Progress Seminar 2017.1.26 이준녕.

Presentation on theme: "Progress Seminar 2017.1.26 이준녕."— Presentation transcript:

1 Progress Seminar 이준녕

2 연구 진행 상황 보고서 재활의학과 응급의학과 Bepatch/CPF 2주전 계획 연구 결과 문제점 및 대책 목표 및 계획
ABT 논문 수정 WPR 논문 영문 교정 및 PM&R 투고 워킹 목업 3대 제작 사용자 테스트 센서 성능 검사 HW ver 4 임상시험 >20명 Software development EMBS 1장 연구 결과 ~ 임상시험 10명 (Static exercise &Noisy data) Real-time signal processing algorithm 문제점 및 대책 목표 및 계획 임상시험

3 >20명 임상시험 (데이터 수집) 및 software 개발
SOCOM 과제 진행 Real-time software 개발 >20명 임상시험 (데이터 수집) 및 software 개발 1/13 1/20 1/27 2/3 2/10 2/17 2/24 2/28 1차 PCB 2차 PCB 발주 2차 PCB 3차 PCB 발주 3차 PCB Sample & hold Casing & Electrode 임상시험: rest, static exercise, dynamic exercise (walking or cycling)

4 Real-time software (혈압 추정) 개발 >10 임상시험 dynamic exercise
SOCOM 과제 진행 Real-time software (peak detection) 개발 Real-time software (혈압 추정) 개발 >10명 임상시험 (데이터 수집) 및 software 개발 >10 임상시험 dynamic exercise 1/13 1/20 1/27 2/3 2/10 2/17 2/24 2/28 1차 PCB 2차 PCB 발주 2차 PCB 3차 PCB 발주 3차 PCB Sample & hold Casing & Electrode 임상시험: rest, static exercise, dynamic exercise (walking or cycling)

5 Bepatch: real time data acquisition
t<=14 seconds ... …… t>14 seconds …… ... 2sec 2sec

6 Bepatch: real time data processing 1
10x250 resample buffer 1 250 ECG R-peak detection (10 recent peaks) 14 sec ECG buffer P&T Signal extraction Resampling to length 250 Ensemble average 8 signals Remove 2 signals with smallest & largest variance 1x250 ensemble average buffer Reconstructed ensemble average 1 250 Resample to 1 n PAT & PEP original length detection

7 Bepatch: real time data processing 2
Ensemble average Reconstructed average PAT Ensemble average Outlier smallest variance Outlier greatest variance

8 Bepatch: real time data processing 2
Ensemble average Reconstructed average PAT find max of derivative PEP find min of derivative Find max of signal before min of derivative

9 Bepatch: real time data processing 3
2sec 2sec Search range: Previous min(derivative) ± fs/50 min(derivative) – fs/25 If min(derivative)=peak Search range: Fs/50 ~ fs/8 min(derivative) – fs/25 If min(derivative)≠peak Search range: none Current peak (i.e. PEP) = Previous peak If min(derivative)≠peak If min (derivative) ≠ peak + α conditions If min (derivative) ≠ peak + β conditions

10 Bepatch: real time data processing results

11 Bepatch: real time data processing results (2017-1-24-KJS2)

12 Bepatch: real time data processing results

13 Room for improvement Large error caused by gradual change in search range. Likely due to distorted PPG waveform.

14 Room for improvement Sharp change in BP with slow changes in HR and PEP (usually occurring at recovery phase), adjusting for this change will significantly improve BP estimation. Errors in R-peak detection will lead to errors in PEP and PAT Rising BP, falling PEP, and fluctuating HR during exercise. This relationship is not constant even within a single subject. How to adjust?

15 Bepatch 계획 TECHNICAL: Add template matching to remove low SNR signals prior to ensemble averaging Remove outlier PEP, PAT, PTT values prior to BP estimation Add physiological parameters to BP estimation Add Kalman filtering (maybe) to improve outlier removal Analyze group data to see if new user calibration is possible Investigate calibration frequency to meet project goals (±10mmHg) MILESTONES: Finish peak detection by 1/31 -> send to ADIT EMBS 1 page paper for mini-symposium 1/31 임상시험 by 2/4 -> data analysis by 2/7 BP estimation software by 2/1 -> send to ADIT HW final version with casing by 2/10 -> test by 2/17 -> send to ADIT