Three US lung-nodule clinics recorded more new attendees and more patients undergoing invasive diagnostic procedures after introducing AI support. Differences in lung cancer diagnoses, early-stage cancer diagnoses and time to diagnosis were not statistically significant.
These findings come from a study published in PLOS One on 14 October 2026. The researchers compared clinic activity before and after implementation, without a concurrent control group. This design cannot establish that AI caused the changes.
What changed in the clinics
Optellum Virtual Nodule Clinic combines searches for nodule mentions in radiology reports, CT-based risk estimates and a follow-up dashboard. Staff contacted patients or referring clinicians to arrange referrals; clinical decisions remained with doctors. The study assesses this combined pathway and cannot isolate the contribution of an individual module.
Across the comparison periods, new patients attending the clinics increased from 138 to 232. Patients undergoing at least one invasive diagnostic procedure increased from 61 to 100. The second measure counts people, rather than every procedure performed.
The researchers compared the same calendar months one year apart: 12.5 months pooled across the three clinics in each phase, rather than 12.5 months at each clinic. New-attendance data covered 2022–2024, with follow-up at one site extending to August 2025. The October date refers to the study’s publication.
What the findings leave open
Higher activity alone does not establish patient benefit or whether procedures were appropriate. The analysis of time from baseline CT to confirmed lung cancer diagnosis included only 37 patients before implementation and 45 afterwards. A non-significant difference does not establish equivalent outcomes. Cost savings and staff-time savings were not measured.
Clinic capacity and imaging volume were not consistently tracked, and one clinic lost a key cancer referral source. These changes, along with differences in patient mix, could have affected the comparison.
The system’s developer, Optellum, participated in the study design, analysis, interpretation, decision to publish and manuscript preparation. Three authors work for the company and hold stock options; one also owns shares. The study was funded through commercial agreements with Johnson & Johnson Interventional Oncology and Intuitive Surgical.
A takeaway for clinic managers
Our editorial takeaway concerns how to assess referral and follow-up processes. A report on this kind of implementation should show new attendances, patients undergoing procedures and health-relevant outcomes separately. Alongside them, record changes in referral sources, staffing, imaging volume and available appointment capacity. This helps managers see what was measured and which other changes could have affected the figures. It is an editorial suggestion for operational review, not treatment advice or a product recommendation.
Read also: a separate review of evidence from cardiovascular AI trials.
Prepared by the NADALI editorial team using AI
Source checked on 15 October 2026. Study published on 14 October 2026.
