Cluster randomized crossover trials with very few clusters but multiple periods: which analyses for continuous outcomes should be used?
研究评估了多种统计方法在处理仅有少数集群的随机交叉试验中的表现，发现准确建模集群-周期相关性比选择固定或随机集群截距更重要。
0 citationsRead paper
Selected work
Representative Papers
An Auditable Pipeline for Fuzzy Full-Text Screening in Systematic Reviews: Integrating Contrastive Semantic Highlighting and LLM Judgment
Systematic reviews suffer from low full-text screening efficiency, primarily because critical evidence is dispersed across heterogeneous long documents, making static binary inclusion/exclusion rules inadequate. Method: We propose an auditable fuzzy screening pipeline that integrates contrastive semantic embeddings with large language model (LLM)-based judgment, supporting multi-label, progressive eligibility assessment. For the first time, we combine Mamdani-type fuzzy logic control with dynamic thresholding and confidence decay to enable fine-grained, interpretable, multi-criteria decision-making. The pipeline incorporates domain-adapted embeddings, contrastive cosine similarity, semantic highlighting, and LLM-based judgment modules. Results: On a fully positive dataset, recall per eligibility criterion reaches 75.0%–87.5%; strict adherence to all criteria achieves 50.0% inclusion rate; average screening time per document drops from 20 minutes to under 1 minute; and human-AI agreement reaches 96.1%. Our approach significantly improves recall, traceability, and screening efficiency.
0 citationsRead paper
An AI-Driven Live Systematic Reviews in the Brain-Heart Interconnectome: Minimizing Research Waste and Advancing Evidence Synthesis
Systematic reviews in the brain–heart interconnection (BHI) domain suffer from low synthesis efficiency, inconsistent methodological quality, and redundant research efforts. Method: We developed the first AI-driven, dynamic systematic review platform for BHI, integrating (i) automated PICOS extraction via a Bi-LSTM model (87% precision, 95.7% recall), (ii) knowledge graph–based semantic retrieval, (iii) LDA topic modeling, and (iv) RAG-augmented generation—where RAG-GPT-3.5 outperforms GPT-4 on graph- and topic-based queries. Contribution/Results: The platform fully automates literature screening, risk-of-bias assessment, evidence synthesis, and knowledge gap identification. It supports real-time updating, redundancy detection, and interactive exploration—enhancing timeliness, transparency, and clinical decision support. Its modular architecture ensures cross-domain adaptability across biomedical disciplines.
0 citationsRead paper
Recent publications
Latest Papers
Cluster randomized crossover trials with very few clusters but multiple periods: which analyses for continuous outcomes should be used?
研究评估了多种统计方法在处理仅有少数集群的随机交叉试验中的表现，发现准确建模集群-周期相关性比选择固定或随机集群截距更重要。
0 citationsRead paper
An Auditable Pipeline for Fuzzy Full-Text Screening in Systematic Reviews: Integrating Contrastive Semantic Highlighting and LLM Judgment
Systematic reviews suffer from low full-text screening efficiency, primarily because critical evidence is dispersed across heterogeneous long documents, making static binary inclusion/exclusion rules inadequate. Method: We propose an auditable fuzzy screening pipeline that integrates contrastive semantic embeddings with large language model (LLM)-based judgment, supporting multi-label, progressive eligibility assessment. For the first time, we combine Mamdani-type fuzzy logic control with dynamic thresholding and confidence decay to enable fine-grained, interpretable, multi-criteria decision-making. The pipeline incorporates domain-adapted embeddings, contrastive cosine similarity, semantic highlighting, and LLM-based judgment modules. Results: On a fully positive dataset, recall per eligibility criterion reaches 75.0%–87.5%; strict adherence to all criteria achieves 50.0% inclusion rate; average screening time per document drops from 20 minutes to under 1 minute; and human-AI agreement reaches 96.1%. Our approach significantly improves recall, traceability, and screening efficiency.
0 citationsRead paper
An AI-Driven Live Systematic Reviews in the Brain-Heart Interconnectome: Minimizing Research Waste and Advancing Evidence Synthesis
Systematic reviews in the brain–heart interconnection (BHI) domain suffer from low synthesis efficiency, inconsistent methodological quality, and redundant research efforts. Method: We developed the first AI-driven, dynamic systematic review platform for BHI, integrating (i) automated PICOS extraction via a Bi-LSTM model (87% precision, 95.7% recall), (ii) knowledge graph–based semantic retrieval, (iii) LDA topic modeling, and (iv) RAG-augmented generation—where RAG-GPT-3.5 outperforms GPT-4 on graph- and topic-based queries. Contribution/Results: The platform fully automates literature screening, risk-of-bias assessment, evidence synthesis, and knowledge gap identification. It supports real-time updating, redundancy detection, and interactive exploration—enhancing timeliness, transparency, and clinical decision support. Its modular architecture ensures cross-domain adaptability across biomedical disciplines.
0 citationsRead paper