Lung screenings are an important diagnostic tool for early detection of cancer, but low-dose CT scans cannot determine whether the pulmonary nodules identified in the imaging are cancerous or benign.
Specially trained radiologists, pulmonologists and thoracic surgeons analyze the imaging and then decide whether invasive biopsies are warranted to make that determination — a process called risk stratification of indeterminate pulmonary nodules.
An estimated 10% to 15% of the resections of these nodules turn out to be surgeries for benign tissue. Risk prediction software using artificial intelligence, such as the Lung Cancer Prediction Score, which was developed by Optellum, a lung health technology company, and approved by the Food and Drug Administration in 2021, aids clinicians in determining whether surgical resections are necessary. In a recent study, Vanderbilt Health researchers determined that AI-assisted decision-making with this software is cost-effective compared to clinician assessment alone.
The study, published March 5 in PLOS ONE, showed that AI-assisted decision-making resulted in an incremental cost-effective ratio of $4,485 per life year gained.
Eric Grogan, MD, MPH
“Artificial intelligence-based tools offer promising assistance to busy clinicians who evaluate suspicious lung nodules and seem to be cost-effective,” said the study’s corresponding author, Eric Grogan, MD, MPH, Ingram Professor of Cancer Research and professor of Thoracic Surgery at Vanderbilt Health.
To determine cost-effectiveness, the researchers constructed a decision model assuming guideline-based care from a payer perspective with a lifetime horizon. The base case is a 1.1 centimeter indeterminate pulmonary nodule in a 60-year-old patient who benefits from surgery. This nodule’s risk for lung cancer is about 65%. The model classified patients as low, medium or high risk using either clinician reasoning or clinician-plus-AI reasoning.
Stephen Deppen, PhD
“When we think of these AI clinical decision aids, they may not really help the true clinical expert, the thoracic radiologist or pulmonologist who sees 20 of these a day. Where the larger health care system impact occurs is when generalist physicians can rely on these tools to remove the easy, cancer and not cancer cases, so they can focus or get a consult on the most difficult,” said the study’s senior author, Stephen Deppen, PhD, associate professor of Thoracic Surgery and co-director of the Early Detection Research Network Lung Group’s National Clinical Validation Center.
Other Vanderbilt Health authors are Caroline Godfrey, MD, MPH, Ashley Leech, PhD, MS, Kevin McGann, MD, Jinyi Zhu, PhD, MPH, Hannah Marmor, MD, MPH, Sophia Pena, Fabien Maldonado, MD, MSc, Evan Osmundson, MD, PhD, and Stacie Dusetzina, PhD. The researchers received support from National Institutes of Health grants T32CA106183, K01DA050740, R01CA253923, P30CA068485, U01CA152662, R01CA252964 and U01CA152662.
Vanderbilt Health and Bertis, an artificial intelligence-driven proteomics-based precision medicine company, have announced a joint research and co-development collaboration. The endeavor marks a significant milestone in oncology by advancing the convergence of AI, spatial biology and translational cancer research.
By integrating Vanderbilt Health’s Molecular AI Initiative capabilities with Bertis’ proprietary deep proteomics and AI-enabled target discovery technologies, the collaboration will build an advanced, spatially resolved dataset to identify novel therapeutic targets and predictive biomarkers.
Traditional target discovery often relies on bulk tissue analysis, which loses the critical context of how cells are organized within a tumor. Vanderbilt Health’s Molecular AI approach changes this paradigm by employing sophisticated computational spatial analysis to generate high-resolution spatial molecular maps. This AI-driven spatial biology allows researchers to visualize and decode the complex architecture of the tumor microenvironment, specifically identifying how tumor, immune and stromal (connective tissue) cells interact in biologically and therapeutically relevant regions. By mapping the precise locations and spatial relationships of these cells, the Molecular AI platform can isolate the key cell populations responsible for treatment response or resistance.
These advanced spatial insights are then integrated with Bertis’ cutting-edge proteomics capabilities. While Vanderbilt Health maps the critical spatial context, Bertis will conduct deep proteomic and metabolomic profiling, applying its proprietary AI-enabled computational models to prioritize the most viable, druggable targets.
