Chenwei Wu

Zhejiang University of Science and Technology · China

Unclaimed · AI-structured profileOpenAlex

Early-career LLM researcher: post-training for telecom reasoning and auditable multi-agent clinical decision support

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-17) 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 scored: the snapshot provides no reliable total output count for this author — h-index and i10-index both stand at 0. This axis would measure cumulative publication volume and productivity across venues.

Impact—

Not scored: citation data is insufficient — h-index and i10-index are both 0 and no citation totals appear in the snapshot. This axis would measure citation influence and uptake of the author's work.

Focus100/100

Focus scores 100/100, the maximum. The OpenAlex field distribution lists exactly one field — natural language processing / large language models — indicating no disciplinary dispersion in the scored record. Topic counts reinforce this: topic modeling (3) and NLP techniques (3) sit alongside applied health topics (AI in healthcare and education, ML in healthcare, multimodal ML applications, 4 each), so concentration lies in language-model methods applied to healthcare and telecom rather than scattered interests.

Collaboration66/100

Collaboration scores 66/100. The snapshot records 27 distinct coauthors, with twelve recurring partners — including Yung Wei Shueh, Zhi-Jie Chen, Chia-Hsuan Hsu, Donghua Zhang, Tongze Zhang, and Shihao Yang — suggesting stable working clusters rather than one-off ties. All three cataloged papers place Chenwei Wu in a middle author position, none as corresponding author, consistent with a contributor role inside larger teams.

Activity67/100

Activity scores 67/100. OpenAlex counts-by-year attribute 9 works to 2026 — the only year in the snapshot — matching recent3yWorks of 9, so all recorded output is concentrated in one burst. The volume is solid for a first indexed year, but the single-year window means there is no multi-year history to confirm a sustained publication cadence.

Reproducibility55/100

Reproducibility scores 55/100, computed from only one scored paper (reproPapers: 1, reproMean: 55), so the base is thin. Within the catalog, TelecomGPT-R1 carries medium reproducibility readiness with solid ablations but flagged contamination and evaluation-setting caveats; DIASENTINEL is rated low due to private data and no released code. Public data and code releases would most directly move this axis.

Research summary

Chenwei Wu (Zhejiang University of Science and Technology, China) is an early-stage researcher working on large language models; the OpenAlex snapshot attributes all recorded output to a single year, 2026 (countsByYear: 9 works; recent3yWorks: 9). The profile is name-resolved from public bibliographic data, with no ORCID linked. Two arXiv papers anchor the reviewed catalog. TelecomGPT-R1 (2026) presents unified post-training — reinforcement learning with verifiable rewards, DAPO, and chain-of-thought distillation — for reasoning across heterogeneous telecom tasks; Evolve's review notes a reported 89.6% mean on GSMA Open Telco, compared against GPT-5 and Claude, while flagging contamination and evaluation-setting caveats (reproducibility: medium). DIASENTINEL (2026) is an auditable multi-agent system for guideline-grounded type 2 diabetes risk screening over electronic health records, built with retrieval-augmented generation and hallucination verification; the review credits honest reporting but notes private data, no released code, and XGBoost beating it on AUPRC (reproducibility: low). A third 2026 paper, MedCache on temporally valid memory for longitudinal clinical agents, is listed but not yet reviewed. Topic counts — AI in healthcare and education (4), multimodal ML applications (4), ML in healthcare (4), topic modeling (3), NLP techniques (3) — support an applied-language-model focus on healthcare and telecom. All cataloged papers are middle-author contributions, none yet as corresponding author.

Collaboration pattern

Collaboration scores 66/100 with 27 distinct coauthors recorded. Twelve recurring partners stand out — Yung Wei Shueh, Zhi-Jie Chen, Chia-Hsuan Hsu, Hsin-Ling Hsu, Donghua Zhang, Jun-En Ding, Tongze Zhang, Shihao Yang, Pengfei Hu, Fang-Ming Hung, Feng Liu, and Hei Ting — indicating several stable working clusters. On all three cataloged papers Chenwei Wu appears in a middle author position and never as corresponding author, a contributor pattern within larger teams.

Research keywords

Large language modelsLLM post-trainingReinforcement learning with verifiable rewardsDomain-specific LLMsChain-of-thought distillationTelecommunications applicationsClinical decision supportMulti-agent systemsRetrieval-augmented generationElectronic health recordsTopic modelingHallucination verification

Highlights

TelecomGPT-R1: unified post-training for reasoning across heterogeneous telecom tasks
Combines reinforcement learning with verifiable rewards, DAPO, and chain-of-thought distillation. Evolve's review notes a reported 89.6% mean on GSMA Open Telco, ahead of GPT-5 and Claude, with solid ablations — but contamination and evaluation-setting caveats persist. Reproducibility readiness: medium.
DIASENTINEL: auditable multi-agent system for guideline-grounded diabetes risk screening
A multi-agent LLM system for type 2 diabetes risk screening over electronic health records, using retrieval-augmented generation and hallucination verification. The review credits honest reporting, but private data, no released code, and XGBoost beating it on AUPRC cap the real-world evidence. Reproducibility readiness: low.
MedCache: temporally valid memory for longitudinal clinical agents
Listed in the Evolve catalog as a 2026 arXiv paper on efficient, temporally valid memory for longitudinal clinical agents. It has not yet been reviewed — no verdict, keywords, or reproducibility readiness are available.

Timeline

2026OpenAlex records 9 works for 2026, the only publication year in the snapshot (recent3yWorks: 9).
2026Published DIASENTINEL on arXiv: an auditable multi-agent system for guideline-grounded type 2 diabetes risk screening.
2026Published TelecomGPT-R1 on arXiv: unified post-training for reasoning across heterogeneous telecom tasks.
2026Co-authored MedCache on efficient, temporally valid memory for longitudinal clinical agents (listed in the catalog, not yet reviewed).

Papers in the Evolve catalog

Frequent co-authors

Yung Wei Shueh×1 Zhi-Jie Chen×1 Chia-Hsuan Hsu×1 Hsin-Ling Hsu×1 Donghua Zhang×1 Jun-En Ding×1 Tongze Zhang×1 Shihao Yang×1 Pengfei Hu×1 Fang-Ming Hung×1 Feng Liu×1 Hei Ting×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.

All bibliometrics come from a single OpenAlex snapshot (identity: openalex) with no ORCID linked; "Chenwei Wu" is a common name, so attribution is name-resolved from public bibliographic data and may be imperfect.
Output and impact dimensions are unscored: h-index and i10-index both stand at 0 and no citation totals are available in the snapshot.
The reproducibility score averages only one scored paper (reproPapers: 1), so the value of 55 rests on a thin base.
Only 3 papers are in the Evolve catalog versus 9 works indexed for 2026; MedCache has no review verdict, keywords, or reproducibility readiness yet.