Publications
& writing
My work in oncology, AI evaluation and governance, alongside conference contributions, correspondence and writing on research culture.
Publication status and authorship roles are indicated throughout. Updated September 2026.
ORCID research recordJournal articles
Published articles, accepted manuscripts and work under review. Status is shown for each entry.
Externalizing Epistemic Governance from Stateless Clinical LLMs: Why It Is Necessary and Why It Must Be Constrained
Philosophy & Technology
Single author
A constrained architecture for oncology decision support that separates evidence authority from permission and reimbursement, keeping uncertainty and clinical overrides inspectable.
Publication date pending.
Read research summaryEarlier public version: Toward an Audit-Ready, Constraint-Based Architecture for Oncology Clinical Decision Support with Large Language Models
Read earlier preprintMore papers, fewer ideas: How evaluation asymmetries drive scientific convergence
Journal of Scholarly Publishing
Single author
Publication date pending. Linked version: SocArXiv preprint.
Read the preprintClinician expertise and prompt engineering enhance cancer information extraction in electronic health records by small language models
Communications Medicine
Co-author
44 authors; Corso, Peppoloni and Mazzeo contributed equally.
View publicationClinical usability of an explainable AI decision support tool and evaluation of multimodal models in NSCLC
Nature Medicine
Consortium author
View publicationArtificial intelligence-assisted Mendelian randomization: a methodological framework for causal inference with applications in oncology
Infectious Agents and Cancer
Co-author
View publicationComposite clinical scores in real-world extensive-stage small-cell lung cancer patients treated with chemo-immunotherapy
Tumori Journal
Co-author
View publicationComprehensive genomic profiling by liquid biopsy in refractory metastatic colorectal cancer patients candidate for anti-EGFR rechallenge therapy (CAVE-2 GOIM trial)
ESMO Open
Co-author
Contribution: eCRF clinical data management.
View publicationRadiation therapy, tissue radiosensitization, and potential synergism in the era of novel antibody–drug conjugates
Critical Reviews in Oncology/Hematology
First author
View publicationA comprehensive update of hormone-related pharmacokinetic variations associated with breast cancer drugs
Expert Opinion on Drug Metabolism & Toxicology
Co-first author
View publicationTransformation of non-small cell lung cancer into small cell lung cancer: a systematic review with exploratory analysis
Clinical Lung Cancer
Co-author
Conferences & abstracts
Congress abstracts and presentations, with individual authorship roles.
The Preservation Gap: The NAT-Clinical Framework for Governing Clinical Large Language Models Across Updates
ESMO AI & Digital Oncology Congress, Berlin
First and presenting author
Abstract #460. ESMO Merit Award. Based on Continuity Is All You Need: Why Clinical AI Governance Requires Structural Persistence.
Role of the pan-immune-inflammatory value in stage III NSCLC treated with chemoradiotherapy with or without durvalumab (305P)
ESMO Open
Co-author
View publicationLungTrial-LLMX: An explainable on-premises LLM system for clinical trial matching in lung cancer
ESMO Real World Data and Digital Oncology
Co-author
View publicationComprehensive benchmarking of large language models on oncological records reveals variations to prompt engineering and medical expertise
Annals of Oncology · ESMO Congress
Co-author
View publicationAssessing first-line immunotherapy duration in advanced NSCLC
Journal of Thoracic Oncology · ELCC
Co-author
View publication
Preprints & working papers
Conceptual and methodological work. Preprints are identified separately from published journal articles.
The bias is in the eye of the beholder: an epistemological reframing of LLM fairness research
Zenodo
Single author
Distinguishes individual outputs, conditional behaviour and system-level claims, proposing robustness checks for evaluating claims about LLM bias.
Under review at Discover Artificial Intelligence.
Read preprintThe Unevaluable Device: Why Standard Evaluation Methods Are Structurally Inadequate for Clinical Large Language Models
Zenodo
Single author
Examines how non-reproducibility, opacity, unrestricted behaviour and global plasticity challenge the assumptions of clinical AI evaluation.
Read preprintNobody’s Fault: Where Liability Collapses in Clinical LLM Deployment
SSRN
Single author
Examines how clinical LLM architecture complicates causal attribution and the allocation of responsibility across deployment chains.
SSRN abstract 6819179.
Read preprintContinuity Is All You Need: Why Clinical AI Governance Requires Structural Persistence
Zenodo
Single author
Proposes a preservation requirement and a risk–verifiability model for governing validated clinical AI properties across interactions and updates.
Read research summaryRead preprintExternalising Epistemic Governance for Stateless Large Language Models: The CUL/TCL Architecture
Zenodo
Single author
A two-layer architecture that determines verification requirements and assigns evidence-based truth states before language generation.
Read preprintThe Narrative Continuity Test: A Conceptual Framework for Evaluating Identity Persistence in AI Systems
arXiv
Single author
A conceptual framework for assessing whether an AI system preserves memory, goals, self-correction, semantic stability and role continuity over time.
In revision at AI & Ethics.
Read research summaryRead preprintCommentary on the ESMO Guidance for LLM Use in Oncology: Narrative and Process Divergences
Zenodo
Single author
Originally submitted to Annals of Oncology.
Read preprintArtificial intelligence in oncologic pathology: lessons from radiomics and the challenges of responsibility and accountability
Zenodo
Single author
Originally submitted to Annals of Oncology.
Read preprint
Correspondence
Published comments and eLetters. Links labelled “Original article” lead to the work being discussed.
When Commercial Influence Enters the Answer
JAMA
Single author
In response to: Singh N, Mello MM. Advertising in AI-Powered Clinical Decision Support Tools. The link opens the original article; its DOI is not a DOI for this response.
Original article“What comes next” is the wrong question: we should ask first whether trials, oversight, and monitoring work for clinical LLMs
Science
Single author
In response to: Hopkins AM, Cornelisse E. AI can reason like a physician—what comes next? The link opens the original article; its DOI is not a DOI for this response.
Original articleWhen the Reference Cannot Hesitate: A Note on Metacognitive Feedback in AI-Augmented Training
JAMA
Single author
In response to: Keren R, Desai BR, West DC. Promoting Clinical Expertise in the Age of AI: No Struggle, No Mastery. The link opens the original article; its DOI is not a DOI for this response.
Original articleBeyond regulatory evasion: clinical LLMs are unevaluable by design
JAMA
Single author
In response to: Aaron DG, Robertson C. The First AI Drug Prescriber. The link opens the original article; its DOI is not a DOI for this response.
Original articleFrom cytotoxic intensification to “living” viral sentinels in metastatic PDAC
BMJ Open
Single author
In response to: Portales F, Gourgou S, Fiess C, et al. GABRINOX-2 protocol: a French, prospective, multicentre, randomised phase II trial evaluating gemcitabine/nab-paclitaxel followed by FOLFIRINOX versus FOLFIRINOX alone as first-line treatment for metastatic pancreatic cancer. The link opens the original article; its DOI is not a DOI for this response.
Original article
Journal of Secondary Endpoints
Scientific satire, editorials and writing on research culture. These works belong to the journal’s satirical and critical publishing programme.
Accept in Current Form
Journal of Secondary Endpoints
Editorial signed as “Editors”
2(1):25–26. Published in an independent journal of scientific satire and research culture.
View publicationCumulative Sanctity: A Multinational Multidenominational Protocol for Assessing Lifetime Liturgical Incense Exposure and Respiratory Carcinogenesis Across Religious Career Trajectories
Journal of Secondary Endpoints
Single author
2(1):27–42. Published in an independent journal of scientific satire and research culture.
View publicationRequiem in LaL Major: A Systematic Review of the Death of Large Language Models
Journal of Secondary Endpoints
Single author
2(1):43–50. Published in an independent journal of scientific satire and research culture.
View publicationAnti-Germination Stereotactic Radiosurgery in Solanum tuberosum: A Phase 0 Feasibility Study
Journal of Secondary Endpoints
Single author
2(1):51–58. Published in an independent journal of scientific satire and research culture.
View publicationDivide et Impera: When Will Peer Review Become Modular and Accountable?
Journal of Secondary Endpoints
Single author
2(1):59–67. Published in an independent journal of scientific satire and research culture.
View publicationEditorial: How We Addressed the Main Unmet Need in the Medical Literature
Journal of Secondary Endpoints
Editorial signed as “Editors”
1(1):1–2. Published in an independent journal of scientific satire and research culture.
View publicationTherapeutic turricephaly for glioblastoma: randomised sham-controlled trial (XXL-GBM)
Journal of Secondary Endpoints
Single author
1(1):3–8. Published in an independent journal of scientific satire and research culture.
View publicationA Phase I, Randomized, Double-Blind, Vehicle-Controlled Trial of Virulence-Attenuated Pseudomonas aeruginosa (P.a-Mel) for the Ecological Displacement of Axillary Microbiota: Data on Results, Impact, Safety, and Exploratory Outcomes (The DEODORISE Trial)
Journal of Secondary Endpoints
Single author
1(1):9–14. Published in an independent journal of scientific satire and research culture.
View publicationExsudatio urinosa: the man who never passed urine
Journal of Secondary Endpoints
Single author
1(1):15–20. Published in an independent journal of scientific satire and research culture.
View publicationThe Outcome-Equity Mandate: A Satirical Policy Paper on Publication Bias
Journal of Secondary Endpoints
Single author
1(1):21–24. Published in an independent journal of scientific satire and research culture.
View publication
No matching works. Try a different search or choose All work.