Home

Overview

Personalized lung cancer screening requires advanced stratification algorithms

Lung cancer is a significant US health issue being the 2nd most frequently diagnosed cancer and the leading cause of cancer deaths.¹While breast cancer screening rates exceed 75%², lung cancer screening in high-risk populations has seen limited improvement.³Current screening guidelines largely exclude additional risk factors⁴, creating a need for population-enrichment algorithms to make screening more efficient and effective. LungFlag* is a computational software that uses clinical and electronic health record data to identify individuals at increased risk.⁵

High lung cancer burden

In the US, 229,410 new lung cancer cases and 124,990 deaths are projected for 2026.¹

Low screening compliance

Despite the proven benefits of low-dose computed tomography (LDCT), the screening rate for lung cancer remains low, averaging at 18% in the US.³

Need for stratification

Applying a population-enrichment algorithm allows for a high-incidence selection of individuals, focusing screening where it is needed most.⁵

navify LungFlag algorithm suite overview showing lung cancer screening workflow and clinical benefits.

How it works

Using clinical data to assess lung cancer risk

LungFlag is a computational software intended to aggregate, analyze, and organize health-related information from adults’ electronic health records at a population level to identify individuals who may be at higher risk of lung cancer. LungFlag analyzes information documented in electronic health records or collected by health care professionals, including demographics, lifestyle factors (e.g., smoking status), weight, body mass index (BMI), spirometry results, blood test results (including white blood cells and platelets) and documented information related to respiratory health.

Download brochure

Healthcare professional reviewing medical data on laptop in modern clinical office environment.

Risk-based pre-screening and re-engagement support

LungFlag can be used upstream of screening protocols to identify individuals at higher risk of NSCLC and support risk-based selection.⁶ This approach may help focus screening resources more efficiently.

LungFlag can also be applied to individuals who are not compliant with screening guidelines, providing personalized risk information that may support discussions with a physician about screening.⁷

Medical illustration of lung anatomy highlighting cancer detection and screening procedures.

Benefits

Supporting risk-based lung cancer screening strategies

LungFlag supports the identification of individuals at increased risk, including earlier stages of disease.^5,6 LungFlag has been evaluated in retrospective studies, including comparisons with PLCOm2012**, the most studied pre-screening algorithm.⁶ Analyses have also assessed performance across demographic groups and explored potential economic impact in healthcare settings: using LungFlag to identify candidates for lung cancer screening was estimated to save USD 2.8 million over 5 years from a US commercial health plan perspective, largely attributable to reduced costs of treating advanced NSCLC^***.⁸

X2 lung cancer detection rate

LungFlag demonstrated 53% sensitivity compared to 26.6% for USPSTF criteria, at a specificity of 91.4%.⁹

Flag to screen 3–12 months earlier

May help flag patients for screening, ahead of clinical diagnosis.⁶

$2.8M in savings over 5 years

Estimated from a US commercial health plan perspective, largely attributable to reduced costs of treating advanced NSCLC.⁸

Integration via navify Algorithm Suite

Experience a single integrated platform designed for healthcare providers and laboratories that simplifies IT complexity while reducing the risk of vulnerabilities. With integrated Roche and partner medical algorithms, we provide a comprehensive solution that streamlines processes and enhances collaboration.

"A medical flowchart showing the LungFlag Algorithm process from blood sample to lung cancer risk stratification.

A laptop screen displaying a digital patient dashboard with data tables, clinical information, and medical icons.

Single point of contact for customer support

Roche offers centralized customer support for all algorithms in our portfolio to ensure consistency and reliability. We manage issues that arise for all hosted algorithms, streamlining the process and eliminating the need for customers to engage with individual providers.

"A technical diagram on a laptop screen showing cloud infrastructure integration between healthcare providers and third-party algorithms.

Security and data privacy

Security and data privacy are central to Roche's operations, founded on a "Security and Privacy by Design" philosophy and ISO/IEC 27001 certification. Our dedicated technical team performs ongoing risk assessments, penetration tests, and network monitoring to minimize IT complexity and vulnerabilities, prioritizing data confidentiality to protect patient information across all partners.

FAQs

Frequently asked questions

If you don’t find answers to your questions here, we’re happy to provide more information and discuss your needs in detail.

Is LungFlag a diagnostic tool?

What data does LungFlag use?

LungFlag uses a limited set of parameters, such as age, sex at birth and smoking status. Additional optional clinical data may be used to increase the performance of the LungFlag model.

