甲氧西林耐药溶血性葡萄球菌的RAPD分型研究

目的建立随机扩增多态性DNA技术（RAPD）对甲氧西林耐药溶血性葡萄球菌（MRSH）的分型方法.方法收集本院2002年4月至2003年4月临床分离的MRSH 81株.共设计了5组引物进行RAPD,电泳条带采用SPSS 13.0软件分析,根据树状图分型.结果除引物H12和M13组无法扩增出条带外,其余的4种引物均可以将81株MRSH进行分型.其中引物ERIC2将其分为13型,而引物ERIC1R将其分为8型.两种引物均可以辨别出主要流行株.结论 RAPD技术采用引物ERIC2对MRSH分型具有快速、敏感、稳定性高等特点.在有适合引物的条件下,RAPD可以作为判断MRSH医院感染流行的初筛工具.     

基于随机森林算法检出hs-cTnT项目小偏移的实时质控方法

目的 利用机器学习随机森林（random forest，RF）算法，建立一种准确识别检测小偏移的实时质控方法，并以移动异常值之和的方法（Moving sum of outlier，MovSO）作为参比方法，评价新算法效能。 方法 收集来自航空总医院实验室信息系统导出的2016年1月至2021年8月在罗氏化学发光E601设备检测的hs-cTnT项目检测结果，按照规定的数据清洗规则筛选出54243个结果作为无偏数据，人为引入10个不同大小的偏移，生成相应的有偏数据，每种偏移下用RF与MovSO两种算法进行实验。采用分类模型标准及临床指标对算法进行评价。 结果 RF算法在步长为10时，对10个小偏移均能检出，FPR在4.0% ~ 4.7%，MNPed在12以下。除了在±1ng/L偏移时准确度为85%，其余8个偏移检出准确度均在90%以上；MovSO算法的最优步长为200，除了在1ng/L时偏移无法检出，对其他9个偏移可检出，FPR在3.5% ~ 4.6%之间，MNPed均在100以上，仅在5ng/L偏移时，识别准确度方可达89%。RF算法总体显著优于MovSO，RF可准确识别hs-cTnT的测量小偏移。 结论 基于机器学习算法建立的质控方法可以改进类似hs-cTnT等临床对检测质量要求较高的项目的测量准确度，为实验室质控方案提供了新思路。     

运动想象脑电多视角深度森林解码算法

针对运动想象脑电信号特征提取操作繁琐及解码精度低等问题,提出一种基于多视角深度森林的运动想象脑电解码算法。首先,通过子频带滤波及时间窗口划分对原始信号进行细粒度分析,生成空时频能量特征。然后,对上述空时频能量特征分别进行稀疏选择和时序扫描得到重要的浅层能量特征及多示例先验类别特征。继而,将上述两类特征进行融合构建运动想象脑电多视角特征集。最后,利用级联森林的逐层特征变换挖掘深层次的抽象特征进行脑电解码。根据脑机接口竞赛数据和自行采集的数据进行算法测试,并与单视角特征模型、传统共空间模式方法以及深度神经网络算法进行对比。在2个脑机接口竞赛数据集和1个真实数据集上分别取得了91.4%、75.2%和70.7%的最高平均分类准确率,结果表明该文所提多视角深度森林算法具有更优的分类识别准确率。     

人眼追踪远距离裸眼立体视觉检测系统的研制和测试

目的应用人眼追踪技术研制裸眼随机点立体视觉检查系统（glasses-free random dot stereo-test-system,GFRDSS）,检测正常视力男性青年的5 m距离立体视锐度。方法在不同照度环境下检测230名正常视力、正常眼位男性青年的立体视锐度。观察组采用GFRDSS检测方法;对照组采用Distance Randot（DR）检测方法。结果GFRDSS可以检测视差范围40″~800″的5 m远距离立体视锐度。100~300 cd/m2照度下,与对照组[41.3%(95/230)]比较,观察组有83.91%（193/230)受检者立体视锐度达到正常水平（40″~60″）,差异有统计学意义(Z=-9.569,P<0.05)。350~600 cd/m2照度下,与对照组[75.22%(173/230)]比较,观察组有84.78%（195/230)受检者立体视锐度达到60″,差异有统计学意义(Z=-4.048,P<0.05)。结论采用人眼追踪技术GFRDSS可以实现5 m距离无辅助立体视锐度检测,正常人群测试结果显示GFRDSS敏感性高于DR,且可以有效避免环境照度对检测的干扰。     

Medical end-of-life decisions: Does its use differ in vulnerable patient groups? A systematic review and meta-analysis

