Deep Learning can Predict Lung Cancer Risk from Single LDCT Scan

A deep learning model was able to predict future lung cancer risk from a single low-dose chest CT scan, according to new research published at the ATS 2025 International Conference.

The model, called Sybil, which was originally developed using National Lung Screening Trial (NLST) data by investigators from the Massachusetts Institute of Technology and Harvard Medical School, could be used to guide more personalized lung cancer screening strategies. It could be an especially valuable tool in Asia, where incidence of lung cancer in nonsmokers is rising, and many people without conventional risk factors don’t meet screening guidelines, researchers said.

"Sybil demonstrated the potential to identify true low-risk individuals who may benefit from discontinuing further screening, as well as to detect at-risk groups who should be encouraged to continue screening," said corresponding author Yeon Wook Kim, MD, a pulmonologist and researcher at Seoul National University Bundang Hospital in Seongnam, Republic of Korea.

Current international guidelines do not recommend lung cancer screening for people considered lower-risk, such as individuals who have never smoked. However, lung cancer rates are rising in this group, and the lung cancer burden in this population is significant.

This disconnect between risk and screening is especially a concern in Asia. The region accounts for more than 60 percent of new lung cancer cases and related deaths globally, with a rising incidence among people who have never smoked, Dr. Kim said. He also noted that the epidemiology of lung cancer in Asia is different from the populations where screening criteria were developed and validated. This has led to an increase in screening that is self-initiated or not consistent with guidelines, but there's a lack of data to suggest who should be screened and who should not.

For the new paper, researchers evaluated more than 21,000 individuals aged 50-80 who underwent self-initiated LDCT screening between 2009 and 2021 and followed their outcomes until 2024. The screening tests were analyzed by Sybil to calculate the risk of future lung cancer diagnosis. The model demonstrated good performance in predicting cancer diagnosis at both one and six years, including in never-smokers.

"Sybil's value lies in its unique ability to predict future lung cancer risk from a single LDCT scan, independent of other demographic factors that are conventionally used for risk stratification," Dr. Kim said.

The model could be used to develop personalized strategies for individuals who have already undergone LDCT screening but have not received further recommendations for additional screening or follow-up. Prospective validation will be needed to confirm the model's potential for clinical use.

Researchers plan to follow up on the study.

"Based on our results, we are eager to conduct a prospective study to further validate and apply Sybil in a pragmatic clinical setting, as well as to enhance the model's ability to predict other important outcomes, such as lung cancer-specific mortality," Dr. Kim said.

YW Kim, J Oh, J Park, M Kim, DY Kim, JG Nam, D-H Joo, C-T Lee.
Validation of Sybil Deep Learning Lung Cancer Risk Prediction Model in Asian High- and Low-risk Individuals.
Am J Respir Crit Care Med 2025;211:A5012. doi: 10.1164/ajrccm.2025.211.Abstracts.A5012

Most Popular Now

Philips Foundation 2024 Annual Report: E…

Marking its tenth anniversary, Philips Foundation released its 2024 Annual Report, highlighting a year in which the Philips Foundation helped provide access to quality healthcare for 46.5 million people around...

New AI Transforms Radiology with Speed, …

A first-of-its-kind generative AI system, developed in-house at Northwestern Medicine, is revolutionizing radiology - boosting productivity, identifying life-threatening conditions in milliseconds and offering a breakthrough solution to the global radiologist...

Scientists Argue for More FDA Oversight …

An agile, transparent, and ethics-driven oversight system is needed for the U.S. Food and Drug Administration (FDA) to balance innovation with patient safety when it comes to artificial intelligence-driven medical...

New Research Finds Specific Learning Str…

If data used to train artificial intelligence models for medical applications, such as hospitals across the Greater Toronto Area, differs from the real-world data, it could lead to patient harm...

Giving Doctors an AI-Powered Head Start …

Detection of melanoma and a range of other skin diseases will be faster and more accurate with a new artificial intelligence (AI) powered tool that analyses multiple imaging types simultaneously...

AI Agents for Oncology

Clinical decision-making in oncology is challenging and requires the analysis of various data types - from medical imaging and genetic information to patient records and treatment guidelines. To effectively support...

Patients say "Yes..ish" to the…

As artificial intelligence (AI) continues to be integrated in healthcare, a new multinational study involving Aarhus University sheds light on how dental patients really feel about its growing role in...

Brains vs. Bytes: Study Compares Diagnos…

A University of Maine study compared how well artificial intelligence (AI) models and human clinicians handled complex or sensitive medical cases. The study published in the Journal of Health Organization...

'AI Scientist' Suggests Combin…

An 'AI scientist', working in collaboration with human scientists, has found that combinations of cheap and safe drugs - used to treat conditions such as high cholesterol and alcohol dependence...

Start-ups in the Spotlight at MEDICA 202…

17 - 20 November 2025, Düsseldorf, Germany. MEDICA, the leading international trade fair and platform for healthcare innovations, will once again confirm its position as the world's number one hotspot for...