膽結石是一種很常見的疾病,近年來,由於國人的生活水準提高,西化飲食盛行,與過度攝取高熱量、高膽固醇之食物,導致罹患膽結石的個案每年有向上攀升的趨勢,且根據最新的調查中發現,膽結石的發病族群已逐漸年輕化,除了常好發於老年人、女性以及肥胖者之外,也常發生於二十至三十歲之男女性,在國內,每年健保支付的醫療成本更是不容忽視,故膽囊病變一直是醫療研究的重要議題之一。 我國自全民健保開辦以來,醫療成本逐年增加,在龐大的健保醫療支出中,住院費用就占了32%,其中50%以上的住院醫療費用與住院日數長短有密切關係,本研究主要以腹腔鏡膽囊切除術病患為對象,有效的對該病患作早期的住院日預測,建構一個住院日預警評估模式,以健保申報住院天數門檻,作為住院天數分類的評估標準,對醫師、醫院管理、病患本身及家屬是具有相當的意義與重要性。 本研究分別應用決策樹及類神經網路作為探討腹腔鏡膽囊切除術住院日預警評估之工具。研究結果顯示,決策樹訓練正確率為86.88%;測試正確率為87.04%;而類神經網路準確率可達87.81%。總結而言,本研究結果已顯示系統的實際可行性,藉由住院天數預測評估,提供醫師於診斷時之參考,以提升醫院更完善的醫療資源分配以及讓病患有更好的醫療照護與服務品質。
Gallstones is a common disease. Recently, cases of patients suffering from gallstones are increasing every year due to several reasons, such as the raising living standard, prevalence of Western eating habit, and assimilation of high-calorie and high-cholesterol. According to the latest survey, the age bracket of sufferers of gallstones is getting younger gradually. It occurs frequently not only to elder, female, and obesity, but also occurs to men and women between 20-30 years old. Domestically, the medical treatment cost disbursed from the national health insurance can not be ignored and gallbladder disease becomes one of the most important issues in medical research. Cost of medical treatment is raised every year since the national health insurance has been executed. In the huge amount of medical expenses, cost of hospitalisation accounted for 32 percent where over 50 percent of hospitalisation cost is closely related to number of days for hospitalisation . Patients underwent laparoscopic cholecystectomy are targeted in this study. Beforehand forecast for the length of stay are expected to be detected effectively, and model of early warning estimation is intended to be built. To use threshold of length of stay declared by health insurance application as the criterion of classification is respectably significant to the physicians, hospital management, patients themselves and their families. In this study, Decision Tree and Neural Network are used as a instrument to estimate the early warning of length of stay for laparoscopic cholecystectomy surgery. The results show that the accuracy is 86.88% for the Decision Tree training, and 87.04% for the test. The accuracy for Neural Network reaches 87.81%. In conclusion, results of this study have shown the practical feasibility. The forecast for length of stay can be used as the reference resources for physicians while diagnosing. Medical resource allotment would be promoted and better medical attention and service would be given.