Table of Content

1. Project Overview
2. Business Problem
3. Data Sources
4. Tools & Technologies
5. Key KPIs Tracked
6. Methodology / Analytics Workflow
7. Data Analysis & Findings
8. Predictive Modeling
9. Dashboard & Reporting Solution
10. Business Recommendations
11. Business Impact & Results
12. Stakeholder Communication
13. Project Challenges
14. Why This Project Matters

Executive Summary

A healthcare imaging provider was facing high MRI appointment no-show rates, scanner underutilization, and operational inefficiencies across multiple locations. By combining predictive analytics, operational KPI tracking, and Power BI dashboards, I helped reduce MRI no-shows by 27%, improve scanner utilization from 63% to 81%, and support approximately $1.2M in annual revenue recovery within six months.

Project Overview

Business Problem

A multi-site healthcare imaging provider was experiencing a growing operational and financial problem related to MRI appointment no-shows, last-minute cancellations, and inefficient scanner utilization.

Although MRI scanners are among the most expensive medical imaging assets in hospitals, many appointment slots were going unused due to:

• High patient no-show rates
• Poor appointment scheduling patterns
• Long patient waiting times
• Inefficient resource allocation
• Lack of predictive visibility into cancellation risks
• Manual reporting processes across departments

This problem resulted in:

• Revenue leakage from unused imaging slots
• Increased patient backlog and delayed diagnosis
• Overtime costs for radiology staff
• Reduced patient satisfaction
• Underutilization of expensive MRI equipment

The organization wanted a data-driven solution that could:

1. Predict patients likely to miss appointments
2. Optimize scanner scheduling efficiency
3. Improve operational planning
4. Provide real-time dashboards for management
5. Support better business decisions using data insights

Data Sources

The project involved integrating data from:

• RIS (Radiology Information System)
• PACS operational logs
• Appointment scheduling systems
• EHR patient demographics
• Billing systems
• Staff scheduling records

The dataset contained:

Data Category
Patient Data
Appointment Data
Operational Data
Financial Data
Staffing Data

Tools & Technologies

Tools Used:

• SQL
• Python
• Power BI
• Excel

Libraries Used:

• Pandas
• Scikit-learn
• Matplotlib
• NumPy

Key KPIs Tracked

Operational KPIs

• MRI scanner utilization rate
• Appointment no-show percentage
• Average patient waiting time
• Same-day cancellation rate
• Average scanner idle time
• Number of rescheduled appointments

Financial KPIs

• Monthly lost revenue from no-shows
• Revenue recovered after optimization
• Cost per idle scanner hour
• Overtime reduction percentage

Patient Experience KPIs

• Appointment lead time
• Average scheduling turnaround
• Patient satisfaction trend

Methodology / Analytics Workflow

To address MRI appointment no-shows and scanner underutilization, I followed a structured analytics and business optimization workflow:

1. Data Integration

Integrated operational and patient data from:

• RIS systems
• PACS logs
• EHR databases
• Scheduling systems
• Billing records

2. Data Cleaning & Preparation

Using Python and SQL, I:

• removed duplicate appointment records,
• standardized timestamps,
• handled missing scheduling statuses,
• validated operational data consistency.

3. Exploratory Data Analysis (EDA)

Performed trend and operational analysis to identify:

• high no-show appointment periods,
• scanner idle-time patterns,
• cancellation behaviors,
• location-specific operational inefficiencies.

4. KPI Definition & Tracking

Established operational, financial, and patient-experience KPIs, including:

• no-show rate,
• scanner utilization,
• patient wait time,
• revenue leakage,
• overtime costs.

5. Predictive Modeling

Developed a machine learning model to identify patients at high risk of missing MRI appointments using historical scheduling and behavioral data.

6. Dashboard Development

Built interactive Power BI dashboards for:

• executive leadership,
• operations management,
• finance teams,
  providing real-time operational visibility and forecasting insights.

7. Operational Optimization

Implemented data-driven scheduling improvements, including:

• automated reminders,
• waitlist slot replacement,
• dynamic scheduling adjustments,
• staffing optimization strategies.

8. Post-Implementation Monitoring

Tracked KPI improvements over a six-month period to measure operational and financial impact after implementation.

Data Analysis and Findings

1. No-Show Trend Analysis

After analyzing 18 months of appointment history, I identified patterns showing that:

• Monday morning appointments had the highest no-show rates
• Patients with previous cancellations were significantly more likely to miss appointments again
• Long waiting periods between booking and appointment increased cancellation risk
• Certain locations had consistently higher no-show patterns
• Weather and traffic conditions indirectly affected attendance in urban clinics

Business Insight

The organization was using a “one-size-fits-all” scheduling model without accounting for behavioral risk patterns.

2. MRI Scanner Utilization Analysis

Using operational logs and appointment schedules, I discovered:

• MRI scanners were only operating at approximately 63% effective utilization during business hours
• Certain scanners experienced high idle periods despite long patient backlogs
• Late cancellations created unfilled appointment gaps
• Scheduling teams lacked real-time visibility into available slots

Business Insight

The issue was not a lack of demand, but poor operational optimization and limited predictive planning.

Predictive Modeling

I developed a machine learning model using historical patient and appointment data to predict the probability of appointment no-shows.

Features Used

• Appointment lead time
• Previous cancellation history
• Distance from clinic
• Day of week
• Appointment time
• Insurance category
• Age group
• Seasonal trends

Model Outcome

The predictive model achieved:

• 84% prediction accuracy
• High-risk patient identification before appointment date
• Early intervention opportunities for scheduling teams

Dashboard & Reporting Solution

I designed interactive Power BI dashboards for leadership and operations teams.

