Lung cancer risk models show moderate to excellent discrimination, but few are ready for clinical useLung Cancer Risk Models Show Promise, Need Validation

European respiratory review : an official journal of the European Respiratory Society Published August 14, 2026 Study authors: Rezaeianzadeh Ramin, Leung Crystal, Kim Soo Jeong, Choy Kayly, Johnson Kate M, Kirby Miranda, Lam St… PubMed ↗ DOI ↗ Editorial oversight: Dr. Julia Lee, PhD · Oncology, Genomics & Drug Development

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Interpret lung cancer risk models cautiously; most lack external validation and are not ready for clinical use.

This systematic review and meta-regression evaluated risk prediction models for lung cancer screening selection and post-screening nodule classification. The authors identified 91 models from 2462 records, assessing their performance in terms of discrimination (AUC) and calibration. The review found that model discrimination was moderate to excellent, with AUC values ranging from approximately 0.70 to greater than 0.90. Models that incorporated biomarkers or imaging-enhanced features outperformed those without these components, suggesting potential added value from these data types.

However, the review also highlighted significant limitations. Calibration was inconsistently reported across studies, making it difficult to assess how well model predictions matched observed outcomes. Fewer than half of the models underwent external validation, which is critical for ensuring generalizability to new populations. The authors concluded that most models are insufficiently mature for clinical adoption due to uncertain performance and practical value outside their original study settings.

The meta-regression examined the impact of sample size, model type, validation type, and biomarkers on AUC, but specific effect sizes were not reported. The review did not report on adverse events or other safety outcomes, as it focused on model performance metrics.

Given these findings, clinicians should interpret the performance of lung cancer risk models cautiously. While many models show promise, they require more rigorous external validation and prospective implementation studies before they can be recommended for routine clinical use. The review underscores the need for standardized reporting of calibration and validation practices in future research.

How this fits prior evidence

This review extends prior coverage on lung cancer screening and diagnosis by focusing on risk prediction models. It complements earlier findings on diagnostic tools like rapid on-site evaluation (ROSE), which achieved 88.1% sensitivity and 91.7% specificity for lung cancer detection, by addressing the broader question of risk stratification. The moderate to excellent AUCs reported here align with the high accuracy of ROSE, but the review highlights a gap: many models lack external validation, unlike the well-validated ROSE. It also contrasts with prior coverage on prevention and management, such as prehabilitation and non-pharmacologic therapies, by emphasizing that prediction tools are not yet ready for clinical adoption.

A new systematic review looked at 91 risk prediction models designed to help decide who should get lung cancer screening and how to classify nodules found on scans. These models use factors like age, smoking history, and sometimes biomarkers or imaging features to estimate a person's risk.

The review found that many models performed moderately to excellently at telling apart people who would develop lung cancer from those who would not. Models that included biomarkers or imaging-enhanced features generally did better than those without. However, the review also revealed important gaps.

Fewer than half of the models were tested on outside groups of people, which is a key step to ensure they work in real-world settings. Also, calibration, which is how well the model's risk estimates match actual outcomes, was not consistently reported. This means it is hard to know if these models give accurate risk levels.

The authors conclude that most models are not yet ready for everyday clinical use. They need more external validation and prospective studies to prove their value. For now, these tools are promising but not a replacement for standard screening guidelines.

For patients and doctors, this means that while these models could improve screening decisions in the future, they are not yet proven enough to change current practice. More research is needed before they can be widely adopted.

What this means for you:

Lung cancer risk models show promise, but most need more testing before they can be used in clinics.

Common questions

What are lung cancer risk prediction models?

These are tools that use information like age, smoking history, and sometimes biomarkers or imaging features to estimate a person's chance of developing lung cancer. They help doctors decide who should get screened and how to manage nodules found on scans.

How well do these models work?

The review found that many models had moderate to excellent accuracy, with AUC scores ranging from about 0.70 to over 0.90. Models that included biomarkers or imaging features performed better than those without.

Are these models ready for use in clinics?

Not yet. Fewer than half of the models were tested on outside groups, and calibration was often not reported. The authors say most models are not mature enough for clinical adoption and need more external validation and prospective studies.

Study Details

Study typeMeta analysis

EvidenceLevel 1

PublishedJul 2026

PMID42419775

View Original Abstract ↓

BACKGROUND: Lung cancer (LC) remains the deadliest cancer, often diagnosed at advanced stages. Screening reduces mortality in high-risk individuals. Eligibility criteria in European and US screening guidelines have recently expanded. Therefore, we conducted an updated systematic review of risk-based models for identifying candidates for low-dose computed tomography screening and post-screening nodule classification. METHODS: We systematically searched Embase and Medline (January 2020-January 2026), identifying studies proposing new risk models in the context of LC screening. We separated models by pre- and post-screening risk stratification. Data extraction included study design, population, model type, risk horizon and model performance metrics. We performed an exploratory meta-regression of areas under the curve (AUCs) to assess whether sample size, model type, validation type and inclusion of biomarkers were associated with performance. RESULTS: Of 2462 records, 91 were included. 56 models were for screening selection (30 included biomarkers) and 35 for post-screening nodule classification. Regression-based models predominated, though machine-learning approaches were increasingly common. Discrimination ranged from moderate (AUC∼0.70) to excellent (>0.90), with biomarker and imaging-enhanced models often outperforming models without. Calibration was inconsistently reported and fewer than half underwent external validation. CONCLUSION: We identified 91 risk prediction models for LC, developed after 2020. Although many demonstrated promising discrimination across both screening selection and post-screening management, most remain insufficiently mature for clinical adoption, as their performance and practical value outside the original study setting are uncertain. Future work should prioritise external validation, updating and comparative evaluation of existing models, and prospective implementation studies rather than continued development of additional models.

Lung cancer risk models show moderate to excellent discrimination, but few are ready for clinical useLung Cancer Risk Models Show Promise, Need Validation

How this fits prior evidence

Common questions

Study Details

First-line immune checkpoint inhibitor combinations improve overall survival in extensive-stage small-cell lung cancer

Immune checkpoint inhibitors combined with chemotherapy improve small cell lung cancer survival

Clinical research that matters. Delivered to your inbox.

Lung cancer risk models show moderate to excellent discrimination, but few are ready for clinical useLung Cancer Risk Models Show Promise, Need Validation

How this fits prior evidence

Common questions

More on Lung Cancer

Study Details

First-line immune checkpoint inhibitor combinations improve overall survival in extensive-stage small-cell lung cancer

Immune checkpoint inhibitors combined with chemotherapy improve small cell lung cancer survival

Clinical research that matters. Delivered to your inbox.

Related in Oncology

From Other Specialties