Steven Hart

Mayo Clinic
Country flagUnited States

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (19)

Year:

2026

Year:

2026

Year:

2026

Objectives To develop and validate an agent-based Large Language Model (LLM) system for extracting structured data from breast cancer synoptic pathology reports and assess the performance gap between synthetic and real-world validation. Materials and Methods We developed a modular artificial intelligence (AI) agent-based framework employing sequential specialized LLMs for parsing pathology reports and extracting structured data. We normalized College of American Pathologists (CAP) cancer protocols into 8 sections, 86 subsections, and 229 discrete fields. Seven leading LLMs (gemini-2.5-pro, llama3.3-70b, phi4-14b, deepseek-r1 14B/70B, gemma3-27b, gemini-2.0-flash-lite) were validated using dual evaluation: synthetic validation (864 controlled test cases) and real-world ground truth (6651 annotated fields from 90 pathology reports). Results Synthetic validation demonstrated strong performance (accuracy: 93.8%-99.0%). Real-world evaluation revealed field extraction recall ranging from 61.8% to 87.7%, demonstrating a substantial “reality gap” with performance drops of 11-32 percentage points. The gemini-2.5-pro model achieved the highest real-world recall (87.7%). Model size did not predict performance: the 14B-parameter deepseek-r1 (77.6%) outperformed its 70B-parameter counterpart (70.4%). Discussion The substantial performance degradation from synthetic to real-world data underscores the complexity of authentic clinical documentation. Smaller models can achieve competitive or superior recall, reducing computational costs. With even the best models missing 12%-38% of annotated fields, mandatory human verification is essential for clinical deployment. Conclusion While LLM-based extraction systems show promise for pathology data extraction, synthetic validation alone provides false confidence. Rigorous real-world ground truth evaluation with expert annotation is essential before clinical deployment. These systems are best positioned as screening tools with mandatory human oversight rather than autonomous decision-making systems.

Year:

2026

Purpose: Germline pathogenic variants (PV) in ATM increase the risk of pancreatic ductal adenocarcinoma (PDAC), but the underlying tumor biology of PDAC associated with germline PV in ATM has not been adequately explored. Experimental Design: Whole-genome, whole-exome, and RNA sequencing were performed on PDAC tumors from 25 germline ATM PV carriers diagnosed at Mayo Clinic between 2007 and 2017. Somatic and copy-number alterations, mutational signatures, transcriptomic subtypes, and the immune landscape were evaluated. Results: High-quality whole-exome and whole-genome sequencing were obtained from 21 and 15 tumors, respectively. Biallelic inactivation of ATM was observed in 87%, KRAS PV in 90%, CDKN2A homozygous loss in 60%, and TP53 alterations in <10% of these tumors. A predominant clock-like mutational signature was present in all samples. Whole-transcriptome analysis identified that the aberrantly differentiated endocrine exocrine subtype accounted for 18% of PDAC and was consistently associated with >5-year overall survival. In addition, a 28-gene expression–based signature associated with overall survival was identified and further validated in The Cancer Genome Atlas cohort. Immune landscape analysis through CODEX identified enriched CD4 T-helper cell/tumor interactions and reduced B7H3-high cell/tumor interactions in ATM PV carriers compared with noncarriers. Conclusions: The observed absence of TP53 PV and enrichment for CDKN2A alterations in ATM tumors, along with differences in the mutational signatures, transcriptomic subtypes and immune landscape, improve our understanding of the mechanistic pathways involved in PDAC development in germline ATM PV carriers and help identify potential targeted therapeutic strategies.

