Purpose:This study examined potential differences in strength,muscle morphology,and motor unit(MU)behavior of the abductor digiti minimi(ADM)between normal-fat(NF)and over-fat(OF)males.Methods:Dual-energy X-ray absorp...Purpose:This study examined potential differences in strength,muscle morphology,and motor unit(MU)behavior of the abductor digiti minimi(ADM)between normal-fat(NF)and over-fat(OF)males.Methods:Dual-energy X-ray absorptiometry assessed percent body fat(%BF).Ultrasonography determined muscle cross-sectional area(CSA),echo intensity(EI),and subcutaneous fat(s FAT).MU behavior was assessed during isometric muscle actions at 50%of maximal voluntary contraction(MVC)by analyzing the y-intercepts and slopes for the MU action potential amplitude(MUAPAMP)vs.recruitment threshold(RT)relationships,the A and B terms for the mean firing rate(MFR)vs.RT relationships,and normalized electromyographic amplitude(N-EMGRMS).MU firing times and waveforms were validated with reconstruct-and-test and spike trigger average procedures.Results:%BF was greater for OF(25.70%±5.40%)than NF(16.50%±2.20%;p<0.001).MVC was greater for NF(27.13±7.16)N than OF([19.89±4.96]N;p=0.014).CSA was greater for NF(2.48±0.39)cm^(2)than OF([1.95±0.47]cm^(2);p=0.011).The y-intercepts for the MUAPAMPvs.RT relationships were greater for NF(0.283±0.254)m V than OF([-0.221±0.659]m V;p=0.004).The B terms for the MFR vs.RT relationships were greater for NF(-0.024±0.003)pps/%MVC than OF([-0.031±0.009]pps/%MVC;p=0.038).N-EMGRMSwas similar between groups(p=0.463).Conclusion:Maximal strength,muscle size,and MU recruitment and firing rate patterns for a non-weight bearing muscle differed between normal-fat and over-fat males.展开更多
智能辅助诊治一直是人工智能(artificial intelligence,AI)应用的热点。近年来,大语言模型(large language model,LLM)在医疗领域中提供创新的健康大数据处理方法,提升了医疗实践的效率和效果。然而,由于LLM语料来源混杂,易生成“幻觉...智能辅助诊治一直是人工智能(artificial intelligence,AI)应用的热点。近年来,大语言模型(large language model,LLM)在医疗领域中提供创新的健康大数据处理方法,提升了医疗实践的效率和效果。然而,由于LLM语料来源混杂,易生成“幻觉”现象(即虚构或错误信息),在强调专业性与安全性的医学垂直领域(如临床诊断)中引发担忧,导致其临床应用尚未得到医学界广泛认可。为保证LLM在辅助诊疗过程中有更准确、稳定、专业的结果,研究人员引入了检索增强生成(retrieval-augmented generation,RAG)技术,通过实时检索权威医学知识库(如临床指南、病例数据库)增强生成过程。目前,RAG技术已在慢性病管理、肿瘤诊断辅助及个性化治疗方案推荐等方面验证了其有效性,显著提升了模型输出的准确率与临床一致性。但RAG技术仍面临多模态数据对齐、动态知识库更新延迟等技术挑战,需进一步突破,以推动AI在辅助诊疗中的规模化应用。RAG技术通过有效减少模型“幻觉”现象的发生,使AI辅助疾病诊治的决策更具可解释性与可信度,为医疗健康数字化提供了新的技术路径。展开更多
文摘Purpose:This study examined potential differences in strength,muscle morphology,and motor unit(MU)behavior of the abductor digiti minimi(ADM)between normal-fat(NF)and over-fat(OF)males.Methods:Dual-energy X-ray absorptiometry assessed percent body fat(%BF).Ultrasonography determined muscle cross-sectional area(CSA),echo intensity(EI),and subcutaneous fat(s FAT).MU behavior was assessed during isometric muscle actions at 50%of maximal voluntary contraction(MVC)by analyzing the y-intercepts and slopes for the MU action potential amplitude(MUAPAMP)vs.recruitment threshold(RT)relationships,the A and B terms for the mean firing rate(MFR)vs.RT relationships,and normalized electromyographic amplitude(N-EMGRMS).MU firing times and waveforms were validated with reconstruct-and-test and spike trigger average procedures.Results:%BF was greater for OF(25.70%±5.40%)than NF(16.50%±2.20%;p<0.001).MVC was greater for NF(27.13±7.16)N than OF([19.89±4.96]N;p=0.014).CSA was greater for NF(2.48±0.39)cm^(2)than OF([1.95±0.47]cm^(2);p=0.011).The y-intercepts for the MUAPAMPvs.RT relationships were greater for NF(0.283±0.254)m V than OF([-0.221±0.659]m V;p=0.004).The B terms for the MFR vs.RT relationships were greater for NF(-0.024±0.003)pps/%MVC than OF([-0.031±0.009]pps/%MVC;p=0.038).N-EMGRMSwas similar between groups(p=0.463).Conclusion:Maximal strength,muscle size,and MU recruitment and firing rate patterns for a non-weight bearing muscle differed between normal-fat and over-fat males.
文摘智能辅助诊治一直是人工智能(artificial intelligence,AI)应用的热点。近年来,大语言模型(large language model,LLM)在医疗领域中提供创新的健康大数据处理方法,提升了医疗实践的效率和效果。然而,由于LLM语料来源混杂,易生成“幻觉”现象(即虚构或错误信息),在强调专业性与安全性的医学垂直领域(如临床诊断)中引发担忧,导致其临床应用尚未得到医学界广泛认可。为保证LLM在辅助诊疗过程中有更准确、稳定、专业的结果,研究人员引入了检索增强生成(retrieval-augmented generation,RAG)技术,通过实时检索权威医学知识库(如临床指南、病例数据库)增强生成过程。目前,RAG技术已在慢性病管理、肿瘤诊断辅助及个性化治疗方案推荐等方面验证了其有效性,显著提升了模型输出的准确率与临床一致性。但RAG技术仍面临多模态数据对齐、动态知识库更新延迟等技术挑战,需进一步突破,以推动AI在辅助诊疗中的规模化应用。RAG技术通过有效减少模型“幻觉”现象的发生,使AI辅助疾病诊治的决策更具可解释性与可信度,为医疗健康数字化提供了新的技术路径。
