ML researcher at CUNY spanning time-series forecasting, spatio-temporal earth-system prediction, and clinical AI
4
papers here
4
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
Unscored: no publication-count or output metrics exist in the current bibliometric snapshot. This axis would measure annual output, venue tiers, and the balance between preprints and peer-reviewed papers across the record.
Impact—
Unscored: citation counts and influence metrics are not yet available for this profile. This axis would measure citations to the four reviewed papers, aggregate indices such as h-index, and downstream adoption of the methods.
Focus0/100
The focus axis scores 0, reflecting maximal breadth: the field distribution is an even 0.25 split across four subfields — time-series forecasting, spatio-temporal/earth-system forecasting, uncertainty quantification for earth-system forecasting, and clinical machine learning. Each reviewed 2026 paper sits in a different field. This exploratory, breadth-first pattern is common early on, though shared methodology suggests convergence may follow.
Collaboration59/100
Collaboration scores 59/100. The four reviewed papers involve 18 distinct coauthors, with 12 recurring partners including Yining Liu, Zhixiang Wu, and Chu Lin. Duan appears as a middle author on every catalog entry and as corresponding author on none, consistent with a steady contributor role inside larger project teams rather than solo or lead authorship.
Activity—
Unscored: publishing-rhythm and submission-cadence data are insufficient in the current snapshot. This axis would measure publishing frequency and research momentum; only four 2026 arXiv preprints are observable, so no trend can be established yet.
Reproducibility46/100
Reproducibility averages 46/100 across the four papers Evolve scored (reproMean 46, reproPapers 4): three are rated medium readiness (CoRe, AsyncCouple-Flow, MedTRACE) and one low (the decision-oriented UQ paper, which lacks released code and specifies no backbone). Recurring review flags — missing seeds, significance tests, and confidence intervals, plus padded citations — define the main path to improvement.
Research summary
Huiran Duan is a machine-learning researcher at the City University of New York whose reviewed work spans four application-oriented subfields: multivariate time-series forecasting, multimodal spatio-temporal and Earth-system forecasting, uncertainty quantification for Earth-system foundation models, and tool-augmented clinical decision support. The Evolve catalog includes four 2026 arXiv preprints, on all of which Duan appears as a middle author rather than corresponding author. Recurring methodological threads are visible despite the breadth: frequency-domain and relational learning objectives (CoRe), cross-modal coupling with flow-matching heads for asynchronous multimodal forecasting (AsyncCouple-Flow), agent-based evidence verification for clinical reasoning (MedTRACE), and conformal, decision-oriented risk control. Evolve's reviews are cautiously positive: gains are reported as consistent or promising — for example, a reported 4.2 percentage-point improvement over CARE on multimodal diagnosis and a claimed 18.7% reduction in decision regret — but all four papers are flagged for missing statistical rigor such as seeds, significance tests, and confidence intervals, and the uncertainty-quantification paper additionally lacks released code. Reproducibility averages 46/100 across the four scored papers (three medium, one low readiness), and a focus score of 0 reflects an even 0.25 split across four fields. With 18 distinct coauthors and 12 recurring partners, this is an early-stage, breadth-first, team-science profile that has not yet consolidated around a single niche.
Collaboration pattern
Across four 2026 arXiv papers, Duan worked with 18 distinct coauthors, 12 of whom recur — including Yining Liu, Zhixiang Wu, Chu Lin, Ji Lu, and Xianglong Wang. All catalog appearances are middle-author positions with no corresponding-author record, indicating a steady contributor within rotating project teams rather than a lab lead. Coauthor names span several subfields, consistent with the even four-way field split in the profile.
Research keywords
Highlights
CoRe: a parameter-free spectral and relational loss for multivariate forecasting
Asynchronous cross-modal coupling for spatio-temporal forecasting
Tool-augmented agents lift multimodal diagnosis by 4.2 points over CARE
Decision-oriented uncertainty quantification claims 18.7% regret reduction
Timeline
- 2026Posted the decision-oriented uncertainty quantification paper on risk control for Earth-system spatiotemporal foundation models to arXiv.
- 2026Released MedTRACE, tool-augmented multimodal clinical reasoning agents for evidence-grounded decision-making, on arXiv.
- 2026Posted AsyncCouple-Flow on asynchronous cross-modal coupling and flow matching for spatio-temporal forecasting to arXiv.
- 2026Posted CoRe, coherence and relational alignment for multivariate time-series forecasting, to arXiv.
Papers in the Evolve catalog
CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting
AsyncCouple-Flow: Asynchronous Cross-Modal Coupling and Flow Matching for Spatio-Temporal Forecasting
MedTRACE: Tool-Augmented Multimodal Clinical Reasoning Agents for Evidence-Grounded Decision-Making
Decision-Oriented Uncertainty Quantification for Risk Control in Earth System Spatiotemporal Foundation Models
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.
- Radar scores and bibliometrics derive from an OpenAlex-based snapshot; output, impact, and activity metrics were unavailable for this profile.
- All four reviewed papers are 2026 arXiv preprints; peer-review status is not reflected in the data.
- Duan is a middle author on every catalog entry with no corresponding-author record, so individual contribution share cannot be inferred.
- The reproducibility score (mean 46) covers only the four Evolve-reviewed papers, not the full publication record.