基于RFID的围术期术后康复行为检测

MINING METHOD OF VERIFYING ASSOCIATION RULES BASED ON DIFFERENTIAL PRIVACY

摘要:

针对术后患者康复结果是否达到出院标准的评判具有主观性，即由医生根据自身经验判断患者是否达到出院条件，为此引入RFID技术，提出基于RFID的术后患者行为活动挖掘方法。由于患者在围术期过程中由RFID获得的活动路径信息具有高冗余、高维度、高无用等特性，因此在挖掘之前，首先计算事物数据库中各属性的条件信息熵进行属性约减；为提高挖掘效率，减少无用挖掘过程，在传统的FP-growth基础上进行验证式挖掘，即医生输入想要挖掘的关联规则，并根据挖掘结果检测患者术后的活动行为是否达到标准。实验结果表明该算法的运行效率与挖掘质量较传统的FP-Growth算法有着较为显著的提高，运行内存与CPU也有较大幅度的降低，具有一定的临床可行性和有效性。

Abstract:

The evaluation of whether the postoperative rehabilitation results of patients meet the discharge standards is subjective, that is, the doctors determine whether the patients have reached the discharge conditions based on their own experience. To this end, RFID technology is introduced and a method for mining postoperative patient behavior activities is proposed. Because the patient's active path information obtained by RFID during the perioperative period has characteristics such as high redundancy, high dimensions, and high uselessness, before mining, first calculate the condition information entropy of each attribute in the transaction database to reduce the attributes; To improve mining efficiency and reduce useless mining, verify mining is performed on the basis of traditional FP-growth, that is, doctors input association rules that they want to mine, and use the mining results to check whether the patient's postoperative activity behavior meets the standard. The experimental results show that the operation efficiency and mining quality of the algorithm are significantly improved compared with the traditional FP-Growth algorithm, and the operating memory and CPU are also greatly reduced, which has certain clinical feasibility and effectiveness.