Xi Chen (HKUST): LLM-agent reinforcement learning and physics-informed computational hemodynamics
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-14) 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 data: Evolve has not indexed a full publication record for this researcher, so output is unscored. This axis would measure total publication volume, venues, and authorship roles; only two 2026 arXiv preprints are currently on file.
Impact—
Insufficient data: no citation counts or h-index are available, and 2026 preprints are too recent for citations to accumulate. This axis would measure citation influence, citing works, and field-normalized impact once a fuller record exists.
Focus0/100
The focus score is 0, driven by an exactly even 50/50 field distribution: half of the indexed output sits in machine learning for LLM agents and reinforcement learning, half in computational hemodynamics and medical-imaging AI. These are substantively different research lines — one algorithmic work on agent RL credit assignment, one clinical-modeling work on coronary hemodynamics — so the profile currently reads as a dual-track portfolio rather than a concentrated single agenda.
Collaboration59/100
Collaboration scores 59, with 18 distinct coauthors in the indexed record and a twelve-person frequent-coauthor list spanning both research lines — Qiang Zhang, Zheng-Jun Zha, Kaipeng Zhang, Pengjun Xie, and Jianchuan Yang among them. Authorship roles differ across the two papers: first author on the hemodynamics work, middle author on ArenaFlow, indicating a mix of led and supporting positions within moderately sized teams.
Activity—
Insufficient data: recent publication cadence cannot be established from two same-year preprints. This axis would measure annual output rhythm and ongoing research tempo; it will be scored as more indexed works accrue over time.
Reproducibility55/100
Reproducibility averages 55 across the 2 papers Evolve has assessed, both rated medium readiness. For ArenaFlow, Evolve flags judge circularity in the LLM-as-a-Judge setup and missing variance tests; for the hemodynamics PINN, the reported 2.02% pressure-drop error and 93.8% FFR accuracy are methodologically thorough but rest on a 32-patient cohort. The score rests on a small sample and may move as more papers are reviewed.
Research summary
Xi Chen is a researcher at Hong Kong University of Science and Technology whose indexed output splits evenly — 50/50 — between two distinct lines of work: machine learning for LLM agents and reinforcement learning, and computational hemodynamics with medical-imaging AI. Evolve's catalog holds two 2026 arXiv preprints. The first, a first-author paper on physics-informed hemodynamic modeling, claims angiography-to-3D-hemodynamics prediction with a 2.02% pressure-drop error and 93.8% FFR accuracy; Evolve judges the methodology thorough but flags a small clinical cohort (n=32). The second, ArenaFlow, a middle-authorship contribution, extends tournament-based agent RL with step- and skill-level credit propagation, reported at +19.5 over ArenaRL on Open-Travel, though Evolve notes judge circularity and missing variance tests. Collaboration scores 59, with 18 distinct coauthors in the indexed record; frequent collaborators among the listed group include Qiang Zhang, Zheng-Jun Zha, Kaipeng Zhang, and Pengjun Xie. Reproducibility averages 55 across the two assessed papers, both rated medium readiness. Output, impact, and activity axes remain unscored because the underlying bibliometric data is not yet sufficient, and the indexed record so far consists of recent preprints rather than a mature citation trail.
Collaboration pattern
The indexed record shows 18 distinct coauthors and a collaboration score of 59, with a twelve-person frequent-coauthor list spanning both research lines — Qiang Zhang, Zheng-Jun Zha, Kaipeng Zhang, Pengjun Xie, and Jianchuan Yang among them. Authorship roles vary: first author on the hemodynamics paper, middle author on ArenaFlow, suggesting both led and supporting contributions within larger teams.
Research keywords
Highlights
Hierarchical credit propagation for open-ended agent RL
First-author physics-informed hemodynamic modeling
Timeline
- 2026Posted a first-author arXiv preprint on physics-informed hemodynamic modeling for data-free prediction and sparse-data assimilation.
- 2026Contributed as a middle author to ArenaFlow, extending tournament-based agent RL with hierarchical credit propagation.
- 2026Both papers entered the Evolve catalog, establishing a 50/50 dual-field split and a coauthor network of 18 in the indexed record.
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.
- 'Xi Chen' is a highly common name; this profile is name-resolved from public bibliographic data and may merge or split same-name authors.
- Dimension scores and raw metrics derive from public bibliometric data (OpenAlex-based) and cover only what Evolve has indexed.
- Both catalog papers are 2026 arXiv preprints; no citation counts, h-index, venue record, or career history are available, so output, impact, and activity remain unscored.
- The reproducibility mean of 55 rests on only 2 assessed papers and may shift as more are reviewed.