Combining centralized patient navigation with artificial intelligence more than doubled lung cancer screening uptake at a large U.S. health system — and increased the detection of early-stage tumors. The data were presented at the annual meeting of the American Society of Clinical Oncology (ASCO) in Chicago.
What the study showed
OSF HealthCare, a system with 17 hospitals across rural and urban areas, raised its lung cancer screening rate among eligible patients from 18% to 42% over five years. The jump was achieved with a centralized patient navigation model combined with AI-supported screening technologies. The interventions also increased the detection of earlier-stage cancers.
“A multifaceted approach combining centralized navigation and innovative screening technologies can further improve screening uptake,” said first author and presenter Jun Zhang, MD, PhD, of the OSF HealthCare Cancer Institute.
How centralized navigation and AI work together
The bottleneck in lung screening rarely lies in the exam itself, but in everything around it: identifying who is eligible, contacting the patient, scheduling the low-dose CT, ensuring follow-up and tracking findings. Centralized navigation concentrates these tasks in a dedicated team rather than scattering them across overloaded clinics.
This is where AI reinforces the workflow — supporting the identification of eligible candidates and the analysis of images. Algorithmic tools help flag high-risk patients in records and standardize exam reading, reducing the chance that a suspicious nodule goes unnoticed. The result is fewer patients lost at each step of the screening funnel.
Why lung screening is still underused
Low-dose CT screening is recommended for high-risk current and former smokers — precisely the group in which early diagnosis most changes outcomes. Yet uptake is historically low in almost every health system; an initial rate of 18%, like OSF’s, is far from an exception. Stigma, lack of awareness, logistical barriers and fragmented care keep patients away from the exam.
Lung cancer remains among the deadliest because it is usually diagnosed late. Shifting diagnosis to earlier stages is, in practice, the factor that most affects survival — and that is exactly what a well-coordinated screening program delivers.
The evidence base behind screening
The enthusiasm for screening is not unfounded. The U.S. National Lung Screening Trial (NLST) showed that low-dose CT reduced lung cancer mortality by roughly 20% compared with chest radiography in high-risk smokers. The European NELSON trial later confirmed and extended those findings, cementing low-dose CT as the screening standard for the eligible population.
The problem is that this robust evidence never translated into proportional uptake. It is precisely this gap between what science recommends and what reaches the patient that initiatives like OSF’s aim to close — not by inventing a new exam, but by ensuring the right exam reaches those who need it. Crucially, the mortality benefit only materializes when eligible patients actually complete the scan and any needed follow-up — which is exactly where navigation pays off.
Implications for practice and the broader picture
For radiologists, the message is clear: image quality is necessary but not sufficient. Clinical value comes from the program around it. Standardized nodule reading, classifications such as Lung-RADS and structured follow-up of findings are as decisive as the scanner itself. It is worth recalling what we have discussed about incidental findings on chest CT and cancer risk, where opportunistic detection depends on a reliable follow-up loop.
In settings where organized lung screening is still nascent, the OSF case points to a viable path: investing in navigation and decision-support technology can broaden reach without immediately requiring more hardware. AI applied to reading, as we saw when covering the performance of algorithms on chest X-rays, is one piece of the puzzle — provided it is validated for the local population.
Reaching rural and underserved patients
OSF’s footprint across both rural and urban areas matters. Rural patients face longer travel times, fewer specialists and thinner support networks — all of which depress screening rates. A centralized navigation team can absorb much of that friction, coordinating outreach, scheduling and reminders from a single hub rather than relying on each local clinic to build its own program. When AI is layered on top to triage eligibility and reading, even small or remote sites gain access to a level of consistency that was previously reserved for large academic centers. That combination — central coordination plus distributed access — is arguably the most transferable lesson of the study.
Outlook
The OSF study reinforces a trend that goes beyond radiology: treating screening as a system, not as an isolated exam. As AI matures to identify eligible patients, prioritize queues and standardize reports, and as navigation ensures patients do not fall through the cracks, uptake rates once seen as a ceiling may become the new floor. The next challenge will be replicating these gains in lower-resource systems — and ensuring that increased detection truly translates into more lives saved.
Source: AuntMinnie