Optional inputs may include demographic information (e.g. height, weight), lifestyle risk factors (e.g. smoking history), BMI, spirometry data and selected blood test results. Providing additional parameters may support a more comprehensive risk assessment.

Data should reflect routine clinical measurements and exclude values from acute or inpatient settings, as these may not represent baseline health status.⁵

Who is LungFlag intended for?

LungFlag is designed for use in screening-eligible populations, specifically in individuals who are not compliant with existing screening recommendations. It was developed using data from ever-smokers adults aged 45 to 89 years.⁵

According to USPSTF guidelines, adults aged 50 to 80 years who have a 20 pack-year smoking history and currently smoke or have quit within the past 15 years are eligible for lung cancer screening.¹⁰

How is LungFlag implemented in clinical workflows?

LungFlag can be integrated into existing electronic medical record systems and applied to routinely collected data.

Lung-Flag, like all navify Algorithms on navify Algorithm Suite, can integrate into EHR (Electronic Health Record) or other clinical data repositories like LIS (Laboratory Information Systems). The integration strategy may vary depending on the customer's IT infrastructure. Please refer to your technical Roche representative to evaluate what integration strategy best fits your institution's infrastructure.

How does LungFlag handle data and model transparency?

Roche Diagnostics

At Roche Diagnostics, we understand the need to go beyond our products and see the bigger picture, creating a world where product and services come together. Roche Service Solutions refers to our service offering and encompasses our full service portfolio. Our total solution package means we can provide you with the personalised mix of products and services you need. Our digital services enable you to quickly connect with us and help make your life more manageable.

See all products Visit website

References and notes

American Cancer Society. Key statistics for lung cancer [Internet; cited April 2026]. Available from: https://www.cancer.org/cancer/types/lung-cancer/about/key-statistics.html
National Cancer Institute. Cancer trends progress report: breast cancer screening [Internet; cited April 2026]. Available from: https://progressreport.cancer.gov/detection/breast_cancer
Bandi P, Landy R, Star J, Kratzer TB, Smith RA, Jemal A. Lung cancer deaths prevented and life-years gained from lung cancer screening. JAMA. 2025;334:2225-2227. doi:10.1001/jama.2025.19798.
National Cancer Institute. Cancer trends progress report: lung cancer screening [Internet; cited April 2026]. Available from: https://progressreport.cancer.gov/detection/lung_cancer
LungFlag 1.0 User Guide V1.0. Material number 10787848001
Choman E, Lanyado A, Gould MK. PP01.03 Flagging high-risk individuals with a machine learning model improves non-small cell lung cancer early detection in a USPSTF-eligible population. J Thorac Oncol. 2023;18:e11. doi:10.1016/j.jtho.2022.09.029.
Goshen R, et al. JCO Clin Cancer Inform. 2018;2:1-8. DOI:10.1200/CCI.17.00130.
Heuser C, Gould MK, Choman E, Olghi N, Ogale S, Obradovic M. Budget impact of the LungFlag predictive risk model for lung cancer screening. Pharmacoecon Open. 2026;10(2):331-339. doi:10.1007/s41669-025-00627-6.
Gould MK, Huang BZ, Tammemagi MC, Kinar Y, Shiff R. Machine learning for early lung cancer identification using routine clinical and laboratory data. Am J Respir Crit Care Med. 2021;204(4):445-453. doi:10.1164/rccm.202007-2791OC.
U.S. Preventive Services Task Force. Final Recommendation Statement Lung Cancer: Screening [Internet; cited April 2026]. Available from: https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/lung-cancer-screening

Disclaimer

Not every digital product is available in all markets. The use of any third-party app is subject to a separate license agreement with the respective third-party app developer. Roche gives no warranties (express or implied) with regard to any third-party app. Third-party apps might not be available in your country. This website and its content may be accessible worldwide, Roche assumes no liability with regard to the access to the information, which may not be compatible with legislations or regulations in force in your country.
*LungFlag is not a diagnostic or screening tool. It must not be used as the sole basis for diagnosis or to waive recommended screening or diagnostic procedures. A positive result does not indicate respiratory illness or malignancy. A negative result does not rule out risk, screening or further testing. The product does not preclude other medical conditions. The product shall only be used by healthcare professionals within healthcare organizations. The health care professional should review the result in conjunction with their clinical judgment, the patient’s family medical history and other risk factors. The product should not be used on individuals already in a screening process or patients unable to undergo potential follow-up actions.
**PLCOm2012: PLCOm2012 Logistic Regression model adapted to work on EMR data.
***The cost savings may not reflect the typical purchaser’s experience, and is not intended to represent or guarantee that anyone will achieve the same or similar results.