目的探讨清热利湿方治疗糖尿病肠病疗效及对患者生活质量的影响。方法将48例湿热内蕴证糖尿病肠病患者按随机数字表法分观察组和对照组,每组24例;两组均口服美沙拉嗪肠溶片,观察组加予清热利湿方日2次口服治疗,统计肠炎疾病量表(inflamm...目的探讨清热利湿方治疗糖尿病肠病疗效及对患者生活质量的影响。方法将48例湿热内蕴证糖尿病肠病患者按随机数字表法分观察组和对照组,每组24例;两组均口服美沙拉嗪肠溶片,观察组加予清热利湿方日2次口服治疗,统计肠炎疾病量表(inflammatory bowel disease questionnaire,IBDQ)评分及相关症状体征。结果观察组与对照组IBDQ评分各分项得分治疗后均高于治疗前,但观察组分数增加较为明显,且观察组在全身症状、肠道症状、情感功能、社会功能、总分等方面得分增加均高于对照组,在全身症状改善方面尤为显著。两组在生活质量量表(the MOS item short from health survey,SF-36)量表中生理机能(physical functioning,PF)、生理职能(role physical,RP)、活力(vitality,VT)、精神健康(mental health,MH)、整体健康(general health,GH)等方面治疗前与治疗后差异有统计学意义(均P<0.05)。观察组总有效率91.67%(22/24),两组间疗效比较差异有统计学意义(P<0.05)。两组病例均耐受良好,未见明显不良反应,1年内复发率观察组低于对照组(均P<0.05)。结论清热利湿方可以有效清热利湿,改善患者病情,是针对糖尿病肠病的一种有效治疗方案,可以提高湿热内蕴证糖尿病肠病患者的治疗有效率,明显提高患者的生活质量,值得临床进一步研究及推广。展开更多
在过去十年中,数据量、算法和高性能计算的巨大进步已将人工智能(AI)推向了高效用前沿。尤其是在医学影像和数字病理领域,AI的辅助诊疗系统成为了学术与医疗界共同关注的焦点。然而,随着对隐私保护问题的日益重视、政策法规的完善,数据...在过去十年中,数据量、算法和高性能计算的巨大进步已将人工智能(AI)推向了高效用前沿。尤其是在医学影像和数字病理领域,AI的辅助诊疗系统成为了学术与医疗界共同关注的焦点。然而,随着对隐私保护问题的日益重视、政策法规的完善,数据共享成为亟待解决的问题。联邦学习作为一种“数据不动模型动”的范式,为数据的隐私保护与共享提供了一种新思路。本文提出一种联邦学习框架,在确保医疗影像数据本地化及安全性的同时,充分利用多中心肺炎影像数据集,训练高精度的人工智能模型,辅助肺炎影像诊断。通过本地化模型训练和参数服务器的梯度聚合,在不违反数据隐私的前提下,实现了模型的优化与更新。本研究的成果不仅提高了肺炎诊断的准确性和效率,而且扩大了系统的样本量和数据维度,为医疗大数据应用中高精度模型的构建提供了有力支撑,进而有助于提供更丰富和高质量的医疗服务,为社会公共医疗事业的发展贡献重要价值。In the past decade, the tremendous advancements in data volume, algorithms, and high-performance computing have propelled artificial intelligence (AI) to the forefront of efficiency. This is especially true in the field of medical imaging and digital pathology, where AI-assisted diagnostic systems have become a focal point of attention in both the academic and medical communities. However, with increasing attention to privacy protection and the perfection of policies and regulations, data sharing has become an urgent issue to address. Federated learning, as a paradigm of “model moving without data moving”, offers a new approach to the privacy protection and data sharing. This paper proposes a federated learning framework that, while ensuring the localization and security of medical imaging data, fully utilizes multi-center pneumonia imaging datasets to train highly accurate AI models to assist in pneumonia image diagnosis. Through localized model training and gradient aggregation on parameter servers, model optimization and updating are achieved without violating data privacy. The outcomes of this study not only improve the accuracy and efficiency of pneumonia diagnosis, but also expand the sample size and data dimensions of the system, providing powerful support for the construction of high-precision models for medical big data applications. This, in turn, helps to provide richer and higher quality medical services, contributing significant value to the development of public medical services.展开更多
