Rethinking Organ Allocation Policy with AI & Optimization
How advanced analytics guided a landmark U.S. transplant policy decision and what it means for healthcare leadership
This article was adapted from The Analytics Edge In Healthcare written by Dimitris Bertsimas, Ph.D., Agni Orfanoudaki, Ph.D., Holly Wiberg, Ph.D.
The Challenge: Fairness Meets Complexity in Organ Transplantation
Every year, thousands of patients wait for life-saving organ transplants. The U.S. transplant system must answer a deeply complex, ethically charged question: Who should receive a scarce organ first?
This isn’t just a medical issue. It’s a balancing act between:
Saving the sickest patients while ensuring long-term survival.
Distributing organs equitably across regions, races, and age groups.
Reducing transportation time and cost to protect organ viability.
Using scarce resources efficiently.
Since 1986, the Organ Procurement & Transplantation Network (OPTN) operated by the United Network for Organ Sharing (UNOS) has overseen this process under federal regulations known as the OPTN Final Rule. The Final Rule calls for allocation policies that use “sound medical judgment,” maximize the “best use” of donated organs, and promote equity across the country.
Yet translating these high-level ethical mandates into concrete policy has historically been slow, complex, and often controversial.
From Classifications to Continuous Distribution
For decades, organ allocation relied on a classification-based system: patients were grouped (e.g., local vs. distant, pediatric vs. adult, urgent vs. standard), and organs were offered sequentially within each group.
While practical, this approach faced criticism for geographic inequity. Hard boundaries often meant that a patient’s access to life-saving organs depended heavily on where they lived.
In 2018, the OPTN Board of Directors approved a landmark shift: the move to Continuous Distribution (CD).
Instead of discrete categories, each patient would now receive a Composite Allocation Score (CAS) – a single number blending multiple factors such as:
Medical urgency – likelihood of death without transplant.
Post-transplant survival – expected benefit after transplant.
Placement efficiency – logistical ease and cost of organ transport.
But the biggest challenge remained: how to weight each of these factors to balance ethical priorities and real-world constraints.
The Old Way: Slow, Opaque, and Resource-Heavy
Traditionally, the OPTN uses Simulated Allocation Models (SAMs) to predict how policy changes might affect outcomes like mortality, organ transport burden, and equity.
While powerful, these simulations are:
Time-consuming – each run can take weeks or months.
Limited – only a handful of scenarios can be tested.
Opaque – it’s hard for non-technical stakeholders to understand why results emerge.
As a result, policy development has often been iterative and slow. The Kidney Allocation System (KAS), for example, took nearly a decade and over 30 simulation rounds to finalize.
Breakthrough: Optimization + AI for Fairer, Faster Policy Design
Dynamic Ideas, the parent company of H2O, built a machine learning–powered optimization framework that transforms this process.
How It Works
Learning from Simulation Data
The company started by running a wide set of policy simulations to map different scoring weights to predicted outcomes (e.g., waitlist mortality, geographic equity, transport cost).Machine Learning Prediction
AI models were trained to accurately and instantly predict outcomes for any new policy by eliminating the need for time-consuming simulations every time weights are adjusted.Multi-Objective Optimization
Mathematical optimization was used to search across millions of possible scoring rules to find those that best balance competing goals (e.g., maximize survival while limiting transport burden).Interactive Tradeoff Exploration
Policymakers can define desired outcomes (e.g., “reduce waitlist deaths by 20% without exceeding a 10% increase in transport burden”) and instantly see policy configurations that achieve them. No coding or deep technical expertise required.
This shifts the process from guess-and-test to outcome-first policy design.
Real-World Impact: Guiding U.S. Lung Transplant Policy
Dynamic Ideas applied this framework in close collaboration with UNOS and the OPTN Lung Transplantation Committee to help shape the nation’s first Continuous Distribution policy for lung allocation, a decision that will influence who receives life-saving lung transplants in the U.S. for years to come.
Key Findings That Shaped National Policy
Placement Efficiency:
The tradeoff analysis identified 10% as the optimal weight for placement efficiency. This was the point where reducing transport burden further produced diminishing returns on waitlist mortality reduction.This finding proved robust across different transport metrics and was independently validated by additional simulation studies performed by the Scientific Registry of Transplant Recipients (SRTR) on updated 2018–2019 patient data.
Pediatric Access:
The company determined that a 15–20% weight on pediatric status was the minimum necessary to maintain high levels of access for children under the new system.
These findings were presented to the OPTN Lung Transplantation Committee in March 2021, validated by the SRTR through additional simulation, and ultimately adopted into the committee’s official proposal.
The final continuous distribution policy, published for public comment in August 2021 and approved by the OPTN Board in December 2021, uses:
10% weight for placement efficiency
20% weight for pediatric access
Life-Saving Impact
An optimized CD policy with a 10% placement efficiency weight is projected to:
Reduce waitlist mortality by ≈21% compared to the prior system, saving an estimated 118 additional lives over two years.
SRTR’s independent modeling suggests an even greater reduction of ≈40% waitlist mortality, potentially saving ≈175 lives over the same horizon.
At the same time, the new system eliminates rigid geographic boundaries, making organ access more equitable nationwide.
Beyond Organ Transplants: A Model for Better Health Policy
While this work focused on lung allocation, the methodology is broadly applicable wherever leaders must balance complex, competing objectives. Especially in resource-constrained, ethically sensitive healthcare settings, such as:
Resource allocation (e.g., ICU beds, ventilators).
Surgical scheduling and operating room utilization.
Value-based care reimbursement design.
Drug supply distribution during shortages.
By combining simulation, AI, and optimization, health systems are empowered to make smarter, more transparent decisions – improving outcomes, equity, and efficiency across care delivery.
Why It Matters
Healthcare leaders face mounting pressure to make data-driven, equitable, and efficient decisions often under conditions where delay severely impacts patients. This framework:
Reveals tradeoffs clearly, so stakeholders can make informed ethical choices.
Engages diverse voices, moving debates from technical parameters to values and outcomes.
Accelerates decision-making, cutting years off traditional policy development cycles.
The result is decision-making that’s smarter, fairer, and better aligned with patient needs and operational realities.
Dynamic Ideas’ methodology shows a path forward by demonstrating how AI and optimization can unlock smarter, more equitable, and more impactful decisions across healthcare.