Year:

2025

Objective Despite exponential growth in artificial intelligence (AI) research for laboratory medicine and pathology, a significant gap exists between model development and clinical AI implementation. This article introduces a structured framework, the Clinical AI Readiness Evaluator (CARE), to bridge this gap and support the responsible adoption of AI in clinical laboratory settings. Methods Building upon the Machine Learning Technology Readiness Levels framework, we developed CARE specifically for the clinical laboratory environment by incorporating health care–specific requirements, regulatory considerations, and workflow integration needs. This framework was iteratively refined through practical application across diverse AI use cases within laboratory medicine and pathology. Results The CARE framework provides a systematic approach to AI development and implementation through 8 component workstreams: clinical use case, data, data pipeline, code, clinical user experience, clinical technology infrastructure, clinical orchestration, and regulatory compliance. Unlike generic AI frameworks, CARE distinguishes itself by emphasizing both health care and laboratory workflow integration, regulatory requirements, ethical considerations, and comprehensive validation for clinical contexts. The framework accommodates both internally developed models and commercial AI solutions, providing clear guidance through technology readiness levels and structured review processes. Conclusions The CARE framework addresses the unique challenges of implementing AI in laboratory medicine and pathology by providing a comprehensive roadmap from initial concepts through clinical deployment and maintenance. This article, the first in a series of 4, establishes the foundational AI lifecycle framework, while subsequent articles will explore data documentation, ethical AI considerations, and governance structures. By adopting this structured approach, laboratories can responsibly harness AI’s potential to enhance diagnostic accuracy and operational efficiencies and, ultimately, improve patient care.

Year:

2025

Year:

2025

Supplementary figures displaying supporting evidence Supplemental Figure 1: Results of IHC Staining for ATM expression. Supplemental Figure 2: Variant Summary from Whole Exome Sequencing Supplemental Figure 3: Oncoprint of Somatic Mutations. Supplemental Figure 4: Circos Plot of Copy Number Changes and Structural Arrangements Supplemental Figure 5: Single Base Substitution Profile. Supplemental Figure 6: Single Base Substitution Profile for each Tumor Evaluated through Whole Exome Sequencing Supplemental Figure 7. Characterization of Single Base Substitution Mutational Signatures in Pancreatic Tumors from Germline ATM Carriers based on Whole Genome Sequencing Supplemental Figure 8: Single Base Substitution Profile for each Tumor Evaluated through Whole Genome Sequencing Supplemental Figure 9: Transcriptomic Subtypes of Pancreatic Ductal Adenocarcinoma in germline ATM carriers based on Non-negative Matrix Factorization. Supplemental Figure 10: Transcriptomic Subtypes of Pancreatic Ductal Adenocarcinoma in germline ATM carriers based on predefined genesets Supplemental Figure 11. Heatmap of Tumor and Immune Cell Interactions for Pancreatic Ductal Adenocarcinoma in ATM Pathogenic Variant Carriers and Non-carriers Supplemental Figure 12: Comparison of Tumor and Immune Cell Interactions between ATM Pathogenic Variant Carriers and Non-Carriers

Year:

2025

Purpose: Germline pathogenic variants (PV) in ATM increase the risk of pancreatic ductal adenocarcinoma (PDAC), but the underlying tumor biology of PDAC associated with germline PV in ATM has not been adequately explored. Experimental Design: Whole-genome, whole-exome, and RNA sequencing were performed on PDAC tumors from 25 germline ATM PV carriers diagnosed at Mayo Clinic between 2007 and 2017. Somatic and copy-number alterations, mutational signatures, transcriptomic subtypes, and the immune landscape were evaluated. Results: High-quality whole-exome and whole-genome sequencing were obtained from 21 and 15 tumors, respectively. Biallelic inactivation of ATM was observed in 87%, KRAS PV in 90%, CDKN2A homozygous loss in 60%, and TP53 alterations in 5-year overall survival. In addition, a 28-gene expression–based signature associated with overall survival was identified and further validated in The Cancer Genome Atlas cohort. Immune landscape analysis through CODEX identified enriched CD4 T-helper cell/tumor interactions and reduced B7H3-high cell/tumor interactions in ATM PV carriers compared with noncarriers. Conclusions: The observed absence of TP53 PV and enrichment for CDKN2A alterations in ATM tumors, along with differences in the mutational signatures, transcriptomic subtypes and immune landscape, improve our understanding of the mechanistic pathways involved in PDAC development in germline ATM PV carriers and help identify potential targeted therapeutic strategies.