Tae Hyun Hwang, PhD
The initial focus of this joint research will be on HER2-low tumors (cancers that express low levels of the growth-promoting protein HER2), a historically challenging clinical area, with the potential to expand into additional tumor types based on data outcomes and joint scientific discussions. By layering spatial context over proteome-level data, the teams aim to pinpoint cell surface proteins that are uniquely positioned for emerging therapeutic modalities, including antibody-drug conjugates and cell-based therapies.
This sophisticated AI-driven spatial multimodal and deep proteomics pipeline is spearheaded by Tae Hyun Hwang, PhD, professor of Surgery, founding director of Molecular AI Initiative and director of AI Research in the Section of Surgical Sciences at Vanderbilt Health. Hwang also co-leads gastric cancer atlas efforts within the National Cancer Institute-funded Human Tumor Atlas Network (HTAN) and is spearheading international HTAN collaborations with South Korea’s National Cancer Center.
Highlighting the clinical necessity of this integrated approach, Hwang said, “Identifying therapeutic targets and understanding treatment response require a precise view of proteins, spatial context and tumor biology. By combining Vanderbilt Health’s Molecular AI and spatial analysis capabilities with Bertis’ proteomics and AI-enabled target discovery platform, this collaboration is designed to generate high-confidence therapeutic targets and predictive biomarkers that can support future translational research and therapeutic development.”
Bertis is led by co-CEOs Dong-young Noh and Seung-man Han, who emphasized the collaboration accelerates the global reach of their platform.
“Collaborating with Vanderbilt Health, a leading U.S. academic medical center with strong expertise in Molecular AI, spatial biology and cancer research, is highly meaningful and reflects the growing global recognition of Bertis’ technological capabilities,” Han said. “Through this collaboration, we aim to expand the role of AI-driven proteomics in drug discovery and identify therapeutic targets that may open new possibilities in oncology.”
Researchers at Vanderbilt University Medical Center using artificial intelligence have helped develop two technologies for improving cancer care.
One technology called MSI-SEER, described in a study published in npj Digital Medicine, better predicts microsatellite instability-high status from standard pathology slides and provides clinicians with specific data, including any uncertainties with predictions. The other technology, a breakthrough three-dimensional imaging tool described in a study published in Nature Communications, has transformative potential beyond cancer diagnostics.
These new technologies showcase how VUMC researchers are using the power of AI to meet a wide range of medical needs, said Tae Hyun Hwang, PhD, professor of Surgery, founding director of the Molecular AI Initiative, and director of AI Research for the Vanderbilt Section of Surgical Sciences. He noted that the 3D imaging could significantly advance development of therapeutic drugs, provide more detailed assessments of organ transplant rejections, assist with personalized medicine, and aid with tissue analysis for biopharmaceutical development.
Tae Hyun Hwang, PhD
“This technology fundamentally redefines how we visualize and analyze tissue architecture, moving from traditional two-dimensional views to full 3D microenvironment mapping at the subcellular level,” said Hwang, a corresponding author of the study, who provided senior leadership in the development, validation and translational development of the technology.
The 3D study published in Nature Communications introduced an innovative framework that integrates holotomography with deep learning to generate hematoxylin- and eosin-stained images directly from thick tissue samples. This noninvasive, AI-driven approach preserves tissue integrity, overcomes the traditional 4- to 5-micron thickness limit of routine histology, and enables volumetric visualization of biological structures up to 50 microns thick.
By preserving tissue samples and avoiding chemical alteration, this method also ensures compatibility with downstream molecular assays, such as spatial transcriptomics, proteomics and genomic profiling — enhancing the breadth and depth of diagnostic and research capabilities.
“This is not just a digital copy of hematoxylin- and eosin-staining,” Hwang said. “It is a foundational platform for AI-driven volumetric tissue analysis that accelerates discoveries in oncology, immunology, regenerative medicine and therapeutic development.”