Executive Dashboard

Displayed:

• Revenue leakage trends
• Scanner utilization rates
• Monthly operational efficiency
• Appointment risk forecasts
• Department performance comparison

Operations Dashboard

Displayed:

• Real-time appointment tracking
• High-risk patient alerts
• Scanner downtime monitoring
• Daily scheduling efficiency
• Staff workload distribution

Finance Dashboard

Displayed:

• Estimated revenue recovery
• Cost savings trends
• Overtime reduction
• Operational ROI metrics

Dashboard and Reporting Solution — showing operational KPIs, scanner utilization, no-show trends, predictive model accuracy, and financial impact before and after implementation.

Here’s a breakdown of every panel in the dashboard:

Executive KPI strip — six headline metrics showing the before/after story at a glance: no-show rate, utilization, recovered revenue, wait time, overtime, and model accuracy.

No-show rate by day of week — confirms Monday and Saturday as the highest-risk days (34% and 24%), directly supporting the recommendation to reduce appointment clustering on Mondays.

Scanner utilization before vs after — grouped bars for all four scanners, showing the jump from the 59–68% range up to 78–84% post-implementation.

Monthly no-show trend — a 18-month line chart split at month 13 (the intervention point), with the post-intervention decline shown in blue dashes against the pre-intervention rise in orange.

No-show risk scatter plot — simulates the predictive model’s core insight: longer booking lead times correlate strongly with higher no-show probability, with red/green color-coding to identify high vs low-risk patients.

Revenue impact bars — shows the crossover dynamic post-intervention: lost revenue shrinking and recovered revenue climbing each month.

Feature importance (horizontal bar) — visualizes the ML model’s input weights. Cancellation history (31.4%) and lead time (23.8%) are the dominant predictors, followed by distance and day of week.

Idle scanner hours heatmap — shows Monday AM and Friday PM as the biggest waste windows, guiding targeted scheduling adjustments.

Appointment Fill Rate- Appointment fill rate improved from 68% to 89% through waitlist optimization and predictive scheduling.

Business Recommendations

Based on the analysis, I recommended:

1. Predictive Scheduling Strategy

High-risk patients received:

• Automated reminders
• Confirmation calls
• Flexible rescheduling options
• Waitlist optimization

2. Dynamic Slot Reallocation

Unused MRI slots were automatically reassigned to waitlisted patients.

3. Optimized Scheduling Windows

High no-show time slots were adjusted using historical attendance patterns.

4. Operational Planning Improvements

Managers used utilization forecasts to:

• Reduce idle scanner time
• Optimize staffing schedules
• Minimize overtime costs

Business Impact and Results

Within 6 months of implementation:

Metric	Improvement
MRI no-show rate	MRI appointment no-show rate decreased from 22% to 16%, representing a 27% relative reduction.
Scanner utilization	Increased from 63% to 81%
Revenue recovery	Approximately $1.2M annually (Revenue recovery estimates were calculated using average completed MRI reimbursement rates and recovered appointment capacity).
Patient waiting time	Reduced by 22%
Staff overtime costs	Reduced by 18%
Scheduling efficiency	Improved significantly

Stakeholder Communication

Instead of only presenting charts and dashboards, I focused on translating data into practical business actions.

For example:

KPI Interpretation Example

Observation

Monday morning appointments showed a 34% higher no-show rate compared to mid-week appointments.

Business Interpretation

This indicated that scheduling patterns were contributing to operational inefficiencies.

Recommendation

I recommended:

• Reducing high-risk appointment clustering on Mondays
• Introducing automated reminder workflows
• Reserving flexible slots for urgent rescheduling

This helped management make operational scheduling decisions backed by data.

How I Communicated Insights to Non-Technical Stakeholders

One of the biggest challenges was ensuring that hospital leadership and operational staff understood the insights without technical complexity.

My Communication Approach

1. Simplifying Technical Language

Instead of saying:

“The predictive model achieved 84% classification accuracy using supervised machine learning.”

I explained:

“We can now identify patients who are most likely to miss appointments before the appointment happens, allowing scheduling teams to intervene early.”

2. Using Business-Focused Storytelling

I connected every insight to:

• Revenue impact
• Patient care improvement
• Operational efficiency
• Staff workload reduction

This helped leadership quickly understand the business value.

3. Visual Dashboard Walkthroughs

During stakeholder meetings:

• I used simple dashboard visuals
• Highlighted trends using color-coded KPIs
• Focused on actionable recommendations instead of technical metrics
• Presented before-and-after operational impact scenarios

Simple stakeholder dashboard showing MRI no-show problem, colour-coded KPIs, and before-and-after operational impact for non-technical audiences.

4. Supporting Decision-Making

I worked directly with radiology managers and operations teams to:

• Define measurable targets
• Prioritize operational improvements
• Align analytics with business goals
• Track post-implementation performance

Project Challenges

Challenge 1: Inconsistent Data Quality

Patient records contained:

• Missing appointment statuses
• Duplicate scheduling entries
• Inconsistent timestamps

Solution

I performed:

• Data cleaning
• Deduplication
• Standardization using Python and SQL

Challenge 2: Stakeholder Resistance

Some scheduling teams were hesitant to trust predictive recommendations.

Solution

I conducted:

• Training sessions
• Dashboard walkthroughs
• Pilot testing with measurable outcomes

This improved stakeholder confidence and adoption.

Why This Project Matters

This project demonstrates:

• Real-world healthcare analytics experience
• Business analysis and stakeholder management
• Predictive analytics application in healthcare operations
• Data storytelling and communication skills
• KPI-driven business decision support
• Dashboard development and reporting
• AI-driven operational optimization
• Cross-functional collaboration

It also shows the ability to move beyond basic reporting into delivering actionable business solutions with measurable operational and financial impact.

Let’s connect if you’re looking for data-driven healthcare analytics, operational optimization, or predictive BI solutions.