發展肝臟移植病患術後狀況早期預警系統

Abstract

器官移植加護病房(Transplant Intensive Care Unit, TICU)為醫院中維護器官移植病患生命的醫療單位，移植器官的病患其病情常處於不穩定的狀態，必須仰賴臨床醫護人員對其進行密切照護與即時處置，這也造成醫護人員的工作負荷(Workload)居高不下，因此一方面為了使病患獲得更完善的照護，另一方面又能降低臨床醫護人員的工作負荷，本研究進行智能型早期預警系統開發，發展心血管預測模型以及肝臟狀態推測模型，期望能透過模型的導入，提升醫療與照護品質。 對於心血管預測模型的建立，本研究運用時間序列分析(Time series analysis)中的自回歸整合移動平均(Autoregressive Integrated Moving Average, ARIMA)法，進行病患心血管變數數值的預測，並運用平均絕對誤差值(Mean Absolute Error, MAE)與平均絕對誤差百分比值(Mean Absolute Percentage Error, MAPE)作為評估本模型準確度的衡量方式，而評估的結果多屬於極佳的預測；另一方面對於病患肝臟狀態的推測模型，本研究運用層級分析法(Analytic Hierarchy Process, AHP)結合模糊邏輯(Fuzzy logic)，探討病患心血管數值與肝臟狀態之間的關係，並運用平均絕對誤差值與平均絕對誤差百分比值作為評估兩者對應程度的衡量方式，評估的結果屬於優良的預測，代表病患的心血管系統變化在一定程度上會影響術後病患的肝臟狀態。

Transplant Intensive Care Unit (TICU) is a ward where transplant patients receive special care to prolong their lives and get their bodily functions back as usually. In TICU, patients with transplantation often be in danger due to the high unstable status, it takes physicians and nurses much effort to look after patients and take medical treatments more closely and instantly, however, it makes health care workers in a high workload position. Therefore, the main purpose of this study is to make patients with medical care more completely, and on the other hand, to reduce workload that health care workers with. By developing intelligent early warning system with cardiovascular parameters prediction model and liver function reasoning model, we expect to solve the problem mentioned above, moreover, enhance the quality of medical treatments and care. For developing a prediction model of cardiovascular parameters, our study uses Autoregressive Integrated Moving Average (ARIMA) model, a kind of Time series analysis method, to construct a forecasting value of cardiovascular parameters, and uses Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) to evaluate the degree of accuracy with this model, the result of this model is an highly accurate forecasting model. On the other side, for developing a liver function reasoning model, we use Analytic Hierarchy Process (AHP) and Fuzzy logic method to make a deep investigation of the relationship between cardiovascular and liver status, and use MAE and MAPE to evaluate the degree of accuracy with this model, the result of this model is a good reasoning model.