Griffin Health, a 160-bed community health system in Derby, Connecticut, improved closure rates on incidental imaging findings by 50% and follow-up completion by 17% by pairing AI analysis of radiology reports with a dedicated diagnostic navigator. The gap it closed is a common one: incidental pulmonary nodules are found during imaging ordered for something else entirely, which means the patients who have them fall outside the eligibility criteria, referral pathways, and registries that make low-dose CT (LDCT) lung cancer screening programs work. Industry-wide, roughly half of recommended follow-up imaging is never performed, and nearly half of patients with incidental pulmonary nodules have no follow-up scheduled at all (Kattih et al., 2025).
This case study covers how Griffin surfaced incidental lung nodule patients across more than 100,000 annual imaging studies, how their demographics differed from the screening population, and what the program required operationally.
How are incidentally found lung nodules navigated?
Lung cancer screening works. The infrastructure around it – eligibility criteria, referral pathways, registries, reimbursement, dedicated navigators – represents more than a decade of deliberate program building. It is one of the most structured care pathways in American medicine.
An incidental pulmonary nodule, however, does not arrive through a program. It surfaces in a CT ordered for chest pain, a trauma scan in the emergency department, an inpatient workup for something unrelated. The finding gets documented. The follow-up recommendation is often written directly into the report. And then the patient walks out of a building with no cohort to enroll them in, no registry to track them through, and no one whose job it is to notice they never came back.
Griffin Health, a community health system anchored by a 160-bed hospital in Derby, Connecticut, focused on solving this gap by combining AI enabled technology with clinician expertise and a committed, capable leadership team.
How are lung program populations different from incidental findings populations?
Griffin runs a LDCT lung cancer screening program. Patients who meet age and smoking-history criteria get referred in and monitored on schedule. The incidental findings population is a structurally different problem. There is no referral into a screening census. Patients enter through the emergency department, inpatient units, and outpatient imaging, and the only durable evidence that something needs attention is a sentence buried in a radiology report.
Christine Sylvia Cooper, MHA, vice president of diagnostic and cancer services at Griffin Health, is direct about where the system stood: there was “a real void” in how the organization could evaluate patients with incidental pulmonary nodules, and no technology to fill it (TechTarget, 2026).
What does the evidence say about the gap in incidental lung nodule follow-up?
In a recent cohort study of adherence to nodule follow-up guidelines, nearly half of patients identified with incidental pulmonary nodules had no follow-up scheduled. Among those who went without follow-up but had prior imaging available for comparison, 40% of the nodules had already grown (Kattih et al., 2025).
Broader reviews put follow-up completion for newly detected incidental nodules between 29% and 39% — against a backdrop of hundreds of thousands of nodules detected on CT annually, a figure estimated at roughly 150,000 Americans per year as far back as 2000, before two decades of imaging volume growth (Schmid-Bindert et al., 2022).
Griffin’s experience matched that baseline. Todd Liu, executive vice president and COO, noted that industry-wide roughly half of recommended additional imaging never gets performed, driven by inconsistent handoffs, absent centralized tracking, and the seams between radiology and primary or specialty care (Healthcare IT News, 2025).
There is an equity dimension that makes this more than an operational inefficiency. The American Cancer Society National Lung Cancer Roundtable has documented that populations at highest risk for lung cancer are paradoxically underscreened, diagnosed at more advanced stages, and less likely to receive molecular testing or surgical resection — and that patients with lung nodules who are medically underserved experience fragmented evaluation that compounds those disparities (Barta et al., 2024).
Griffin observed the same pattern from the other direction. The demographic profile of its incidental nodule population differs from its LDCT screening population. Because incidental findings arrive through the emergency department and unscheduled care rather than through physician referral, the program reaches a more diverse group, including patients without reliable access to primary care.
That observation has independent support. In a community cohort of more than 43,000 patients across the Mississippi Delta, only 13% of patients enrolled in an incidental pulmonary nodule program met 2021 USPSTF screening eligibility criteria (Smeltzer et al., 2025). An earlier analysis from the same program found Black patients made up 28.7% of the screening-ineligible nodule cohort compared with 18.6% of the LDCT cohort (Osarogiagbon et al., 2023).
What workflow best addresses incidental lung nodule follow-up?
Griffin implemented Inflo Health’s platform and paired it with navigator-led follow-up. The platform mines radiology reports for language indicating a significant finding and generates a daily worklist of patients with documented incidental pulmonary nodules. Patients are organized into predefined pathways based on the type of follow-up needed and its timing, with provider outreach automated to confirm the recommended order is placed in the EHR.
