演算法首先利用公開資料進行驗證。HeartCycle 資料庫中，各 case 平均估測 SV_ICG 與 NICCOMO SV 之相關係數為 R = 0.848；Radar Vital Signs Dataset 經 10-beat averaging 後，估測 SV_ICG 與 Task Force Monitor (TFM) SV 之相關係數為 R = 0.663。進一步於 ICU 臨床場域進行量測後，本研究估測 SV_ICG 與 FloTrac SV 於代表性 case 中呈現相似之時間變化趨勢，相關係數可達 R = 0.751。
綜合上述結果，本研究所開發之可攜式 ECG–ICG 同步量測系統可於實際臨床環境中取得可供分析之訊號，且所建立之特徵擷取與 SV 估測方法於不同資料來源下皆能反映血流動力學變化，顯示其具有應用於非侵入式長時間血流動力學監測之潛力。;This study developed a portable system for synchronous electrocardiogram (ECG) and impedance cardiography (ICG) acquisition and evaluated its feasibility for non-invasive hemodynamic monitoring using public datasets and clinical measurements. Hemodynamic parameters such as stroke volume (SV) and cardiac output (CO) are important for assessing cardiac function; however， conventional clinical approaches are often limited by cost， operational requirements， and portability， making long-term monitoring difficult.
The proposed system integrates ECG and ICG acquisition modules， low-noise power management， an STM32U575 microcontroller， SD-card storage， and Bluetooth communication. Synchronization was evaluated using a 10-Hz sinusoidal signal， resulting in an inter-channel time difference of approximately 1.07 ms. ECG R peaks were used as the primary timing reference， while T peaks assisted ICG feature localization. The B， C， and X points were identified from the dZ/dt waveform， and Z_0， LVET， and (dZ/dt)_max were extracted for SV estimation.
The algorithm was validated using two public datasets. In the HeartCycle dataset， the correlation between case-averaged estimated SV_ICG and NICCOMO SV was R = 0.848. In the Radar Vital Signs Dataset， estimated SV_ICG correlated with Task Force Monitor (TFM) SV at R = 0.663 after 10-beat averaging. Clinical evaluation in an intensive care unit (ICU) showed similar temporal trends between estimated SV_ICG and FloTrac SV in a representative case， with R = 0.751.
Overall， the proposed ECG–ICG system successfully acquired synchronized signals in a clinical environment， while the developed feature extraction and SV estimation method showed consistent relationships with reference SV measurements. These results support its potential for non-invasive and long-term hemodynamic monitoring.
