手控標靶注射與輸注(TCMII) --- propofol藥物動力指引系統的 實作性與可行性的初步評估

Abstract

目的：到目前為止用群體藥物動力學與藥效學模型控制靜脈麻醉的深度，只能使用TCI程控幫浦。不過在麻醉工作中仍常需以手工給藥或用傳統幫浦以 rate control infusion的方法注射藥劑，因此我們嘗試以藥物動力模型開發手控靜脈注射的TCMII guider 以協助這方面的工作。由於手控給藥需要考慮長時間的藥物累積效應，需要設計比TCI更加有效率的演算法。為了協助人員在手工給藥或設定幫浦流速之前，判斷給藥對作用部位濃度(Ce)的影響，此系統首先要能夠快速計算Ce的走勢(Ce trend)，以協助人員判斷與選擇。此外為了降低人員的工作負荷，此系統要能夠即時計算出最佳的注射劑量(OD)與最佳輸注速率(OR)將Ce維持於目標區(Cet range)之內。 研究內容與方法：基於Schnider的propofol群體藥物動力學模型與藥效學模型，在Java ME平台上實作此系統。推算Ce走勢的方法有三，解析法線性搜尋(ALS)、數值法線性搜尋(NLS)以及實根分離法(RRI)，以t-test比較三者所需時間以求其最快者並得知能推算的Ce trend有多長久。尋找OD與OR的方法有二，線性搜尋法與二分搜尋法，也以t-test比較二者所需的時間，以求其最快者。再以各種大小的Cet range與劑量及流速單元量模擬兩小時以上的target control，以了解使用TCMII時所需的人員工作量以初步評估此系統的可行性。 結果：時間範圍愈大，計算Ce trend所需的時間愈久。當預測的時間範圍長達3小時以上時，RRI仍能在0.2秒內推算Ce走勢，而ALS與NLS則否。二分搜尋法能在1秒內找出OD與OR，線性搜尋法則僅能找出OD。如欲將propofol的Ce控制在目標的90-110%範圍內，只要每2分鐘左右給予一次最佳注射劑量即可。對於同範圍，使用定速幫浦注射，則在穩態之前僅需調整流速為最佳輸注速率3次即可。 結論：TCMII能即時提供Ce trend以及OD與OR，以協助麻醉醫師以最少的人員工作量將Ce控制於Cet range中。它是目前唯一能將藥物動力模型應用於手控處置靜脈注射以施行target control的方法。

Objective: Population pharmacokinetic/pharmacodynamic (PopPK/PD) models based intra- venous anesthesia need a program-controlled infusion pump so far. But bolus injection or traditional rate control infusion pump is still applied on anesthesia work and that is why a target- controlled manual injection/infusion (TCMII) guider is needed and was developed in this study. Since manual administration has a long term drug accumulation effect, TCMII need a more efficient algorithm to handle effect site concentration (Ce) trend. A user has to read Ce trend before deciding an injection dose and an infusion rate, so this guider has to plot Ce trend timely. Moreover, in order to minimize the human workload, the guider must calculate an optimal injection dose (OD) and an optimal infusion rate (OR) every second to help the user to keep Ce in Cet range. Methods and Material: Based on Schnider's PopPK/PD models of propofol, a TCMII guider was implemented on Java ME platform. T-tests was applied to compare the speed to plot Ce trend by analytical linear search (ALS), numerical linear search (NLS) and real root isolation (RRI) methods to find the most efficient one and how long the Ce trend can be. The times needed to find out ODs and ORs by linear search (LS) and bisection method (BM) were compared with t-tests to find out which one is faster. The frequencies of using ODs and ORs to keep Ce in Cet ranges for two or more hours were simulated in order to evaluate the workload of manual target control and the feasibility of this guider. Result: To plot Ce trend on a larger time domain took more time. If time domain size was more than 3 hours, RRI can plot Ce trend in 0.2 second but ALS and NLS can not. OD can be searched by BM and by LS in 1 second. But OR can be searched in 1 second only by BM. To keep Ce in the range of 90-110% Cet needed to inject an OD every 2 minutes. To keep Ce in the same range for more than 2 hours needed to set the infusion rate as OR for 3 times. Conclusion: TCMII can offer users with Ce trend and OD and OR on demain. It helps to keep Ce in a Cet range with minimal and reasonable workload. As we know, it is the only method to apply PK and PD models on target-controlled manual intravenous administrations.