That worklist then goes to a human. Griffin’s diagnostic navigator, a board-certified radiologic technologist who also oversees the LDCT screening program. She works from a central dashboard tracking cases from initial finding through recommended follow-up, scheduling, and completed study. If the patient received care elsewhere, she can check through Connecticut’s health information exchange. If care is underway, nothing more is needed. If it is not and escalation is required, she acts: contacting the primary care office to coordinate next steps, or reaching the patient directly when there is no primary care physician to call.
The Inflo Health platform plugs into Griffin’s existing processes rather than replacing them thereby increasing capacity to track and act on findings instead of imposing a new workflow on clinicians who did not ask for one.
What results did Griffin Health realize with Inflo Health?
For incidental findings specifically, Griffin realized a 50% improvement in closure rates. Follow-up completion for flagged patients rose 17%. And 18 patients were identified and enrolled in the lung cancer screening program who would not otherwise have been engaged at that moment.
Eighteen is a small number that does something important: it closes the loop between the two populations. A patient who entered through the emergency department with an incidental nodule and no primary care relationship ends up enrolled in a structured screening program. The unstructured pathway feeds the structured one.
Liu framed the stakes plainly: these are real lives potentially extended because a system surfaced the right data at the right time and the teams were equipped to act on it.
What lessons did Griffin Health learn?
Griffin launched without bringing primary care physicians in first and had to explain the program retroactively. Cooper’s advice reflects the cost of that sequencing: identify physician champions, communicate before launch, and set clear expectations for how incidental findings will be managed.
“Doing this in a vacuum doesn’t work,” she said. Once the workflows, communication, and provider support are in place, the recommendation is to expand — to additional categories of incidental findings, not just nodules.
What is Griffin Health’s biggest takeaway?
Detection was never Griffin’s problem. What the organization lacked was the ability to treat scattered documentation as a managed population — and the staffing model to act on it once it could.
That is the shape of the opportunity in every complex workflow sitting outside a formal program. Structured programs get built because they are fundable and measurable. The care between programs — incidental, unscheduled, arriving through whatever door the patient happened to use — gets managed manually or not at all. It is precisely where AI earns its keep, and precisely where a worklist without a navigator behind it accomplishes nothing.
Cooper’s measure of success is the right one: not whether patients were identified, but whether they received the care they needed.
Want to see how this works for your system? Schedule a Conversation
References
Barta, J. A., Farjah, F., Thomson, C. C., Dyer, D. S., Wiener, R. S., Slatore, C. G., Smith-Bindman, R., Rosenthal, L. S., Silvestri, G. A., Smith, R. A., & Gould, M. K. (2024). The American Cancer Society National Lung Cancer Roundtable strategic plan: Optimizing strategies for lung nodule evaluation and management. Cancer, 130(24), 4177–4187. https://doi.org/10.1002/cncr.35181
Kattih, Z., Moore, J. A., Wilson, B., Gajjala, S., Schwartz, J., Kushner, J., Zajac, S., Mahajan, A., Leung, T., & Makkar, P. (2025). Rate of incidental lung nodule follow-up: A cohort study evaluating adherence to guideline recommendations. American Journal of Medicine Open, 13, 100091. https://doi.org/10.1016/j.ajmo.2025.100091
Osarogiagbon, R. U., Liao, W., Faris, N. R., Fehnel, C., Goss, J., Shepherd, C. J., Qureshi, T., Matthews, A. T., Smeltzer, M. P., & Pinsky, P. F. (2023). Evaluation of lung cancer risk among persons undergoing screening or guideline-concordant monitoring of lung nodules in the Mississippi Delta. JAMA Network Open, 6(2), e230787. https://doi.org/10.1001/jamanetworkopen.2023.0787
Schmid-Bindert, G., Vogel-Claussen, J., Gütz, S., Fink, J., Hoffmann, H., Eichhorn, M. E., & Herth, F. J. F. (2022). Incidental pulmonary nodules — What do we know in 2022. Respiration, 101(11), 1024–1034. https://doi.org/10.1159/000526818
Smeltzer, M. P., Liao, W., Goss, J., Qureshi, T., Johnson, S., Harris, A., Dortch, K., Fehnel, C., Ely, S., Ray, M., & Osarogiagbon, R. U. (2025). Reducing smoking requirements for lung screening to address health disparities in a community cohort. JAMA Network Open, 8(6), e2517149. https://doi.org/10.1001/jamanetworkopen.2025.17149
Siwicki, B. (2025, May 22). At Griffin Health, AI helps point out patients that clinicians should screen for cancer. Healthcare IT News. https://www.healthcareitnews.com/news/griffin-health-ai-helps-point-out-patients-clinicians-should-screen-cancer
Stricker, E. (2026, July 28). Griffin Health combines AI & navigation in lung nodule follow-up program. TechTarget Healthtech Analytics. https://www.techtarget.com/healthtechanalytics/feature/Griffin-Health-combines-AI-navigation-in-lung-nodule-follow-up-program