Medical end-of-life decisions, defined as end-of-life practices with a potential or certain life-shortening effect, precede almost 50% of deaths in Western countries, and receive ample medical-ethical attention. This systematic review aims to detect whether there are differences in the prevalence of medical end-of-life decisions in 'vulnerable' patient groups. In 2009, five major databases were scrutinized for publications containing original data on the prevalence of euthanasia/physician-assisted suicide, life-ending without explicit patient request, intensified symptom alleviation, non-treatment decisions and palliative sedation by social factors (eg age, gender and SES). Heterogeneous findings were pooled using a random effects model. We identified 6377 papers of which 51 papers were selected, involving over 1.09 million patients. Most publications reported the prevalence of non-treatment decisions. The most studied social factors were age and gender. Among patients older than eighty years, non-treatment decisions occurred more frequently compared with younger patients, while intensified symptom alleviation, palliative sedation, euthanasia/physician-assisted suicide and life-ending without explicit request were practiced less often. Similar patterns of association, although less strong, were found for female patients compared with males and those with lower levels of education versus more highly-educated patients. We conclude that the administration of medication with a potential or certain life-shortening effect seemed generally to be practiced less often among the elderly, females and less well-educated patients compared with younger, male or more educated patients, while decisions that include the withdrawal or withholding of treatments seem to be more common in these groups. Further studies should focus on investigating whether these differences reflect less than optimal end-of-life care for specific patient groups.     

随机系数发展模型与协方差模式模型在纵向数据分析中的应用及R实现

目的 应用随机系数发展模型与协方差模式模型分析社区卫生服务中心纵向数据,探讨纵向数据分析的问题,为社区随访数据处理提供科学方法.方法 使用R软件对社区卫生服务中心糖尿病重复测量数据分别拟合随机系数发展模型,协方差模式模型以及传统线性回归模型,并比较3种模型的分析结果.结果 随机系数发展模型和协方差模式模型的分析结果与传统线性回归模型不同,2模型较传统线性回归更多的考虑了数据的变异来源.随机系数发展模型与协方差模式模型变量的估计系数结果相近,2者在固定效应的估计上区别往往不是很大,2模型相比,信息标准统计相差也不大.随机系数发展模型倾向于解释组间随机效应,协方差模式模型更关注组内观测之间的联系.R软件nlme package相比于SAS proc mixed,其相应的结果比较与可视化的函数使用更为灵活方便,同时GLS函数提供更多的组内方差协方差模式以供选择.结论 随机系数发展模型与协方差模式模型都能较好的处理重复观测数据组内相关性的问题.2者处理组内相关性的出发点不同,如果强调组内观测之间的联系性,则选择协方差模式模型.相反,如果更关注组间的异质性,强调组间的随机效应,则选择随机系数发展模型.R软件nlme是比较完善的处理混合结构数据的分析包.     

Pathways to forest wealth in Nepal

ABSTRACT

Nepal is one of the leading countries embracing community forestry with about 45% of households being members of community forest user groups. However, there has been a failure to deliver the full potential of forest wealth because of a lack of proper silvicultural management, a constraining policy environment and a complex socio-institutional context. Meanwhile, mid-hill agriculture has not kept pace with the changing economy and out-migration. Food insecurity is rife in a landscape of under-utilised forests and under-utilised land. Australian development assistance between 1978 and 2006 supported the establishment of 21 000 ha of community forests and significant contributions to community forest institutions. In the light of the under-performance of this sector, the Australian Centre for International Agricultural Research established the project Enhancing Food Security and Livelihoods through Agroforestry and Community Forestry in Nepal, locally known as EnLiFT, which ran from 2013 to 2018. This paper aims to explain Australia’s contribution to Nepal’s forestry, with a focus on more recent achievements supporting pathways to realise the potential wealth in Nepal’s forests. It begins with an outline of the early Australian support and origins of community forestry in Nepal via the Nepal–Australia Forestry Project, and then the current status of community forestry. It then describes the research process of the EnLiFT project starting with new conceptual models and methods such as: (1) the Pathways Approach to link forest and food security; (2) the EnLiFT Bioeconomic Model of the Farm-Forest Interface; (3) the Silvo-Institutional Model for Scientific Forest Management; (4) Active and Equitable Forest Management; (5) Rapid Silvicultural Appraisal; (6) the Strategic and Inclusive Planning process and (7) EnLiFT Policy Labs. We also highlight many significant development impacts of EnLiFT. The demonstration and training of silvicultural methods released considerable forest wealth into the community. This occurred around the time of the 2015 earthquake when timber was in need for reconstruction. It was also associated with the re-vitalisation of a defunct sawmill by facilitation of community-private partnership. It was responsible for turning the public debate from resistance to acceptance of scientific forest management. It also developed inclusive planning processes for the revision of operational plans of community forests. On privately owned land, EnLiFT demonstrated: (1) the possibility for marked and rapid changes in livelihoods from relatively simple agroforestry interventions based on horticultural commodities and tree fodders; (2) an even greater potential for livelihood enhancement through private forestry and (3) articulated the current institutional and regulatory constraints on sale of trees from private land. We conclude by highlighting the contributions of EnLiFT in policy debate which led to policy outcomes that further the improvement of community forestry, agroforestry and bringing under-utilised land back into productive use.