Enterprise

AI Identifies Lung Cancer Screening Candidates Missed by EHRs

Artificial intelligence can flag eligible patients whose smoking histories are poorly documented, but nurses must remain the critical human connection.

Omega Editorial · November 20, 2026 · 3 min read

Key takeaways

  • Pack-year smoking histories in EHRs were inaccurate in 96.2% of lung screening referrals in one study, hiding eligible patients from screening programs
  • AI can identify screening-eligible patients with poorly documented smoking histories, but nurses must remain the human connection for empathic communication
  • Early-stage lung cancer has a 65% five-year survival rate versus 10% for late-stage disease, making early detection critical
  • Training data bias poses equity risks—developers must examine whether their AI models represent the populations being served
  • Nurses should use new drug approvals as opportunities to educate patients and families about lung cancer screening eligibility

Artificial intelligence systems may help healthcare organizations identify patients eligible for lung cancer screening who slip through the cracks due to incomplete electronic health record documentation, according to Lisa Carter-Bawa, PhD, MPH, APRN, ANP-C, FAAN, FSBM, director of the Cancer Prevention Precision Control Institute at the Hackensack Meridian Center for Discovery and Innovation.

The challenge is significant: in one study of patients referred for lung cancer screening, pack-year smoking history recorded in the medical record disagreed with the history obtained during shared decision-making conversations in 96.2% of cases—230 out of 239 patients. When smoking status and pack-years are inaccurately documented, eligible patients remain invisible to screening programs.

Why it matters

Lung cancer survival depends heavily on early detection. The five-year survival rate for early-stage lung cancer is 65%, compared to just 10% for late-stage disease, according to the American Lung Association. Yet unlike mammography for breast cancer, lung cancer screening lacks broad public awareness, and healthcare systems struggle to identify eligible candidates when foundational data is missing or wrong.

AI as a discovery tool

Carter-Bawa, who also serves as deputy associate director for community outreach and engagement at Georgetown Lombardi Comprehensive Cancer Center, explained that AI could be leveraged to identify screening-eligible individuals within EHR systems and support initial outreach efforts. "If you don't know who the person is, you don't even know how to reach out," she told Oncology Nursing News in an October 2026 interview, first reported by the publication.

The technology offers a practical solution to a documentation problem that has long hampered screening programs, potentially surfacing patients who would otherwise go unscreened until symptoms appear.

The irreplaceable human element

Despite AI's potential to improve patient identification, Carter-Bawa emphasized that nurses must remain "the human in the loop." The human connection and empathic communication that nurses provide cannot be automated away. "I am not a big proponent of AI replacing everything because I think the human connection is so incredibly important," she said.

This perspective aligns with recent concerns in oncology nursing about algorithmic bias and overreliance on technology as barriers to AI adoption.

Equity risks demand scrutiny

Carter-Bawa raised critical concerns about training data bias. Large language models learn from existing evidence "whether it's good or bad," and that evidence reflects historical disparities in who gets represented in research and clinical data. "If AI is being trained on a majority, there are people who are not represented in that training data. And so it's going to potentially widen the disparities in some areas," she warned.

This isn't theoretical: a widely used algorithm that used health costs as a proxy for health needs systematically underestimated the needs of Black patients, rating them at the same risk level as healthier White patients. Carter-Bawa urged developers to examine their training data and address inequities before deployment.

Linking treatment news to screening

Carter-Bawa also encouraged nurses to use each new drug approval as an opportunity to educate the public about screening. While the person a nurse speaks with may not be eligible themselves, their spouse or family member might be. Building the same general awareness for lung cancer screening that exists for mammography remains an important goal.

These details were first reported by Oncology Nursing News.

  • #lung cancer screening
  • #artificial intelligence in healthcare
  • #oncology nursing
  • #ehr documentation
  • #algorithmic bias
  • #health equity

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Enterprise