Cold Ischemia Foundation · Project Donor-Flow · 2025

The Intelligence Gap:
How AI Can End America's
Living Donor Crisis

17 Americans die every day waiting for a kidney transplant. More than 80% of willing donors never reach surgery — not because of medical disqualification, but because the evaluation system was designed to screen them, not support them. Project Donor-Flow deploys 8 AI technologies across 8 pipeline stages to close the gap that is costing lives.

17 Americans die daily on the kidney waitlist OPTN, 2024
>80% Of interested donors never reach surgery SRTR, 2023
2.15σ Current pipeline process sigma level CIF Analysis; Lentine et al., 2021
+40% Projected donor conversion rate improvement CIF Framework Projection
Section I — The Problem

A System That Fails the People It Was Built to Serve

The living donor crisis is not a shortage of willing donors. It is a failure of intelligence — the failure to deploy AI where it is now demonstrably capable of preventing preventable attrition at every stage of the pipeline.

💀
103,000+

Patients on the National Kidney Waitlist

As of March 2024, 103,000+ Americans await a kidney transplant. Pretransplant mortality is 5.0 deaths per 100 patient-years (2023). Among adults newly listed in 2018–2020, a combined 26% had died or been removed due to deterioration within three years of listing.

OPTN, 2024 · Lentine et al., 2025 (SRTR Fig. KI 22)
🔬
90% vs 82%

5-Year Graft Survival: Living vs. Deceased Donor

Five-year graft survival is 90.0% for LDKT vs. 82.2% for deceased donor transplant (ages 18–34). A 2024 UK propensity-matched cohort of 10,915 transplants found LDKT associated with 6.03% lower five-year graft failure risk (95% CI 4.71–7.35%).

Lentine et al., 2025 · Buse et al., 2024 (British Journal of Surgery)
⚠️
15.2%

Candidates Withdraw — Fully Eligible — Before Any Decision

The LDCPR (2,107 consecutive candidates, 10 programs) confirmed 15.2% of non-approvals were candidate withdrawal before any formal decision. An additional 10.3% were denied for psychosocial reasons the literature characterizes as addressable support needs — not inherent medical barriers.

Lentine et al., 2021 (AJKD 78:3) — LDCPR
📊
255,000

DPMO — Current Pipeline Process Quality

Preventable attrition (withdrawal 15.2% + psychosocial 10.3%) = 25.5% of non-approvals = 255,000 Defects Per Million Opportunities. The current pipeline operates at ≈2.15σ — more than 3 standard deviations below the minimum acceptable healthcare standard of 3σ.

CIF Analysis · Lentine et al., 2021
🌍
37%

Lower LDKT Likelihood for Black Recipients

A multicenter study of 19,287 recipients found Black recipients had 37% lower LDKT likelihood (aRR 0.63; 95% CI 0.59–0.67; p<0.001) independent of community vulnerability. Of all waitlisted Hispanic patients, only 5.2% received LDKT vs. 11.4% of non-Hispanic White patients.

Axelrod et al., 2021 (JAMA Surgery) · Waterman et al., 2022
🧠
The Gap

Data Without Intelligence = Continued Inaction

OPTN collects program-level data. SRTR publishes it. USRDS tracks ESRD prevalence. The LDCPR documents failure modes. The problem is not absent data — it is the absence of computational intelligence to act on that data at the speed, scale, and specificity required to prevent the attrition it predicts.

CIF Framework Analysis, 2025
Section III — The Architecture

The 8-Stage AI-Enabled Donor Pipeline

From pre-inquiry population identification to post-donation longitudinal monitoring — every stage deploys the specific AI technology most appropriate to its data type, regulatory environment, and interpretability requirements.

1

Predictive Donor Mapping

Identifies ZIP codes, faith institutions, FQHCs, and community touchpoints with the highest concentration of individuals whose ML-derived profiles predict elevated donation intent among underrepresented populations — before any individual self-identifies.

SUPERVISED ML
2

NLP Social Media Monitoring

Classifies publicly available social media content to identify pre-inquiry donation intent, classify motivation archetype, and route potential donors toward individualized outreach before they contact any transplant center.

NLP · BERT
3

AI-Enabled Concierge

Guides potential donors through pre-screening completion, manages scheduling friction, surfaces educational resources at motivationally appropriate moments, and provides proactive support when disengagement signals are detected.

CONVERSATIONAL AI
4

NLP Medical Record Review

Extracts and synthesizes relevant clinical signals from medical records into structured summaries for evaluating physicians — augmenting clinical review rather than replacing clinical judgment.

NLP · DOCUMENT AI
5

Donor Readiness Score (DRS)

A composite psychosocial-readiness metric identifying candidates at elevated risk of withdrawal before withdrawal occurs. Equity-adjusted to prevent encoding of historical disparate-denial patterns. Trained with mandatory human clinical review layer.

XGBOOST · EQUITY-ADJ.
6

Velocity Monitoring

Tracks each candidate's pipeline progress against the National Donor Velocity Benchmark, alerting coordinators to barrier signals before informed withdrawal becomes uninformed dropout.

