行動擴增實境對高齡者用藥績效影響之研究

Abstract

隨著年齡增長所導致所的疾病發生，高齡者將面對越來越複雜的用藥情況。因年齡而產生的認知退化將可能使高齡者用藥風險提高，並降低用藥遵從性。本研究旨在探討擴增實境科技輔助對高齡者用藥遵從績效與用藥認知負荷的影響。本研究聚焦於使用分藥藥盒進行分藥管理之任務，並依據資訊處理歷程、感知負荷與認知負荷之相關理論設計分藥輔助擴增實境科技，針對高齡者使用該科技執行分藥任務進行實驗、觀察、問卷調查與訪談。

本研究招募54位55至74歲的高齡受試者，參與使用擴增實境科技進行分藥任務之實驗。該實驗以三種不同介面設計為組間設計因子，包含感知擴增介面(Perceptual augmentation interface, PAI)、反應擴增介面(Response augmentation interface, RAI)，以及處理歷程擴增介面(Processing augmentation interface, PRAI)。在分藥任務上，分別以兩種難易程度為組內設計因子，包含中用藥複雜度與高用藥複雜度。受試者須先接受心智狀態與視力評估，符合正常狀態才能開始實驗。首先受試者須填寫用藥能力與經驗問卷，接著進行分藥與介面操作的練習，完成練習後，便使用其中一種擴增實境介面分別進行中用藥複雜度與高用藥複雜度之分藥任務操作。兩次實驗操作完成後須分別填寫NASA Task Load Index，全部實驗完成後填寫 System Usability Scale與完成隨後訪談(follow-up interview)。

藉由描述性統計、無母數統計分析與質化觀察資料之分析發現，三種介面在用藥遵從績效與認知負荷上有所差異。其中以使用處理歷程擴增介面（PRAI）之用藥遵從績效最好，不僅能確保分藥結果的正確性，即使產生操作性的失誤也多能由系統互動來恢復。另一方面，不同的用藥複雜度會對用藥遵從績效與認知負荷產生差異。高用藥複雜度的分藥任務出現分藥結果錯誤的情況，操作失誤的情況也較多。在任務時間與任務負荷上也顯著高於中用藥複雜度的分藥任務。然而這些差異並不完全達統計上顯著。

為促進受試者能閱讀並理解用藥指示，進而整合過往經驗以建立系統與任務操作的基模，首要設計重點為感知輔助，此設計可藉由摘要（abstraction）與外生提示（exogenous cues）原則來達成。然而為確保分藥結果的正確性與降低任務過程中的操作性失誤，必須應用提示（notification）、轉換（transition）、內生提示（endogenous cues）與平行搜尋（parallel visual searching）的設計來提供完整的認知輔助設計，以讓受試者執行合適的回應行動（response execution）。另外也需要依據分藥任務的複雜度提供合適的輔助設計。在執行中用藥複雜度的任務時，雖然注意力容易集中，但卻也容易因為快速的學習經驗而疏忽當前任務狀況的評估，而產生操作錯誤，所以需要非持續但即時性的操作行為提示以減少操作性的失誤。而在執行高用藥複雜度的任務時，則能較謹慎評估任務現況與步驟，但也因此注意力容易分散而導致操作失誤。因此在高用藥複雜度任務的認知輔助上需要設計持續性的注意力分配引導機制，以及提供操作執行情況之回饋。

本研究之研究結果驗證分藥輔助擴增實境科技的不同設計以及用藥複雜度對於用藥遵從績效與認知負荷之影響。此研究結果不僅為高齡用藥輔助科技提供設計原則之建議，也補充資訊處理歷程、感知負荷理論與認知負荷理論之實證結果與理論應用之參考。

The number of chronic conditions tends to raise with age, which results in an increased number of multiple-drug regimens for older adults. Age-related perceptual and cognitive impairments affect elderly patients’ unintentional non-adherence through the management of adherence, which raises the risk of health issues. The purpose of this study is to explore how the AR technology, which was designed in line with perceptual and cognitive theories affects the medication adherence performance and the cognitive load of allocating prescribed pills into pill dispensers and the design requirements in different medication regimen complexities.

This study conducted quasi-experiment, task observations, surveys, and interviews to collect the data. Fifty-four participants aged 55 to 74 years were recruited. Two one-factorial experiments are conducted. The first one is a between-subject design to explore the difference of medication adherence performance and cognitive load with using the treatments of perceptual augmentation interface (PAI), response augmentation interface (RAI), and processing augmentation interface (PRAI). The second one is a within-subject design to explore the difference of medication adherence performance and cognitive load when performing the medium and high regimen complexity tasks respectively. The participants were first required to take the evaluations of mental state and visual acuity. Second, they were asked to complete the questionnaires of competencies of medication use and medication experience and practice the using of the systems. Third, they were assigned to use one kind of treatment to perform the medication allocating tasks in both medium and high regimen complexity trails respectively. The NASA Task Load Index (NASA-TLX) was used to measure their task loads after preforming the medium and high complexity trails respectively. Finally, the participants were required to complete the System Usability Scales (SUS) and the follow-up interviews.

According to descriptive statistics, nonparametric statistics, and qualitative analysis, the results show the obvious differences in medication adherence performance and cognitive load with using the PAI, the RAI, and the PRAI. Using PRAI supports the greatest medication adherence performance and the lowest cognitive load of medication allocation tasks between the three treatments. Meanwhile, the differences in medication adherence performance and cognitive load showed between performing the medium and high regimen complexity trails as well. The errors, task time, and task load increase when the medication regimen complexity is high.

Different design principles are required to overcome the obstacles of using the pill dispensers for managing medication adherence. Reducing the perceptual load is the first importance so that the appropriate design principles are abstraction exogenous cues. In order to ensure the correct medication allocating, both perceptual and response supports are required. That is, applying design principles of notification, transition, endogenous cue, and parallel visual searching are demanded. In terms of different medication regimen complexity tasks, the results reflect the different design requirements for medication allocation tasks with medium and high medication regimen complexities respectively. The support designs of attention direction and correctness feedback are needed through the complete task process when performing the medication allocation task with high medication regimen complexity. In contrast, when performing the medium one, just a real-time feedback design for ensuring updated plan executions is satisfied.

This study proposed the design principles of AR technology for supporting medication allocation tasks. In addition, this study suggested the recommendations of applying the information processing model, the perceptual load theory, and the cognitive load theory as theoretical bases for design.