Health-AI methodologist whose 2026 first-author study rethinks how methodological progress in clinical ML is evaluated
1
papers here
1
reviewed by Evolve
1
first-author
0
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—
Not yet scored: the catalog currently indexes one reviewed paper and no publication-count or productivity data, so this axis — measuring total research output over time — cannot be computed until more venues are linked.
Impact—
Not yet scored: no citation counts, h-index, or venue-level influence data are available in the current snapshot; this axis would measure the uptake and citations of Pollet's work across the health-AI literature.
Focus100/100
Scores 100 on focus: every indexed activity sits in a single field, machine learning for healthcare (EHR modeling), with no dispersion across topics. With only one catalogued paper, this perfect concentration reflects a very narrow evidentiary base rather than a proven long-term specialization.
Collaboration14/100
Scores 14 on collaboration: the indexed work lists only one distinct coauthor, Matthew McDermott, indicating a strongly two-person working pattern. With a single paper in scope, this is a preliminary signal of a tightly paired collaboration rather than evidence of limited broader networking.
Activity—
Not yet scored: the snapshot lacks publication-frequency or yearly productivity measures, so this axis — tracking how actively Pollet publishes across years — remains unavailable; the single 2026 arXiv entry is the only temporal data point so far.
Reproducibility55/100
Scores 55, averaged over one assessed paper. The 2026 evaluation study is rated medium for reproducibility readiness: credible methodology and a clear benchmark design, but code had not yet been released at review time, leaving a key gap for independent verification.
Research summary
Florent Pollet is a researcher working at the intersection of machine learning and healthcare, focused on electronic health record (EHR) modeling and, specifically, on how progress in health AI is evaluated. Evolve's reviewed catalog currently contains one paper, a 2026 arXiv preprint titled 'Rethinking How We Evaluate Methodological Progress in Health AI,' on which Pollet is first author (not corresponding). The study asks whether algorithm rankings on EHR benchmarks are stable: it finds that rankings largely transfer from random tasks and across two ICU datasets, with a wide-sparse XGBoost configuration emerging as the strongest performer. Evolve's review rates the work credible but small-scale, with code not yet released and medium reproducibility readiness. Bibliometrically the profile is narrow: a single catalogued field (machine learning for healthcare), one distinct coauthor — Matthew McDermott, the recurring collaborator — and a perfect focus score of 100 alongside a low collaboration score of 14 and a reproducibility score of 55 from one assessed paper. Output, impact, and activity cannot be scored from current data. No institutional affiliation or ORCID is publicly linked, so identity is name-resolved from bibliographic records. This is an early-stage profile expected to sharpen as more of Pollet's work is indexed and cited.
Collaboration pattern
Bibliometric data for this profile show exactly one distinct coauthor: Matthew McDermott, also the recurring collaborator Evolve identifies. The collaboration dimension scores 14/100, reflecting an extremely concentrated two-author working pattern rather than a broad network. With only one catalogued paper, this is an early and narrow signal — the true breadth of Pollet's collaborations cannot be assessed until more papers with full author lists are indexed.
Research keywords
Highlights
First-author study rethinks health-AI evaluation
Rankings transfer across tasks and ICU datasets
Credible but small-scale; code still pending
Timeline
- 2026First-author arXiv preprint 'Rethinking How We Evaluate Methodological Progress in Health AI' enters the Evolve catalog — the only indexed paper in this profile so far.
- 2026The study's evaluation shows EHR algorithm rankings largely transfer from random tasks and across two ICU datasets, with wide-sparse XGBoost winning.
- 2026Evolve review rates the paper medium reproducibility readiness (score 55), with code not yet released; indexed collaboration is confined to one coauthor, Matthew McDermott.
Papers in the Evolve catalog
Frequent co-authors
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.
- Bibliometrics in this profile are derived from OpenAlex-sourced public data only; no ORCID, institutional, or country information was available for name resolution.
- The reviewed catalog contains a single paper (2026 arXiv), so every dimension rests on a very small evidence base and the timeline covers only one year.
- Output, impact, and activity axes are unscored because publication counts, citations, and productivity data are not present in the snapshot.
- The reproducibility score of 55 averages over just one assessed paper and reflects code not yet released at review time.