CONTROL CHARTS
7

AI-Organized Donor Passport

Aggregates the candidate's complete clinical and psychosocial documentation into a structured, portable summary that eliminates repeated-documentation burden — a primary driver of process-fatigue withdrawal.

DOCUMENT AI · FHIR
8

Post-Donation Longitudinal Monitoring

ML models monitor post-donation renal outcomes against population-matched baselines, feeding outcome data into the Closed-Loop Refinement System that continuously retrains and rebalances the DRS.

ML · CLOSED-LOOP
Section IV — Process Engineering

Lean Six Sigma DMAIC Framework

Lean Six Sigma provides the engineering backbone — ensuring every AI deployment is organized around measurable outcomes, data-driven root cause analysis, and continuous improvement.

255,000
Current DPMO
≈2.15σ process quality
3+ std dev below acceptable
153,000
Projected DPMO at 24 Months
40% relative improvement
from full AI-PDLM deployment
<66,807
Target DPMO (3σ)
Healthcare minimum standard
74% reduction from baseline
Conversion Rate Model — CIF Framework 2025
C₀ = 0.18       // Baseline: 18% of inquiries reach surgery (SRTR, 2023)
C₁ = C₀ × (1 + Δ)  // Target after AI-PDLM deployment
Δ = 0.40           // Conservative weighted intervention effect
C₁ = 0.18 × 1.40 = 0.252 → 25.2% conversion rate at 24 months
State Intelligence — CIF Priority Targets

The Five Critical States

CIF analysis of OPTN state data, USRDS ESRD prevalence, and AKF Report Card legislative grades identifies five states requiring priority Project Donor-Flow deployment.

MS
AKF Grade F

Obesity: 40.8% (highest nationally)
Diabetes: 16.9% (highest nationally)
Zero improvement 2021–2023
No anti-discrimination law

LA
AKF Grade D

Obesity: 37.1% (2nd nationally)
No anti-discrimination law
No job protection
Critical decline trend

AL
AKF Grade D

Obesity: 36.3%
UAB nationally significant
Zero state-level support
Modifiable with the right system

TX
AKF Grade C

2nd most populous state
Hispanic LDKT: 5.2%
vs. White LDKT: 11.4%
Documented equity crisis

FL
AKF Grade C — improved

CIF Home · FL-16
2022 legislation passed
3rd most populous state
Marie's Lifeline Act target

Section VII — National Impact

What a 40% Conversion Improvement Means

Applying the projected conversion rate improvement to the current U.S. living donor base produces measurable, calculable national impact — conservative by design.

2,490

Additional Donors per Year

6,226 (2023 baseline) × 0.40 = 2,490 additional annual donors from improved conversion alone — before accounting for increased inquiry volume from AI-guided outreach at Stages 1–2.

8,716

Projected LDKT Volume at 24 Months

Up from 6,226 in 2023 — surpassing the 2019 all-time peak of 6,856 and exceeding HRSA's 2024 total of 7,030 within the first two deployment years.

1,158

Fewer Waitlist Deaths Annually

2,490 × (17 deaths/day ÷ 91,000 waitlisted) × 365 = 1,158 preventable deaths avoided per year. Conservative: constant inquiry volume, no Stage 1-2 uplift, no Year 2–3 compounding.

About the Cold Ischemia Foundation

Independence Is the Source of Authority

The organizations that dominate the national conversation about kidney disease receive substantial funding from the dialysis and pharmaceutical industries. CIF does not — by founding constraint, not circumstance.

Why Independence Matters

That independence eliminates the structural conflicts of interest that prevent other organizations from naming these failures and proposing solutions of this scope.

"The intelligence gap is real. The evidence base is complete. The framework is specified. What remains is institutional will — and the urgency of 17 deaths every day."

CIF advances Project Donor-Flow and Marie's Lifeline Compensation Act simultaneously — the only organization doing so from a fully independent position. AI optimizes the pipeline; legislation expands the population that enters it. The two interventions compound each other.

Zero pharma funding Zero dialysis funding Zero insurance funding Founding constraint Ellenton, FL · FL-16

Jeff Parke

Co-Founder & Executive Director

Three-time kidney transplant recipient. Author of Unbroken: Rising Above Chronic Kidney Disease. Jeff's lived experience as a patient provides primary-source insight into the systemic failures that published data documents but cannot fully explain. The LDCPR's finding that 50.7% of withdrawals cited "undisclosed" reasons is not a data gap to CIF — it is a description of shame, financial embarrassment, and overwhelm that the system's architecture produced and that Project Donor-Flow's design eliminates.

Marie Parke, RN

Co-Founder, Primary Author & Namesake

Registered nurse in radiation oncology. Namesake of Marie's Lifeline Compensation Act — the national legislative framework for direct living donor compensation modeled on New York State's 2022 law. Marie's clinical expertise in care delivery anchors the psychosocial, care partner, and post-donation dimensions of this framework.