Research Project Title: Machine-to-Machine Application of Telehealth Using Automatic Electrocardiographic Analysis System by Digital Signal Processing Methods 1. Team member Principal Investigators: Assoc Prof. Yi-Lwun Ho Co-PI: Assist Prof. Jian-Jiun Ding; Assist Prof. Yen-Hung Lin; Dr. Ying-Hsien Chen 2. Discussion with champions Time: 4/26 10:00-11:00 a. discuss the current work: ECG analysis (the algorism for differentiation of normal and abnormal ECG including APC, VPC, atrial fibrillation) b. Assess different criteria which differentiate atrial fibrillation fgrom ventricular fibrillation (Cited from reference: Liu CS, Tseng WK, Lee JK, Hsiao TC, Lin CW. The differential method of phase space matrix for AF/VF discrimination application. Medical Engineering & Pysics 2010;32: 444–453) 3. Progress between last month and this month (the algorithm for extracting the feature points and differentiation of normal and abnormal ECG including APC, VPC, atrial fibrillation) a. Summary of key findings and innovation, and/or The results of the previous month: For the database acquired from NTHU Feature Points R Points Q and S Points P and T Points Sensitivity 98.06% 97.09% 93.93% FP rate 0% 0% 2.60% Computation time: 0.000264 second for a one-second length ECG signal (by Matlab). The results of this month: For the database acquired from NTHU Feature Points R Points Q and S Points P and T Points True Number 206 412 346 Detected (TP) 206 412 343 False Detected (FP) 0 0 2 Sensitivity 100% 100% 99.13% FP rate 0% 0% 0.58% Computation time: 0.000292 second for a one-second length ECG signal (by Matlab). If only the R points should be detected, the computation time is 0.000130 second for a one-second length ECG signal (by Matlab). For MIT/BIH Arrhythmia Database (used worldwide) True number of R points: 110159 Sensitivity: 99.82% Detected R Points: 109964 a.We have done the paper survey and found that the best result in the open literature is 99.80%. The computation time of our method is about 0.3 second (by Matlab). However, the computation time of the existing method is 3 second. We develop an advanced method for base line extraction. The accuracy of base line extraction significantly affects the performance of ECG feature point extraction. We find that, with the proposed base line extraction method, the P, Q, R, S, and T points can be detected very accurately. b. Negative results and their consequences, and/or There are some cases that the algorithm does not work perfectly, such as the cases where the interval of R points is less than 1/3 second and the baseline varies too fast. c. Cross-project synergy We propose the idea that the ECG sensor can be implemented in the steering wheel of the car. This can monitor the condition of the driver. If the driver feels sleepy, the RR interval becomes larger. If the driver is drunk, the RR interval becomes smaller. In these cases the system in the car will inform the driver not to derive again. This is helpful for improving the safety of the driver. 4. Brief plan for the next month (1) We will calculate the number of instructions for ECG signal analysis and try to minimize it. (2) We will further improve the accuracy of the ECG feature point extraction algorithm. (3) We will try to design the algorithm to analyze the arrhythmia problems, including the cases of atrial fibrillation (AF), ventricular fibrillation (VF), and the atrioventricular block (AV block). a. Updated view of planned milestones, deliverables, and success criteria Date Milestone Deliverables Technical Success Criteria/ Objectives Confirm the research direction Study and environment Presentation of the study setting. Make solid power estimation to re-address the first risk of this project 2012Q4 Finish the first version of Technical report of the system A technique report the ECG interpretation analysis result with detailed algorithm. Software for the interpretation description Define some joint develop items Clinical trial IRB with other subproject. application 2012Q5 Develop a very accurate and Technical report of the system For MIT/BIH very fast algorithm for ECG analysis result Arrhythmia Database, feature point extraction. Software for the interpretation sensitivity = 99.80% and Define some joint develop items computation time for with other subproject. each 30 minute length Discuss the arrhythmia problem ECG data is 0.3 second. 5. Research byproducts 2012Q2 Since the ECG feature point extraction algorithm we develop has both higher accuracy and very less computation time than the existing methods, we plan to write it as a journal paper in the summer.