Lunit INSIGHT CXR finds 13 resectable lung cancers incidentally

AI system

System: Lunit INSIGHT CXR
Model: Not assessed
Release: Lunit INSIGHT CXR versions 2 and 3.
Application developer: Lunit Inc.
Role: The software flagged suspicious nodules on chest X-rays for clinicians to review.
Configuration: The studies used versions 2 and 3 of the system.
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: Routine AI analysis of chest radiographs with results visible to ordering clinicians in real time. A prompt pulmonology referral pathway for AI-flagged nodules.
Human oversight: Radiologists also interpreted the images, and clinicians decided on consultations and further care.
Relation to AI: The software flagged all 13 nodules and its results were visible to ordering clinicians on PACS immediately after the radiograph. The authors assume from medical-record timestamps that the AI results aided the same-day consultations in 8 of 13 patients. They acknowledge that whether clinicians acted on the AI result or on the radiologist's report cannot be settled from the study. Radiologists also read the films. In 12 of 13 cases the radiologist's report described an abnormality at the site. There the contribution is earlier attention to findings nobody was looking for. In one (patient 12) the report read “No active lung disease”, so the AI flagged a cancer that the formal report missed.
AI contribution: Lunit INSIGHT CXR flagged all 13 nodules and made the results available to clinicians immediately after the chest X-rays. In one patient, it flagged a nodule that proved cancerous where the radiologist had reported no active lung disease.

Benefits

Outcome identifiers: Not assessed
Benefit: 13 (17.3%) had cancers found incidentally because the software flagged a nodule
Category: Improved Physical Health
Status: Not assessed
Beneficiary groups: People. Organizations.
Beneficiaries: Not assessed
Reach: Not assessed
Distribution and access: Not assessed
Duration: Not assessed
Expected?: Yes
Costs and harms: Not assessed

Measurement

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: 13 (17.3%) had cancers found incidentally because the software flagged a nodule
Comparison: Not assessed
Results: 13 (17.3%) had cancers found incidentally because the software flagged a nodule
Additional value from AI: Not assessed
Limitations: The authors inferred from timestamps that AI aided eight same-day consultations. They could not establish whether clinicians acted on the software’s result or the radiologist’s report. Radiologists described an abnormality at the same site in 12 of the 13 cases. Table 1 does not fully agree with the reported count of eight same-day consultations.
Denominators: Not assessed
Materials for replication: Not assessed

Money

Kind of value: Not assessed
Amount: Not assessed
Currency and prices: Not assessed
Period and unit: Not assessed
Realization: Not assessed
Recipients and payers: Not assessed
Gross or net: Not assessed
Included costs: Not assessed
Comparison: Not assessed
Method: Not assessed
Return and payback: Not assessed

People and places

Participants: Developer: Lunit Inc. Operator: Yongin Severance Hospital, Yonsei University College of Medicine.
Jurisdictions: Benefit: Yongin, South Korea. Developer: South Korea. Operator: South Korea.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Follow-up

Conditions for replication: Clinicians review AI flags on films ordered for non-pulmonary reasons. CT follow-up and tissue diagnosis for flagged nodules.
Replication attempts: Not assessed
Later findings: No later report on these 13 patients or on the Yongin Severance deployment appears in the cited paper.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Evidence

Source numbers: AIBD-0063-source-0001–0002
Citations: Numbered references below.
Language: Not assessed
Source relationships: Operator
Original evidence: Not assessed
Access and versions: Not assessed
Supported findings: Not assessed
Evidence links: Not assessed
Evidence strength: Not assessed
Missing evidence: Not assessed

Grades

Assessed benefits: People. Organizations.
Depth: People: 3. Organizations: 1.
Reasons: People: Thirteen patients had early, resectable lung cancer diagnosed and operated on. Seven were stage I and needed no further treatment. A survival benefit is plausible but not demonstrated in the paper. Organizations: Eight patients were seen by a pulmonologist on the day of the radiograph, before the formal report, shortening the diagnostic pathway.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: The deployer's account

Entry details

Kind

Not assessed

Related entries

Same system class and task, AI classification of routine chest radiographs: the classifiers studied showed gender, racial and socioeconomic bias that underdiagnoses patients, the group Lunit's incidental nodule flags benefited.

Field

Healthcare - thoracic radiology / lung cancer. AI applied to routine hospital chest radiographs.

Purpose

Lunit INSIGHT CXR is built to detect and localize abnormalities, including nodules, on chest radiographs. Incidental detection of lung cancer is its intended function.

Dates

Date basis unassessed: 2020 (precision unassessed). Updated: 21 October 2026.

Stage