Running Zhao

University of Chicago

Unclaimed · AI-structured profileORCID OpenAlex

Running Zhao: LLM multi-agent systems for research reproducibility and wearable digital health (University of Chicago)

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) 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—

No publication counts, venue tiers, or cumulative output data exist in the current snapshot, so output is unscored. This axis would measure total papers, authorship roles, and venues over time; the two reviewed preprints are too thin to compute it.

Impact—

Impact is unscored: the snapshot contains no citation counts, h-index, or highly-cited-paper flags. This axis would measure how much the community builds on the work; the 2026 preprints are too recent for citation accumulation, so no honest estimate is possible.

Focus0/100

Evolve scores focus at 0, and the underlying field distribution explains why: the cataloged output splits exactly evenly (0.5 and 0.5) between artificial intelligence — LLM-based multi-agent systems and NLP / digital health (LLMs for wearable data). With only two papers reviewed, one per field, there is no measurable concentration yet; the 0 reflects an even two-way split rather than confirmed lack of direction, and future output may consolidate.

Collaboration62/100

The collaboration score is 62, based on 22 distinct coauthors recorded in the bibliometric snapshot. Twelve collaborators recur across the record — including Weijie Liu, Wenhao Yuan, Jinfeng Xu, Zhanfeng Xu, Xiaoxi Zhang, Edith Cheuk-Han Ngai, Yusheng Tan, Sofia Angel, Ninghui Hao, Ash Arian, Nikita N. Dulin, and Jay Lin. Both cataloged papers are multi-author, middle-position contributions, consistent with team-based work rather than solo or leading-author output.

Activity—

Activity is unscored because the snapshot lacks per-year publication cadence, submission frequency, or timestamps beyond paper years. This axis would measure research tempo and recency over a multi-year window; two same-year 2026 entries cannot establish a rhythm.

Reproducibility38/100

Reproducibility is scored 38, the mean across the 2 reviewed papers. The DuDe paper–code discrepancy detector rates medium readiness, while the personalized sleep guidance work rates low — Evolve flags evaluation circularity from self-referenced metrics there, and a single-benchmark, LLM-judged setup in DuDe. A 38 is a modest starting point: artifacts and evaluation protocols are not yet demonstrated to be independently reusable.

Research summary

Running Zhao is a researcher at the University of Chicago whose visible work spans two areas recorded evenly in the bibliometric snapshot: LLM-based multi-agent systems, and NLP for digital health with wearable data. The Evolve-reviewed catalog holds two 2026 arXiv preprints, both with Zhao in a middle, non-corresponding author position — a contributor rather than lead role in the visible record. The first, DuDe, is a dual-detection multi-agent system for paper–code discrepancy detection, reporting F1 gains of up to +18.7 points that Evolve notes rest on a single benchmark and an author-validated LLM judge. The second distills a multi-agent sleep-guidance pipeline into 0.8B-parameter small language models that beat commercial LLMs on self-referenced metrics, with evaluation circularity tempering the headline claims. Collaboration is the strongest scored axis at 62, drawing on 22 distinct coauthors and 12 recurring collaborators; the 50/50 field split yields a focus score of 0, and reproducibility readiness averages 38 across the two papers (one medium, one low). Output, impact, and activity metrics are missing from the snapshot, so this is explicitly an early-stage profile: conclusions reflect only the reviewed papers and should be read as provisional rather than established.

Collaboration pattern

The visible pattern is team-based: both cataloged papers carry middle, non-corresponding authorship within multi-author groups, and the snapshot records 22 distinct coauthors with 12 recurring collaborators — Weijie Liu, Wenhao Yuan, Jinfeng Xu, Zhanfeng Xu, Xiaoxi Zhang, Edith Cheuk-Han Ngai, Yusheng Tan, Sofia Angel, Ninghui Hao, Ash Arian, Nikita N. Dulin, and Jay Lin. Work spans LLM multi-agent methods and wearable digital health, suggesting participation across groups rather than a single lab line.

Research keywords

LLM-based multi-agent systemsPaper–code discrepancy detectionResearch reproducibilityClaim verificationSalience filteringLLM distillationSmall language modelsWearable healthPersonalized sleep guidanceSynthetic data generation

Highlights

DuDe: dual-detection multi-agent system spots paper–code discrepancies
The 2026 arXiv paper proposes a dual-detection multi-agent LLM system for detecting discrepancies between papers and code, reporting F1 gains of up to +18.7 points. Evolve's verdict tempers the headline: the gains rest on one benchmark and an author-validated LLM judge; reproducibility readiness is rated medium.
0.8B small language models distilled for personalized sleep guidance
The 2026 arXiv paper distills a multi-agent LLM pipeline into 0.8B-parameter small language models that generate personalized sleep guidance from wearable data, outperforming commercial LLMs on self-referenced metrics; Evolve flags evaluation circularity, and reproducibility readiness is rated low.

Timeline

2026Co-authored 'DuDe: A Dual-Detection Multi-Agent System for Paper-Code Discrepancy Detection' (arXiv preprint) in a middle, non-corresponding author position; Evolve notes F1 gains of up to +18.7 points, resting on one benchmark and an author-validated LLM judge.
2026Co-authored 'Toward Personalized Sleep Guidance from Wearable Data Using Language Models' (arXiv preprint), distilling a multi-agent pipeline into 0.8B small language models for personalized sleep guidance from wearable data; reproducibility readiness is rated low.
2026Both cataloged works appear in the same year, marking the current start of the visible record; output, impact, and activity metrics remain unscored in the snapshot.

Papers in the Evolve catalog

Frequent co-authors

Weijie Liu×1 Wenhao Yuan×1 Jinfeng Xu×1 Zhanfeng Xu×1 Xiaoxi Zhang×1 Edith Cheuk-Han Ngai×1 Yusheng Tan×1 Sofia Angel×1 Ninghui Hao×1 Ash Arian×1 Nikita N. Dulin×1 Jay Lin×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.

Radar scores and coauthor counts come from a public bibliometric snapshot; the input does not specify the underlying index, and no citation, output, or activity data was included.
Only two papers (both 2026 arXiv preprints) are in the reviewed catalog; journal/conference publication, output volume, and career history fall outside this record.
Neither cataloged paper lists Running Zhao as first or corresponding author, so the extent of individual contribution cannot be assessed.
Identity is name-resolved from public bibliographic data via ORCID 0000-0003-2496-3429; country is not recorded in the input.