「胎兒窘迫」代表胎兒在子宮內缺氧的象徵，這種狀況往往影響胎兒的神經系統，導致胎兒罹患腦性麻痺的機率提升，嚴重甚至會導致胎兒死亡。目前臨床上使用床邊胎兒監測器，將「胎兒心率」與「子宮收縮」訊號顯示於紙張或螢幕上，作為評估胎兒是否缺氧的參考。不過監測器本身體積龐大、不易移動，且價格較為昂貴，孕婦大多必須前往醫院或是婦產科診所才能使用。此外，產檢時醫師大多仰賴目視進行長時間的資料判讀。除了不容易發現細微的訊號差異外，也容易加入主觀結果。本研究利用自行研發的正向力子宮收縮與都卜勒超音波感測技術，結合無線傳輸與智慧分類方法，設計一套穿戴式孕婦胎心率與子宮收縮偵測裝置，並可即時進行訊號分類與警示，將結果傳送至手機顯示。智慧分類方法以臨床三層分類標準為主，先以法則分類系統進行三類判讀，接著利用序列分析結合人工類神經網路(ANN)的方法針對臨床較未知的II類訊號，進一步分類成IIa與IIb訊號，提供臨床更精準的依據來提高胎兒窘迫的偵測準確度。本研究所研發的正向力子宮收縮與都卜勒超音波感測技術，與市售設備均有相當高的相關度。在68位臨床孕婦的分類實驗中，法則分類系統與醫生判讀Kappa統計平均結果為0.72。序列分析結合人工類神經網路的方法，可以再區分臨床較未知的II類訊號切割成IIa與IIb，其發生胎兒窘迫的機率分別為25%和75%，可見本系統的臨床應用潛力。未來可以加入其它臨床生理參數，提高系統預測準確度，期待本研究的成果可以提高臨床照護品質與降低人力需求，並作為未來孕婦精準醫療之用。

Fetal distress is a symbol of fetal hypoxia in the uterus, which might affect the fetal nervous system function, leading to an increased risk of cerebral palsy during fetal growth, and could be serious enough to cause fetal death. At present, bedside fetal monitors which are used clinically to display fetal heart rate (FHR) and uterine contraction (UC) signals on sheets of ordinary paper or monitor screen are recognized as a reference in the assessment of fetal hypoxia. The bedside fetal monitors are pretty heavy, quite difficult to move, and expensive, therefore, most pregnant women must process the monitors only in hospitals or in maternity clinics. The bedside fetal monitors are pretty heavy, quite difficult to move, and expensive, therefore, most pregnant women need to go to the hospitals or maternity clinics examination. Additionally, most doctors mainly rely on visual observation for the interpretation of long-term data during the prenatal examination. In addition to the fact that it was not easy to find subtle signal differences, and was rather prone to make an assessment with their own subjective thinking. This study used a self-developed positive force UC and Doppler ultrasound sensing technology, and combined with wireless transmission and intelligent classification methods to design a wearable maternal fetal heart rate (FHR) and UC detection device. The proposed device could immediately carry out signals classification and warning signs, the results were transmitted and displayed on the phone. The intelligent classification method was based on the clinical three-tier classification and used the method of sequence analysis and artificial neural network (ANN) method to further classified into IIa and IIb based on the unknown clinical signals (type II). The more precise reference would be provided to improve the detection accuracy of fetal distress in clinical trials. The positive force UC and Doppler ultrasonic sensing technology developed by the institute, it had the same high-degree of correlation in comparison with commercially available equipment. In the classification experiment of 68 pregnant women in clinical trial, the average Kappa score of the rule classification system and the doctor's interpretation was 0.72. In combination with sequence analysis and artificial neural network method, it could distinguish the unknown clinical signals (type II) into IIa and IIb which their probability of fetal distress were 25% and 75%, respectively. Hence, it could be seen that the clinical application potential of this system could be expected. The study could further to add other clinical physiological parameters to improve the accuracy of system prediction. It was expected that the results of the present study could improve the quality of clinical care and reduce the requirements for Nurse Staffing, and serve as an optimal medical treatment for pregnant women in the future.