Jiandong Gao

Jiandong Gao


Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 November 2022 in Journal of Clinical Sleep Medicine
Verified
Journal Article
Closed access
YDYiming Ding
Yiming Ding
YSYuechuan Sun
Yuechuan Sun
YLYanru Li
Yanru Li
HWHuijun Wang
Huijun Wang
QFQiang Fang
Qiang Fang
DHDemin Han
Demin Han
Study Objectives:To screen all of the obstructive sleep apnea (OSA)-characteristic pronunciations, explore the pronunciations which provide a better OSA classification effect than those used previo...

31 March 2025 in Frontiers in Endocrinology
Verified
Journal Article
Open access
PDF file
ZLZhaoxiang Liu
Zhaoxiang Liu
LZLianhao Zhou
Lianhao Zhou
WZWenhui Zhao
Wenhui Zhao
LJLixia Jin
Lixia Jin
JZJinping Zhang
Jinping Zhang
JGJiandong Gao
Jiandong Gao
Purpose: This study investigates the incidence, predictors, and preventive strategies for microvascular complications in type 2 diabetes patients in community settings.Methods: Data were collected from 3,008 type 2 diabetes patients enrolled across 31 clinics in Beijing and Hebei. Prevalence and incidence of diabetic kidney disease (DKD), diabetic retinopathy (DR), and diabetic peripheral neuropathy (DPN) were assessed. Predictors were identified using XGBoost and Cox regression, and the impact of lifestyle and multifactorial interventions (MFI) was analyzed.Results: The prevalence of DKD, DR, and DPN were 39.5%, 26.2%, and 27.1%, respectively, with incidences of 74, 21, and 28 per 1000-person year. XGBoost identified that diabetes duration, age, HbA1c, FBG, triglyceride, BP, serum creatinine, proteinuria, aspirin and statin use were associated with those microvascular complications. The risk of DKD increased more rapidly as HbA1c exceeded 7.5% and decreased as blood pressure was maintained below 120/70 mmHg. Cox regression models showed that community-based intervention, including lifestyle modifications, were associated with a lower risk of DR and DPN. The study also found that higher variability in HbA1c and albumin-to-creatinine ratio (ACR) was associated with an increased risk of microvascular complications.Conclusions: Community-based interventions significantly reduce the of DR and DPN, highlighting the need for individualized glycemic and BP management in primary care. The findings emphasize the importance of comprehensive management strategies to prevent the development and progression of microvascular complications in type 2 diabetes patients.Clinical trial registration: http://www.chictr.org.cn/, identifier ChiCTR-TRC-13003222.

21 January 2026 in arXiv
Verified
Preprint
Open access
PDF file
YCYutong Chen
Yutong Chen
JGJiandong Gao
Jiandong Gao
JWJi Wu
Ji Wu
Training Large Reasoning Model (LRM) is usually unstable and unpredictable, especially on hard problems or weak foundation models. We found that the current post-training scaling strategy can still improve on these cases. We propose CoScale-RL, a novel scaling strategy with better data and computational efficiency. We first scale up solutions to make problems solvable. The core idea is to collect multiple solutions for each problem, rather than simply enlarging the dataset. Then, we scale up rollout computation to stabilize Reinforcement Learning. We further leverage a model merge technique called Re-distillation to sustain or even improve computational efficiency when scaling up. Our method significantly improves data and computational efficiency, with an average 3.76$\times$ accuracy improvement on four benchmarks. CoScale-RL is able to improve an LRM's ability boundary without an extensive SFT dataset. Our method provides a new scaling direction to further improve LRM's reasoning ability.

29 January 2026 in arXiv
Verified
Preprint
Open access
PDF file
MYMutian Yang
Mutian Yang
ZZZisen Zhan
Zisen Zhan
YCYutong Chen
Yutong Chen
HLHaolin Li
Haolin Li
KWKaiwen Wang
Kaiwen Wang
JWJi Wu
Ji Wu
Catastrophic forgetting during knowledge injection severely undermines the continual learning capability of large language models (LLMs). Although existing methods attempt to mitigate this issue, they often lack a foundational theoretical explanation. We establish a gradient-based theoretical framework to explain catastrophic forgetting. We first prove that strongly negative gradient similarity is a fundamental cause of forgetting. We then use gradient similarity to identify two types of neurons: conflicting neurons that induce forgetting and account for 50%-75% of neurons, and collaborative neurons that mitigate forgetting and account for 25%-50%. Based on this analysis, we propose a knowledge injection method, Collaborative Neural Learning (CNL). By freezing conflicting neurons and updating only collaborative neurons, CNL theoretically eliminates catastrophic forgetting under an infinitesimal learning rate eta and an exactly known mastered set. Experiments on five LLMs, four datasets, and four optimizers show that CNL achieves zero forgetting in in-set settings and reduces forgetting by 59.1%-81.7% in out-of-set settings.

24 July 2025 in arXiv
Verified
Preprint
Open access
PDF file
MYMutian Yang
Mutian Yang
JGJiandong Gao
Jiandong Gao
JWJi Wu
Ji Wu
While large language models (LLMs) leverage both knowledge and reasoning during inference, the capacity to distinguish between them plays a pivotal role in model analysis, interpretability, and development. Inspired by dual-system cognitive theory, we propose a cognition attribution framework to decouple the contribution of knowledge and reasoning. In particular, the cognition of LLMs is decomposed into two distinct yet complementary phases: knowledge retrieval (Phase 1) and reasoning adjustment (Phase 2). To separate these phases, LLMs are prompted to generate answers under two different cognitive modes, fast thinking and slow thinking, respectively. The performance under different cognitive modes is analyzed to quantify the contribution of knowledge and reasoning. This architecture is employed to 15 LLMs across 3 datasets. Results reveal: (1) reasoning adjustment is domain-specific, benefiting reasoning-intensive domains (e.g., mathematics, physics, and chemistry) and potentially imparing knowledge-intensive domains. (2) Parameter scaling improves both knowledge and reasoning, with knowledge improvements being more pronounced. Additionally, parameter scaling make LLMs reasoning significantly more prudent, while moderately more intelligent. (3) Knowledge primarily resides in lower network layers, while reasoning operates in higher layers. Our framework not only helps understand LLMs from a "decoupling" perspective, but also provides new insights into existing research, including scaling laws, hierarchical knowledge editing, and limitations of small-model reasoning.