A dual-channel single-mode-multi-mode-single-mode (SMS) fiber optic sensor encapsulated by polydimethylsiloxane (PDMS) was proposed for the first time, for the simultaneous monitoring of the brachial and radial arteries for accurate blood pressure prediction. With the help of the machine learning algorithm Support Vector Regression (SVR), the SMS fiber sensor can continuously and accurately monitor the systolic and diastolic blood pressure. Commercial sphygmomanometers are used to calibrate the accuracy of blood pressure measurement. Compared with the single-channel system, this system can extract more pulse wave features for blood pressure prediction, such as radial artery transit time (RPTT), brachial artery transit time (BPTT), and the transit time difference between the radial artery and the brachial artery (DBRPTT). The results show that the performance of dual-channel blood pressure monitoring is more accurate than that of single-channel blood pressure monitoring in terms of the absolute value of the correlation coefficient (R) and the average value of the difference between SBP and DBP. In addition, both the single-channel and dual-channel blood pressure monitoring are in line with the Association for the Advancement of Medical Devices (AAMI), but the average deviation (DM, 0.06 mmHg) and standard deviation (SD, 1.54 mmHg) of dual-channel blood pressure monitoring are more accurate. The blood pressure monitoring system has the characteristics of low cost, high sensitivity, non-invasive and capability for remote real time monitoring, which can provide effective solution for intelligent health monitoring in the era of artificial intelligence in the future.