James Zou

Stanford University · United States

Unclaimed · AI-structured profileORCID OpenAlex

Stanford AI researcher spanning healthcare, computational biology, and LLM agents for scientific discovery.

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—

Insufficient data to score. This axis would measure total publication volume, per-year output and authorship roles; the current OpenAlex snapshot lacks a complete verified works list for James Zou, so no output score is computed here.

Impact—

Citation data is insufficient to score this axis. Impact would measure citation counts, citation trajectory and highly cited works; the feed supplies an h-index (80) and i10-index (259), but per-paper citation evidence remains incomplete.

Focus0/100

Focus scores 0, reflecting deliberately broad coverage rather than concentration. The OpenAlex topic distribution is near-even across five clusters — AI in Healthcare and Education (89 works), Topic Modeling (71), Ethics and Social Impacts of AI (52), single-cell and spatial transcriptomics (51), and XAI (49) — and the observed field mix splits roughly into equal thirds across computational biology, AI for healthcare, and network science × LLM reasoning. Work moves between domains rather than deepening a single niche.

Collaboration67/100

Collaboration scores 67. Over the most recent three-year window the snapshot records 28 distinct co-authors and a recurring core of twelve frequent collaborators, including Lei Liu, Yikun Zhang, Jialin Chen, Wanjia Zhao, Rex Ying, and Wengong Jin. On all three Evolve-cataloged papers James Zou appears as a middle author without corresponding-author role, consistent with contributing expertise within larger, team-led projects rather than leading small groups.

Activity100/100

Activity scores 100, the strongest signal in this profile. The snapshot logs 208 works in the most recent three years, and counts by year climb almost continuously from 2 works in 2010 and 12 in 2015 to 26 in 2018, 65 in 2022, and 76 in 2023, peaking at 92 in 2024 and holding at 91 in 2025, with 25 already recorded for 2026. This indicates exceptionally sustained, rising research throughput.

Reproducibility85/100

Reproducibility scores 85, a mean across three Evolve-reviewed papers. Two of the three — LabAgent and the BioGlyph network-topology work — carry high reproducibility readiness, with execution-validated skill libraries and rigorous study design noted; SleepFM-2 is rated medium. The verdicts still flag single-run evidence for LabAgent and untested fairness for SleepFM-2, so the score reflects strong readiness rather than completed independent replication.

Research summary

James Zou is a researcher at Stanford University whose work, as indexed in public bibliometric data, spans AI for healthcare, computational biology, and large language model agents for science. The OpenAlex-derived snapshot shows a substantial and accelerating record: an h-index of 80, an i10-index of 259, and 208 works over the most recent three years, with annual output climbing from single digits in the early 2010s to 92 works in 2024 and 91 in 2025. Topical breadth is wide — the largest OpenAlex topics are Artificial Intelligence in Healthcare and Education (89 works), Topic Modeling (71), Ethics and Social Impacts of AI (52), single-cell and spatial transcriptomics (51), and Explainable AI (49) — so the profile reads as a multi-field bridge rather than a narrow specialist. Three 2026 arXiv papers in the Evolve catalog, on each of which he appears as a middle author, illustrate the current agenda: LabAgent (execution-validated skill libraries for reproducing published research), SleepFM-2 (a foundation model trained on two million hours of polysomnography), and BioGlyph (language-encoded network topology for LLM graph reasoning). Reproducibility readiness is comparatively strong, with a mean score of 85 across the three reviewed papers, though reviews flag single-run evidence and untested fairness in places.

Collaboration pattern

The collaboration score of 67 sits alongside 28 distinct co-authors in the last three years and a stable recurring group of twelve names, among them Lei Liu, Yikun Zhang, Jialin Chen, Wanjia Zhao, Rex Ying, Wengong Jin, and Hua Xu. All three cataloged 2026 papers list James Zou as a middle, non-corresponding author, suggesting contribution of specialized expertise inside larger team efforts rather than sole-leadership projects.

Research keywords

LLM agentsscientific reproducibilityAI for healthcaresleep foundation modelspolysomnographyLLM graph reasoningnetwork sciencesingle-cell transcriptomicsexplainable AIAI ethicstopic modelingself-supervised learning

Highlights

Execution-validated agent framework for reproducing published research
Evolve's review finds LabAgent pairs execution-validated skill libraries with run memory to reproduce published results, leading commercial agents, though evidence is single-run; reproducibility readiness is rated high.
Sleep foundation model trained on two million hours of polysomnography
SleepFM-2 learns transferable human physiology from two million hours of sleep data, beating interpretable baselines and transferring to wearables; the review notes impressive scale but retrospective design and untested fairness. Reproducibility readiness: medium.
Language-encoded network topology for LLM graph reasoning
The BioGlyph work encodes graph topology as named structural roles, boosting frozen 8B LLM performance on graph QA over protein interaction networks; gains fade on frontier models and sparse graphs. A rigorous study with high reproducibility readiness.

Timeline

2026LabAgent introduces execution-validated skill libraries and run memory for AI agents that customize research hubs and reproduce published results.
2026SleepFM-2 presents a foundation model trained on two million hours of sleep data, transferring to wearables and phenome-wide disease prediction.
2026BioGlyph shows that language-encoded structural roles let frozen 8B LLMs reason about complex networks such as protein interaction graphs.

Papers in the Evolve catalog

Frequent co-authors

Lei Liu×1 Yikun Zhang×1 Jialin Chen×1 Wanjia Zhao×1 Rex Ying×1 Wengong Jin×1 Hua Xu×1 Tianyu Liu×1 Hongyu Zhao×1 Rahul Thapa×1 Christopher Sun×1 William Theodor Lehn-Schioler×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.

Bibliometric figures (h-index 80, i10-index 259, yearly counts, 208 recent works) derive from an OpenAlex-sourced snapshot and may lag behind live citation records.
The profile is name-resolved from public bibliographic data; 'James Zou' is a shared name, so identity should be confirmed against ORCID 0000-0001-8880-4764.
Output and impact dimensions are unscored because per-publication citation data is insufficient in this snapshot.
All three cataloged papers are 2026 arXiv preprints and may not yet have completed peer review.