探讨大数据技术在老年人衰弱综合征粗筛中的应用价值。

选取2017年7月至8月新疆维吾尔自治区人民医院收治的118例老年患者作为研究对象，均完成Fried衰弱表型评估。再利用大数据技术（Hadoop）收集、整理患者病例资料中的各项指标，以累计缺陷衰弱指数法（frailty index, FI）为基础，并结合其他评估工具建立评价指标体系，模拟计算FI值。通过不断调整指标项目，确定与Fried衰弱表型评估在相关性和一致性等方面契合的指标集合。

Fried衰弱表型分期：无衰弱28例，衰弱前期68例，衰弱22例；大数据技术自动评估粗筛：无衰弱8例，衰弱110例（不区分衰弱前期与衰弱）。利用大数据技术对本组老年患者计算的模拟FI平均值为0.58±0.20。FI值的增加与Fried分期呈正相关（r=0.607，P<0.01）。以Fried衰弱表型分期结果为标准，将本组老年人分为无衰弱和衰弱（衰弱前期+衰弱）两组，代入最终确定的FI值绘制ROC曲线，得到曲线下面积（AUC）为0.912（95%CI：0.861-0.963，P<0.01），得出FI临界值为0.28，其灵敏度、特异度及Youden指数分别为99%、75%、0.74。将FI=0.28引入大数据技术评估模型，其对老年人群的衰弱筛查结果与Fried衰弱表型分期结果的一致性Kappa值为0.169（P<0.01）。

采用大数据技术建立模型对入院老年人进行衰弱粗筛是可行的，其筛查结果与Fried衰弱表型分期结果具有一定的一致性，可以为临床诊断提供参考。

To explore the application value of big data technology in coarse screening of elderly debilitating syndrome.

A total of 118 elderly patients in the People's Hospital of Xinjiang Uygur Autonomous Region from July to August 2017 were selected as subjects. They were all completed with Fried's weak phenotypic assessment. Big data technology (Hadoop) were subsequently used to collect and sort out the indicators. Based on the cumulative defect index (FI), and combine with other evaluation tools, an evaluation index system to simulate the FI value were established. By continuously adjusting the indicator items, the set of indicators that fit the relevance and consistency of the Fried weak phenotypic assessment were determined.

Fried fragility phenotypic staging: 28 cases without fragility, 68 cases with weak pre- fragility, and 22 cases with fragility; Big data technology automatically evaluated coarse screening: 8 cases without fragility and 110 cases with fragility (not distinguishing between pre-fragility and fragility). The mean simulated FI calculated for elderly patients using big data technology was 0.58 ± 0.20. The increase in FI value was positively correlated with Fried stage (r=0.607, P<0.01). Based on the Fried fragility phenotypic staging results, the elderly were divided into two groups: no fragility and fragility (pre- fragility + fragility). The ROC curve was drawn into the final FI value, and the area under the curve (AUC) was 0.912 ( 95% CI: 0.861-0.963, P <0.01), the FI threshold was 0.28, and the sensitivity, specificity and Youden index were 99%, 75%, and 0.74, respectively. FI=0.28 was introduced into the big data technology assessment method, and the Kappa value of the weak screening test results of the elderly group and the Fried weak phenotypic staging results was 0.169 (P<0.01).

It is feasible to use the big data technology to establish a model for the fragility and coarse screening of the elderly in the hospital. The screening results have a certain consistency with the Fried fragility phenotypic staging results, which can provide reference for clinical diagnosis.