Year:

2025

Pathogenic protein-truncating variants of RAD51C , which plays an integral role in promoting DNA damage repair, increase the risk of breast and ovarian cancer. A large number of RAD51C missense variants of uncertain significance (VUS) have been identified, but the effects of the majority of these variants on RAD51C function and cancer predisposition have not been established. Here, analysis of 173 missense variants by a homology-directed repair (HDR) assay in reconstituted RAD51C −/− cells identified 30 nonfunctional (deleterious) variants, including 18 in a hotspot within the ATP-binding region. The deleterious variants conferred sensitivity to cisplatin and olaparib and disrupted formation of RAD51C/XRCC3 and RAD51B/RAD51C/RAD51D/XRCC2 complexes. Computational analysis indicated the deleterious variant effects were consistent with structural effects on ATP-binding to RAD51C. A subset of the variants displayed similar effects on RAD51C activity in reconstituted human RAD51C -depleted cancer cells. Case–control association studies of deleterious variants in women with breast and ovarian cancer and noncancer controls showed associations with moderate breast cancer risk [OR, 3.92; 95% confidence interval (95% CI), 2.18–7.59] and high ovarian cancer risk (OR, 14.8; 95% CI, 7.71–30.36), similar to protein-truncating variants. This functional data supports the clinical classification of inactivating RAD51C missense variants as pathogenic or likely pathogenic, which may improve the clinical management of variant carriers. Significance: Functional analysis of the impact of a large number of missense variants on RAD51C function provides insight into RAD51C activity and information for classification of the cancer relevance of RAD51C variants.

Year:

2025

Year:

2025

Year:

2025

Year:

2024

Year:

2024

Year:

2024

Year:

2023

Year:

2023

Year:

2023

Year:

2023

Collaborators (26)

Jean Yves Masson

Full Professor

Université Laval Faculté de médecine

CANADA
View details

Christopher A. Haiman

University of Southern California

UNITED STATES
View details

Susan Domchek

Professor, Clinician-Educator

University of Pennsylvania

UNITED STATES
View details

Sara Lindström

University of Washington

UNITED STATES
View details

Katherine Nathanson

Professor

University of Pennsylvania

UNITED STATES
View details

Jacques Simard

Full Professor

Laval University

CANADA
View details

Song Yao

Professor

Roswell Park Comprehensive Cancer Center

UNITED STATES
View details

FEI CHEN

Assistant Professor

University of Southern California

UNITED STATES
View details

Irene Ong

Associate Professor

University of Wisconsin-Madison

UNITED STATES
View details

Julie R Palmer

Boston University School of Medicine

UNITED STATES
View details

Daniel D. Billadeau

-

UNITED STATES
View details

gaofeng cui

Assistant professor

Mayo Clinic

UNITED STATES
View details

Peter Kraft

Professor

Harvard T.H. Chan School of Public Health

UNITED STATES
View details

Adrian V. Lee

University of Pittsburgh

UNITED STATES
View details

Siddhartha Yadav

Associate Professor of Oncology

Mayo Clinic

UNITED STATES
View details

Robert McWilliams

Mayo Clinic Minnesota

UNITED STATES
View details

Thomas Durant

Assistant Professor

Yale University School of Medicine

UNITED STATES
View details

Toby Cornish

Professor

Medical College of Wisconsin

UNITED STATES
View details

Yuan-Ping Pang

Professor of Biophysics and Pharmacology

Mayo Clinic

UNITED STATES
View details

Joel Saltz

SUNY Distinguished Professor, Chair

Stony Brook University

UNITED STATES
View details

Mark Zarella

Vice Chair of Digital and Computational Pathology

University of Pennsylvania Health System

UNITED STATES
View details

Jeffrey Weitzel

Professor and Director

University of Kansas Cancer Center

UNITED STATES
View details

Maegan Roberts

Assistant Professor

Ohio State University

UNITED STATES
View details

Alvaro Monteiro

Senior Member - Full Professor

H Lee Moffitt Cancer Center and Research Institute

UNITED STATES
View details

Eric Polley

Associate Professor

University of Chicago

UNITED STATES
View details

Nicholas Boddicker

Research Associate

Mayo Clinic

UNITED STATES
View details