The multi-institutional effort also included researchers from KAIST, Tomocube Inc., Yonsei University College of Medicine and Mayo Clinic. Hwang received funding support from the National Cancer Institute (grants R01CA276690, R37CA265967, U01CA294518).
VUMC researchers developed the MSI-SEER predictor technology in collaboration with Mayo Clinic, Yonsei Severance Hospital and Seoul St. Mary’s Hospital in South Korea. This technology identifies patients who will benefit from an immunotherapy that might otherwise be missed with existing prediction models.
Microsatellite instability-high (MSI-H) status is a well-established biomarker used to identify patients likely to respond to immune checkpoint inhibitors, especially patients with gastrointestinal cancers. However, traditional testing methods — including immunohistochemistry and PCR-based assays — offer only a binary result and often miss focal or heterogeneous MSI-H regions within tumors.
MSI-SEER overcomes this limitation by dividing each pathology slide into thousands of image tiles and generating region-by-region predictions of MSI-H probability. This enables visualization of the tumor’s spatial heterogeneity and quantification of the MSI-H fraction across the tumor. In multiple cases, MSI-SEER identified MSI-H regions in tumors previously classified as microsatellite stability, and those patients subsequently responded to immunotherapy.
“This is analogous to what we say in HER2-low gastric cancer, where patients previously not eligible for targeted therapy are now being treated with agents like trastuzumab deruxtecan,” Hwang said. “Likewise, patients with low or heterogeneous MSI-features may now be reconsidered for immunotherapy if spatially resolved analysis like MSI-SEER is used.”
A key innovation of MSI-SEER is its ability to report not only predictions but the confidence level for each result.
“AI should not dictate clinical decisions; it should support them,” Hwang said. “MSI-SEER gives clinicians both the answer and a measure of how reliable the answer is. It’s not about replacing human expertise but about combining the best of AI computation with physician judgment to drive safe, precise decisions.”
Hwang, who conceptualized the study and is the paper’s senior author, received research support from the National Cancer Institute and the Department of Defense. He also received support from the Eric and Wendy Schmidt Fund for AI Research and Innovation and the American Association for Cancer Research Innovation and Discovery Grant.
Other VUMC researchers who authored the study are Sunho Park, PhD, Minji Kim, MS, Jean Clemenceau, PhD, and Inyeop Jang, PhD.
Experts on the research, clinical use, governance and ethical use of artificial intelligence gathered for the recent Vanderbilt-Ingram Cancer Center 26th Annual Scientific Symposium.
In a twist from years past, graduate students and postdoctoral fellows took the helm in selecting the topics and inviting speakers focused on “Artificial Intelligence in Cancer Research and Clinical Care.”
The keynote speakers were Eytan Ruppin, MD, PhD, chief of the Cancer Data Science Laboratory at the National Cancer Institute, and Gelareh Zadeh, MD, PhD, chair of the Department of Neurologic Surgery at Mayo Clinic.
Ruppin detailed how he is developing computational approaches for advancing precision oncology, and Zadeh explained how she is using integrated multi-platform molecular analysis of brain tumors to predict patients’ responses to targeted therapies. Ruppin participated in panel discussions about artificial intelligence.
“I am enriched talking to you guys,” Ruppin said. “I develop AI materials, but I am not using them to treat patients. I am learning a lot.”
Douglas Flora, MD, executive medical director of Oncology Services at the Yung Family Cancer Center at St. Elizabeth in Edgewood, Kentucky, and the editor-in-chief of AI in Precision Oncology, replied, “All of us are cross pollinating. That’s why I love a symposium like this.”
In opening the first panel discussion that focused on ethics, Ellen Wright Clayton, MD, JD, the Craig-Weaver Professor of Pediatrics, professor of Law and professor of Health Policy at Vanderbilt, framed artificial intelligence from an historical perspective, noting that “decision support is not new to medicine.” She gave specific examples of how clinicians can use artificial intelligence for decision support but stressed that they should not rely solely on it for treatment plans.
