Scientific evidence of commercial artificial intelligence products for pulmonary nodule assessment on CT scans: a systematic review
Medical AI tools for lung cancer screening are increasingly studied at higher levels of clinical evidence, though funding bias and bias risk remain widespread concerns for real-world deployment.
This Erasmus MC systematic review maps the evolution of evidence for FDA-cleared and CE-marked AI tools for pulmonary nodule CT assessment using the RADAR efficacy hierarchy, finding that by 2024 higher-order efficacy studies (clinical thinking, therapeutic, patient outcomes) account for over a third of literature, up from near-zero before 2018. Critical gaps remain: all 95 studies have high bias risk in at least one domain, nearly two-thirds involve vendor funding, and Level-5 patient outcome evidence is minimal—raising regulatory and real-world deployment questions.
What the study was
- Study design
- systematic_review
- Category
- ai_ml_clinical_diagnostics
- Maturity
- Validated
- Journal
- European Radiology
Why it surfaced
Comprehensive evidence map for a major commercial AI application (pulmonary nodule CAD) with direct regulatory implications; identifies systemic evidence quality problems relevant to Pulse's AI diagnostics monitoring. Score 7 (N=2, D=2, P=1, E=2) reflecting rigorous systematic review with actionable regulatory insights.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.