A5024536314- Electroconvulsive Therapy Studies
- Treatment of Major Depression
- Bipolar Disorder and Treatment
- Natural Language Processing Techniques
- Topic Modeling
- Hemoglobinopathies and Related Disorders
- Algorithms and Data Compression
- Data Management and Algorithms
- Neurological disorders and treatments
- Cardiac Imaging and Diagnostics
- Multimedia Communication and Technology
- Coronary Interventions and Diagnostics
- Takotsubo Cardiomyopathy and Associated Phenomena
- Semantic Web and Ontologies
- Thyroid Disorders and Treatments
- Optimal Experimental Design Methods
- Text Readability and Simplification
- Viral Infectious Diseases and Gene Expression in Insects
- Cardiac Arrhythmias and Treatments
- Cardiovascular Disease and Adiposity
- Statistical Methods in Clinical Trials
- Artificial Intelligence in Healthcare and Education
- Vascular Procedures and Complications
Sun Yat-sen University
2022-2023
Southern Medical University
2018-2022
Instruction Fine-Tuning (IFT) significantly enhances the zero-shot capabilities of pretrained Large Language Models (LLMs). While coding data is known to boost LLM reasoning abilities during pretraining, its role in activating internal capacities IFT remains understudied. This paper investigates a key question: How does impact LLMs' stage? To explore this, we thoroughly examine across different proportions, model families, sizes, and domains, from various perspectives. Specifically, create...
Instruction-finetuning (IFT) has become crucial in aligning Large Language Models (LLMs) with diverse human needs and shown great potential medical applications. However, previous studies mainly fine-tune LLMs on biomedical datasets limited diversity, which often rely benchmarks or narrow task scopes, hence significantly limit the effectiveness their instruction-following ability generalizability. To bridge this gap, we propose creating a diverse, machine-generated IFT dataset,...
Objectives ECT is a rapid and effective treatment for depression. While efficacy often remarkable over the initial 3–4 sessions, of later sessions less rapid, side-effects, especially cognitive impairment limit its use. To preliminarily compare acceptability novel hybrid-ECT (HECT) protocol patients with major depressive disorder (MDD) standard ECT, we conducted this pilot trial. Methods Thirty were randomly assigned to or HECT. Both arms received three (phase 1) but, in phase 2, HECT arm...
Objective: Little is known about the effectiveness and cognitive side-effects of electroconvulsive therapy (ECT) in young adults with treatment-resistant depression (TRD). The primary aim this prospective longitudinal observational trial was to examine clinical features outcomes TRD undergoing ECT. Methods: Changes depressive symptoms objective subjective function were assessed using repeated evaluation at baseline, after each ECT session, one-month follow-up Montgomery-Äsberg Depression...
Abstract Background Electroconvulsive therapy (ECT) is the most rapid and effective treatment for patients with depression, ECT can achieve remarkable antidepressant effects in initial 3–4 sessions, but significant side limit its use. However, recent low-charge electrotherapy (LCE) studies have demonstrated or antipsychotic significantly fewer effects. The aim of this study to propose a novel two-step charge set strategy treatment, referred as Hybrid-ECT, decrease by using low while...
Reinforcement learning with human feedback~(RLHF) is critical for aligning Large Language Models (LLMs) preference. Compared to the widely studied offline version of RLHF, \emph{e.g.} direct preference optimization (DPO), recent works have shown that online variants achieve even better alignment. However, alignment requires on-the-fly generation new training data, which costly, hard parallelize, and suffers from varying quality utility. In this paper, we propose a more efficient data...
Instruction Fine-Tuning (IFT) significantly enhances the zero-shot capabilities of pretrained Large Language Models (LLMs). While coding data is known to boost reasoning abilities during LLM pretraining, its role in activating internal capacities IFT remains understudied. This paper investigates a key question: How does impact LLMs' stage? To explore this, we thoroughly examine across different proportions, model families, sizes, and domains, from various perspectives. Specifically, create...
Objective To study the safety of transradial approach for coronary interventions.Methods Coronary intervention was tried to be performed via in 80 patients with heart disease.Results Transradial puncture failed 4 cases and interventions were trarsfemoral or transbrachial instead.Coronary angiography 3 although succeeded.Both succeeded rest 73 cases.Conclusions is feasible,the radial artery some what diificult,and guiding catheter support poor extent,complications suffering are minor after...
In phase II oncology trials, two-stage design allowing early stopping for futility and/or efficacy is frequently used. However, this based on frequentist statistical approaches could not guarantee a high posterior probability of attending the pre-specified clinically interesting rate from Bayesian perspective. Here, we proposed new enabling terminating as well futility. addition to uninteresting and response rate, prior distribution minimum threshold probabilities lengths highest density...