and previous encounter metrics. Traditional rules-based engines fail to capture the complex, non-linear relationships inherent in multi-morbid patient populations.

Technical Note: Prototype Dataset Limitations

This platform serves as a resume-grade portfolio project showcasing end-to-end ML engineering. To ensure data integrity and avoid target leakage, this application uses strictly segregated, independent public datasets for each predictive task. The legacy "fused dataset" architecture has been fully deprecated.

diabetic_data.csv
Readmissions

Used exclusively to train the Logistic Regression model for 30-day readmission classification. Contains 100k+ clinical encounters.

healthinsurance_claims.csv
Financial

Used exclusively to train the Random Forest model for claim amount estimation. Focuses on demographic and lifestyle factors.