Enterprise AI Analysis
Application of a machine learning model to predict the estimated primary care patient time consumption
This deep dive explores the Friedman Score, an ML-driven solution for predicting primary care utilization, enhancing proactive care planning, and optimizing resource allocation within healthcare systems.
Executive Impact
The Friedman Score offers a robust, interpretable tool for identifying high-utilization patients, guiding proactive, data-driven primary care delivery, and improving system resilience.
0%
Increase in Primary Care Utilization (2022-2023)
0
XGBoost Discriminative Ability (Max)
0x
Utilization Ratio (Very High vs Low Use Patients)
0+
Predictive Variables Used
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Overall Utilization
Model Workflow
Model Performance
Key Predictors
Proactive Care
Overall Utilization Increase
Between 2022 and 2023, total primary care time utilization increased significantly by 10%, reflecting growing demand, increasing chronic disease burden, and a post-COVID rebound in care-seeking. This underscores the rising pressure on primary care resources.
10%
Increase in Primary Care Utilization (2022-2023)
Predictive Model Workflow
The Friedman Score utilizes structured EHR data to predict estimated yearly primary care utilization categories. This data-driven approach allows for proactive care planning and targeted interventions by identifying high-utilization patients.
Enterprise Process Flow
Retrieve Primary Care Data & PHI Removal
→
Feature Engineering (39 Variables)
→
XGBoost Model Training & Calibration
→
Utilization Prediction (Low, High, Very High Use)
→
Clinical Workflow Integration & Proactive Outreach
Model Performance Across Algorithms
XGBoost demonstrated superior or comparable performance against five other machine learning algorithms, consistently showing high discriminative ability and robust calibration across both 2022 and 2023 test sets.
| Algorithm | 2022 AUC (Calibrated) | 2023 AUC (Calibrated) |
|---|---|---|
| XGBoost | 0.833 | 0.846 |
| Random Forest | 0.833 | 0.841 |
| Logistic Regression | 0.8062 | 0.791 |
| Support Vector Machine | 0.788 | 0.783 |
| Decision Tree | 0.814 | 0.818 |
| Multilayer Perceptron | 0.823 | 0.832 |
Key Predictors of Utilization
SHAP analysis identified medication usage, age, and chronic disease burden, particularly depression, as the most influential predictors of primary care utilization. This highlights the importance of comprehensive patient data in risk stratification.
21.3%
Patients with Depression (2023 Cohort), a significant predictor.
Proactive Care Redesign with Friedman Score
The Friedman Score's value lies in its integration into clinical workflows to proactively manage complex patient needs. A pilot program demonstrated enhanced care coordination and reduced provider burden through targeted outreach to high-utilization patients.
Case Study: UCSD Health Proactive Outreach Pilot
UCSD Health implemented a pilot where a dedicated NP and MA dyad use the Friedman Score to proactively outreach to Very High Use patients. This involves structured phone calls and messages to assess unmet needs (e.g., health literacy, medication access, mental health), confirm treatment understanding, and coordinate appropriate follow-up with the primary care team. This redesign aims to enable anticipatory care and reduce provider burden through targeted resource allocation.
- Targeted Outreach: NPs and MAs use the score to identify and engage Very High Use patients.
- Needs Assessment: Structured phone calls and messages assess health literacy, medication access, and mental health needs.
- Care Coordination: Ensuring understanding of treatment plans and coordinating appropriate follow-up.
- Anticipatory Care: Shifting from reactive to proactive care delivery.
- Reduced Provider Burden: Optimizing resource allocation for complex patient needs.
Calculate Your Potential ROI
Estimate the potential time and cost savings your organization could achieve by implementing an AI-driven utilization prediction model.
Estimated Annual Savings
$0
Reclaimed Hours Per Year
0
Our AI Implementation Roadmap
A clear, phased approach to integrating AI into your enterprise, ensuring maximum value and minimal disruption.
Phase 1: Discovery & Strategy
In-depth analysis of your current workflows, data infrastructure, and strategic goals to define AI opportunities.
Phase 2: Data Integration & Model Development
Securely integrate your EHR data, develop and validate custom AI models tailored to your specific needs.
Phase 3: Pilot & Workflow Integration
Implement a pilot program, integrate the AI solution into existing clinical workflows, and gather feedback.
Phase 4: Scaling & Continuous Optimization
Expand the solution across your organization, monitor performance, and continuously refine for peak efficiency.
Phase 5: Performance Monitoring & Iteration
Establish ongoing metrics and feedback loops to ensure the model remains accurate and beneficial, adapting to evolving healthcare landscapes.
Ready to Transform Your Primary Care Operations?
Leverage the power of AI to optimize resource allocation, reduce provider burden, and enhance patient care. Book a complimentary consultation to discuss how the Friedman Score can benefit your health system.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat