Martin J. Menten

—

Unclaimed · AI-structured profileORCID OpenAlex

Medical AI safety researcher studying memorisation, privacy and longitudinal clinical image analysis.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-09-20) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Insufficient publication-volume data: this axis would measure total output and its distribution across venues and years. The snapshot records only two 2026 preprints, so output cannot be scored and is left blank rather than estimated.

Impact—

No citation or uptake data is available: this axis would measure citations, downstream reuse, and influence on clinical AI practice. Impact is left unscored because the bibliometric record here is too thin to support any estimate.

Focus100/100

The focus score is 100 because the bibliometric snapshot classifies the entire recorded output within a single field: machine learning for healthcare with a medical AI safety emphasis (fieldDist contains exactly one entry). Both catalogued papers — memorisation bias in medical AI and longitudinal angiographic graph matching — sit inside clinical machine learning, indicating a tightly concentrated research agenda.

Collaboration48/100

The collaboration score is 48, based on 10 distinct co-authors in the snapshot. The recurring circle includes Laurin Lux, Daniel Rueckert, Moritz A. Knolle, Mélanie Roschewitz, Emma A. M. Stanley, Georgios Kaissis, Ben Glocker, Linus Kreitner, and Carmen Baumann, suggesting a stable clinical-ML network; the moderate score reflects a focused rather than broadly dispersed co-authorship pattern.

Activity—

Insufficient recent-activity data: this axis would measure publication cadence and recency over time. Only two 2026 preprints are present in the snapshot, so no reliable activity trend can be computed yet.

Reproducibility100/100

Reproducibility scores 100 (reproMean 100), but from only one assessed paper (reproPapers = 1): 'Memorisation bias in medical AI', whose Evolve repro-readiness rating is high. The perfect score therefore reflects a single strong signal rather than a track record, and should be revisited as more of his work is assessed.

Research summary

Martin J. Menten (ORCID 0000-0001-8261-7810) works in machine learning for healthcare, with the Evolve bibliometric classification placing his recorded output in a single field: medical AI safety. The catalog currently holds two 2026 arXiv papers. 'Memorisation bias in medical AI', on which he is a middle author, tests whether training on patients' past records biases predictions on their future data; Evolve describes the evidence as rigorous and large-scale, spanning ECG deep learning and chest radiograph classification, while judging the clinical harm modest, and rates the paper's reproducibility readiness as high. Its keywords tie memorisation to differential privacy and membership inference. The second paper, 'Temporally Consistent Graph Extraction and Matching for Longitudinal Angiographic Images', where he is the last-listed author, addresses temporal consistency in longitudinal angiographic analysis. The available bibliometric signals are narrow but consistent: a focus score of 100 from a single-field distribution, a reproducibility score of 100 (mean across one assessed paper), a collaboration score of 48 across 10 distinct co-authors, and a recurring circle that includes Daniel Rueckert, Ben Glocker, Georgios Kaissis, Mélanie Roschewitz, and Laurin Lux. Output, impact, and activity axes lack sufficient underlying data and are deliberately left unscored. With institution and country unrecorded, this should be read as an early-stage, name-resolved profile built from a small bibliographic snapshot.

Collaboration pattern

The snapshot records 10 distinct co-authors and a collaboration score of 48. The recurring circle — Laurin Lux, Daniel Rueckert, Moritz A. Knolle, Mélanie Roschewitz, Emma A. M. Stanley, Georgios Kaissis, Ben Glocker, Linus Kreitner, and Carmen Baumann — suggests a stable, specialist network in medical machine learning, with mid-to-last authorship roles rather than a broad, dispersed collaboration footprint.

Research keywords

Medical AI safetyMemorisationDifferential privacyMembership inferenceECG deep learningChest radiograph classificationLongitudinal image analysisAngiographic imagingGraph extractionGraph matchingTemporal consistencyBias in medical AI

Highlights

Large-scale evidence that memorising past records biases medical AI
In 'Memorisation bias in medical AI' (2026, arXiv; middle author), the study supports the claim that training on patients' past records biases predictions on their future data, with rigorous large-scale experiments across ECG deep learning and chest radiograph classification. Evolve judges the clinical harm modest and rates reproducibility readiness high.
Temporal consistency for longitudinal angiographic images
In 'Temporally Consistent Graph Extraction and Matching for Longitudinal Angiographic Images' (2026, arXiv), he appears as the last-listed author, contributing to graph-based extraction and matching that tracks structures across timepoints in angiographic series; it extends his catalog footprint beyond AI safety into clinical image analysis.
Privacy tooling frames memorisation risk
The memorisation paper connects the phenomenon to differential privacy and membership inference, using ECG deep learning and chest radiograph classification as testbeds — a measurement-oriented take on medical AI safety whose high repro-readiness makes it a candidate baseline for follow-up privacy-bias studies.

Timeline

2026Middle author on 'Memorisation bias in medical AI' (arXiv), the catalog's reviewed medical-AI-safety paper, rated high for reproducibility readiness.
2026Last-listed author on 'Temporally Consistent Graph Extraction and Matching for Longitudinal Angiographic Images' (arXiv).
2026Both 2026 preprints enter the Evolve catalog, anchoring a single-field (medical AI safety) classification with focus and reproducibility scores of 100.

Papers in the Evolve catalog

2026-09-17
Temporally Consistent Graph Extraction and Matching for Longitudinal Angiographic Images
2026-09-15
Memorisation bias in medical AI

Frequent co-authors

Laurin Lux×2 Daniel Rueckert×2 Moritz A. Knolle×1 Mélanie Roschewitz×1 Emma A. M. Stanley×1 Georgios Kaissis×1 Ben Glocker×1 Linus Kreitner×1 Carmen Baumann×1 Daniel Rueckert×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Profile is name-resolved from public bibliographic data via ORCID 0000-0001-8261-7810; institution and country are not recorded in the snapshot.
Bibliometric scores come from a small public bibliometric snapshot; output, impact, and activity axes are unscored for insufficient underlying data.
The reproducibility score of 100 rests on a single assessed paper (reproPapers = 1) and may not generalise.
Coverage is limited to two 2026 arXiv papers in the Evolve catalog; earlier or non-preprint work is not reflected.