文摘目的探讨清热利湿方治疗糖尿病肠病疗效及对患者生活质量的影响。方法将48例湿热内蕴证糖尿病肠病患者按随机数字表法分观察组和对照组,每组24例;两组均口服美沙拉嗪肠溶片,观察组加予清热利湿方日2次口服治疗,统计肠炎疾病量表(inflammatory bowel disease questionnaire,IBDQ)评分及相关症状体征。结果观察组与对照组IBDQ评分各分项得分治疗后均高于治疗前,但观察组分数增加较为明显,且观察组在全身症状、肠道症状、情感功能、社会功能、总分等方面得分增加均高于对照组,在全身症状改善方面尤为显著。两组在生活质量量表(the MOS item short from health survey,SF-36)量表中生理机能(physical functioning,PF)、生理职能(role physical,RP)、活力(vitality,VT)、精神健康(mental health,MH)、整体健康(general health,GH)等方面治疗前与治疗后差异有统计学意义(均P<0.05)。观察组总有效率91.67%(22/24),两组间疗效比较差异有统计学意义(P<0.05)。两组病例均耐受良好,未见明显不良反应,1年内复发率观察组低于对照组(均P<0.05)。结论清热利湿方可以有效清热利湿,改善患者病情,是针对糖尿病肠病的一种有效治疗方案,可以提高湿热内蕴证糖尿病肠病患者的治疗有效率,明显提高患者的生活质量,值得临床进一步研究及推广。
文摘在过去十年中,数据量、算法和高性能计算的巨大进步已将人工智能(AI)推向了高效用前沿。尤其是在医学影像和数字病理领域,AI的辅助诊疗系统成为了学术与医疗界共同关注的焦点。然而,随着对隐私保护问题的日益重视、政策法规的完善,数据共享成为亟待解决的问题。联邦学习作为一种“数据不动模型动”的范式,为数据的隐私保护与共享提供了一种新思路。本文提出一种联邦学习框架,在确保医疗影像数据本地化及安全性的同时,充分利用多中心肺炎影像数据集,训练高精度的人工智能模型,辅助肺炎影像诊断。通过本地化模型训练和参数服务器的梯度聚合,在不违反数据隐私的前提下,实现了模型的优化与更新。本研究的成果不仅提高了肺炎诊断的准确性和效率,而且扩大了系统的样本量和数据维度,为医疗大数据应用中高精度模型的构建提供了有力支撑,进而有助于提供更丰富和高质量的医疗服务,为社会公共医疗事业的发展贡献重要价值。In the past decade, the tremendous advancements in data volume, algorithms, and high-performance computing have propelled artificial intelligence (AI) to the forefront of efficiency. This is especially true in the field of medical imaging and digital pathology, where AI-assisted diagnostic systems have become a focal point of attention in both the academic and medical communities. However, with increasing attention to privacy protection and the perfection of policies and regulations, data sharing has become an urgent issue to address. Federated learning, as a paradigm of “model moving without data moving”, offers a new approach to the privacy protection and data sharing. This paper proposes a federated learning framework that, while ensuring the localization and security of medical imaging data, fully utilizes multi-center pneumonia imaging datasets to train highly accurate AI models to assist in pneumonia image diagnosis. Through localized model training and gradient aggregation on parameter servers, model optimization and updating are achieved without violating data privacy. The outcomes of this study not only improve the accuracy and efficiency of pneumonia diagnosis, but also expand the sample size and data dimensions of the system, providing powerful support for the construction of high-precision models for medical big data applications. This, in turn, helps to provide richer and higher quality medical services, contributing significant value to the development of public medical services.
