Clinical Evidence

The Data Behind
LungAI

Peer-reviewed performance benchmarks, external multi-center validation, and an active academic study program at Mass General Brigham, Mount Sinai, and NYU Langone.

Pre-FDA Clearance — All data is retrospective. LungAI has not yet received FDA 510(k) clearance.

ROC Curve — Malignancy Detection (NLST)
0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00 1 − Specificity (False Positive Rate) Sensitivity (True Positive Rate) LungAI — 0.96 Lung-RADS — 0.88 LungAI (Balanced) AUROC 0.96 Lung-RADS v2022 AUROC 0.88
Internal Validation · NLST Test Set

Head-to-Head Against the Standard of Care

LungAI was benchmarked against Lung-RADS v2022 on a held-out test set from the National Lung Screening Trial (NLST). Lung-RADS annotations provided by board-certified radiologist Bradley Mullen, MD, Oatmeal Health Medical Advisory Board.

Model AUROC Sensitivity Specificity Advantage
LungAI — High Sensitivity State of the Art
 0.96 94% 82% +3% sensitivity
LungAI — Balanced
 0.96 93% 84% +2% / +2%
LungAI — High Specificity
 0.96 91% 88% +6% specificity
Lung-RADS v2022 Current Standard
 0.88 91% 82% Baseline
Competitive Landscape

The CADx Landscape

The only CADx platform with modern AI architecture, temporal modeling, full-funnel FQHC distribution, and net-new reimbursed revenue for hospital partners.

Company Use Case FDA Status ML Architecture Temporal Modeling AUROC
Oatmeal Health LungAI CADx
LDCT Screening Pending 3D Vision Transformer + Sybil + Dense Decoder (2026 AI) ✓ YES 0.96
Optellum Virtual Nodule Clinic
Incidental CT only
(Cannot market for screening)		 Cleared Mar 2021 DenseNet CNN (Legacy) ✗ NO 0.867
RevealDx RevealAI-Lung
Incidental CT only
(Cannot market for screening)		 Cleared Feb 2026 Radiomic Features + Traditional ML (Legacy) ✗ NO 0.89
Median Technologies eyonis LCS
LDCT Screening Cleared Feb 2026 2D/3D CNN (Legacy) ✗ NO 0.904

* AUROC figures are sourced from each company's FDA 510(k) submission or published validation study. Studies differ in patient population, dataset, CT acquisition protocol, and methodology and are not directly comparable. Oatmeal Health's AUROC of 0.96 reflects internal validation on a held-out NLST test set and is pending FDA review.

Academic Partners

Tier-1 Institutional Collaborators

Our clinical validation program is led by recognized authorities in thoracic radiology and pulmonology at three of the nation's leading academic medical centers.

Mount Sinai Health System · I-ELCAP
Study focus: Small nodule performance (<6mm and <8mm), photon-counting CT subgroup analysis, I-ELCAP longitudinal cohort.
Small nodule specialist Photon-counting CT
NYU Langone Health
Study focus: Lung-RADS stratum-specific AUROC validation, demonstrating LungAI utility across the full risk spectrum from low-risk (2/3) through high-risk (4X).
Lung-RADS strata 2–4X
Mass General Brigham · Harvard Medical
Study focus: Leads the pivotal MRMC reader study, the FDA-required demonstration that LungAI improves physician clinical decision-making.
6 readers Randomized crossover
For Imaging Centers and Health Systems

Ready to validate this with your own data?

We partner with imaging centers and health systems to pilot LungAI on your patient population.
Integrates into your existing CADe workflow and generates reimbursable AI overreads on every scan.

~$650/scan under CPT 0721T
Integrates into your existing CADe
We refer the patients
Hospitals bill directly
See how it works → Talk to our clinical team ↗