“It is not OK simply to get the AI output and just do what it says,” Clayton said. “Maybe it is OK, but it is always required to see if that’s the right advice. Always.”
In another twist from years past, the Mission Moment, which is a personal testament from a patient, was presented by a pediatric cancer survivor for the first time. Easton Reeder, 13, who has undergone surgery and chemotherapy for pilocytic astrocytoma — a type of brain tumor — shared about his experiences living with cancer. He told his story vividly with flashes of humor, describing how being tossed in the air like a rodeo clown by a Great Dane led to his diagnosis.
Clinicians initially concluded that he had a concussion because of persistent headaches that followed, but his mother, who is a nurse, insisted on a brain scan. Reeder, a committed athlete, who continued playing sports even while undergoing chemotherapy, was given a jersey signed by Vanderbilt baseball players.
“I learned that tomorrow isn’t a promise, and I have to make the best of every moment I have,” Reeder said. “I also learned that there is no ‘normal button.’ I have been trying to learn that power since forever, until I realized that power is not to be . . . God has proven to me that anything is possible through him.”
In his welcoming remarks, Vanderbilt-Ingram director Ben Ho Park, MD, PhD, emphasized the importance of training new generations of cancer researchers and clinicians.
“This is an opportunity for us to celebrate all the cancer research going on at Vanderbilt-Ingram,” Park said. “For all of you who don’t know, we really run the spectrum of everything research: clinical, population science, laboratory science and everything in between. This is our time of the year when we get to showcase and highlight not only the great science that our external panelists and presenters are going to bring — but you will be duly impressed, as I always am, by what our trainees bring to the table. The future really is bright, and we have to keep sustaining our future by encouraging and mentoring the next generation, which will ultimately lead to more cures.”
From left are Christopher Williams, MD, PhD, associate director of Research Education at VICC, Michael Robinson, MD; Guochong “Damon” Jia, PhD, postdoctoral scholar of the year, Katie Brown, PhD, co-chair of the Vanderbilt-Ingram Scientific Symposium, Candace Grisham, MS, co-graduate student of the year, Xiaopeng Sun, PhD, co-graduate student of the year, Jared Rhodes, co-chair of the Scientific Symposium, Ben Ho Park, MD, PhD, director of VICC. (photo by Donn Jones)
Two tied in voting for the Graduate Student of the Year. Candace Grisham, MS, received the honor for her research into brain tumors, including a study she authored that was published in Clinical Neurology and Neurosurgery. Xiaopeng Sun, PhD, is the other co-awardee for his research into biomarkers to predict immunotherapy outcomes in patients, and his prolific contributions to that field of study, including 12 studies published in scientific journals.
Guochong “Damon” Jia, PhD, MPH, is the Postdoctoral Scholar of the Year. He was selected for his high-impact research that has advanced the understanding of cancer genetics and epidemiology, including the largest genetic study ever conducted on breast cancer in African ancestry populations, which was published in Nature Genetics.
The poster exhibition was one of the largest ever for the annual event. Sarah Reed took home the overall winner award for her entry “Identifying Genotype-Specific Effects of CHIP on Solid Tumors Using Chimeric Mouse Modeling and Clinical Data.”
In the Translational Science Category, Julia Steele won first place; Alexander Kwiatkowski, PhD, won second place; Heather Beasley, PhD, won third place; and honorable mention went to Jacey Marshall.
Robust participation in the Basic Science Category resulted in duplicate prizes due to the number of entries. Rachel Sinard and Lincoln Brown won first place prizes. Emily Green and Logan Vlach received second place awards, andAnna Gilbert and Alyssa Jarabek received third place awards. Honorable mentions went to Sydney Bates, Nicholas Eleuteri, Sarah Glass, PhD, Gabriela Gonzalez Vasquez, Maxwell Hamilton, Yash Pershad, Jared Rhodes and Brenda Rios.
In the Population Science Category, Michael Betti received first place; Duc Huy Le, MD, MBA, received second place; and there was a tie for third place with Melissa Goldin and Jiajun Shi, PhD, both receiving awards. Honorable mention went to Grace Xu.
