📊 Patient Disenrollment Analysis

This repository contains an analysis of patient disenrollment from healthcare services. The goal of the project is to identify key factors driving disenrollment and provide actionable insights to improve member retention.

---

📌 Features

• Exploratory Data Analysis (EDA) to understand patterns in membership duration, health plans, and care interactions.
• Chi-Square and Correlation Analysis to identify statistically significant factors.
• Logistic Regression (single and multivariate) and Random Forest for predictive modeling.
• Actionable Insights and recommendations for reducing disenrollment rates.

---

📊 Key Analyses

1️⃣ Membership Duration

• Finding: Members with shorter durations are more likely to disenroll.
• Recommendation: Target newer members with engagement strategies in the first 3 years.

2️⃣ Health Plan Analysis

• Finding: Certain plans (e.g., "Humana") have higher disenrollment rates.
• Recommendation: Gather feedback and address plan-specific dissatisfaction.

3️⃣ Primary Care Provider (PCP) Changes

• Finding: Frequent PCP changes correlate with disenrollment.
• Recommendation: Improve continuity of care by minimizing unnecessary PCP changes.

4️⃣ Authorization Denials

• Finding: Members experiencing prior authorization denials are more likely to leave.
• Recommendation: Streamline authorization processes and educate members.

---

🛠 Methodology

1. Data Preprocessing
   • Handled missing values and scaled numerical variables.
   • Encoded categorical features like health plans.
2. Statistical Analysis
   • Used Chi-Square tests for categorical variables (e.g., health plans).
   • Correlation Analysis for numerical variables (e.g., membership duration).
3. Predictive Modeling
   • Applied Logistic Regression and Random Forest to predict disenrollment likelihood.
   • Evaluated models using ROC AUC, accuracy, and coefficients.

---

📈 Results

Model Performance

• Logistic Regression (Multivariate):
  • Accuracy: 74%
  • ROC AUC: 0.78
• Key Features:
  • Membership Duration (memberMonthsCount): Strong positive impact on retention.
  • PCP Changes: Frequent changes negatively impact retention.
  • Authorization Denials: Higher denial rates increase disenrollment likelihood.

---

📝 Reports

The following reports provide detailed insights:

• Disenrollment Analysis Report: Overview of key findings and recommendations.
• Enhanced Analysis Report: Focus on Random Forest feature importance.
• Patient Disenrollment Methodology: Step-by-step statistical and predictive modeling methodology.

---

🔥 Recommendations

1. Engage New Members: Prioritize members in their first 3 years with personalized outreach.
2. Focus on PCP Stability: Reduce unnecessary PCP changes to maintain care continuity.
3. Optimize Authorization Processes: Simplify prior authorizations to minimize dissatisfaction.
4. Plan-Specific Improvements: Gather feedback from members of high-disenrollment plans to address concerns.