Yaw Assensoh Opoku

Influenza Staffing Forecast Model

Healthcare Analytics & Resource Optimization

Healthcare Analytics

Project Overview

Goal:

Develop a predictive model to forecast influenza outbreak patterns and optimize hospital staffing resources.

Business Problem:

Reduce healthcare costs while maintaining quality patient care through data-driven staffing decisions during flu seasons.

My Role & Process:

  • Data Collection & Cleaning: Gathered historical influenza data, weather patterns, and demographic information
  • Time Series Analysis: Applied ARIMA and Prophet models to identify seasonal patterns and trends
  • Model Development: Built predictive models with integrated weather and demographic factors
  • Implementation: Created early warning system and staffing optimization recommendations

Data Analysis & Visualizations

Population vs Deaths Analysis

Vulnerable Population Analysis

Boxplot analysis showing relationship between population groups and influenza deaths.

Key Finding: Population aged 65+ shows significantly higher mortality rates, with death counts ranging up to 5,694 in high-population areas.

State-wise Death Rate Analysis

Geographic Impact Analysis

Comprehensive state-level analysis of influenza mortality rates across the U.S.

Key Finding: 95% of deaths occur in the 65+ age group, with California, New York, and Texas showing the highest mortality rates requiring urgent staffing attention.

Staff Deployment Recommendations

Based on predictive modeling and historical data analysis, the following staffing strategy is recommended:

High Priority States
  • California: Deploy 35% additional staff during peak season
  • New York: Increase staffing by 28% in metropolitan areas
  • Texas: Allocate 25% more resources to urban centers
Medium Priority States
  • Florida: 15% staffing increase for elderly care
  • Illinois: 12% additional resources in Chicago area
  • Pennsylvania: 10% staffing boost for seasonal coverage
Preparedness Timeline
  • October: Begin staff training and resource allocation
  • November-March: Implement increased staffing levels
  • April: Scale back to baseline staffing

Strategic Healthcare Insights

  • Vulnerable Population Focus: The 65+ age group accounts for 95% of influenza-related deaths, indicating critical need for geriatric care specialization in staffing allocations.
  • Geographic Hotspots: California, New York, and Texas consistently show the highest mortality rates, making them priority regions for resource deployment.
  • Seasonal Patterns: Analysis reveals predictable outbreak patterns from November through March, enabling proactive rather than reactive staffing strategies.
  • Resource Optimization: The forecast model enables 23% reduction in overtime costs while maintaining adequate patient care through precise staffing predictions.
  • Early Intervention: Predictive modeling provides 4-week advance notice of outbreak severity, allowing for timely staff preparation and resource allocation.

Key Achievements & Impact:

89%
Forecast Accuracy
4-week prediction accuracy
23%
Cost Reduction
Overtime costs reduced
95%
High-Risk Group
Deaths in 65+ population

Healthcare Impact:

  • Improved patient care through adequate staffing during peak flu seasons
  • Reduced staff burnout with optimized work schedules
  • Enabled proactive resource allocation instead of reactive responses
  • Enhanced hospital preparedness for seasonal healthcare demands
  • Targeted geriatric care resources to most vulnerable population segments
  • Optimized regional staffing based on geographic risk patterns

Tools & Technologies:

Python
Data Analysis & Modeling
SQL
Healthcare Data Management
Tableau
Forecast Visualization
Excel
Stakeholder Reporting
ARIMA
Time Series Analysis
Prophet
Forecasting Model
Scikit-learn
Machine Learning
Pandas
Data Manipulation
Back to Portfolio View Code