Journal of Digital Policy (디지털정책학회지)

The Society of Digital Policy and Management (한국디지털정책학회)
권호 표지
DOI QR코드

DOI QR Code

The study of blood glucose level prediction using photoplethysmography and machine learning

PPG와 기계학습을 활용한 혈당수치 예측 연구

  • Received : 2022.11.17
  • Accepted : 2022.12.20
  • Published : 2022.12.28
Download PDF
⟨ Previous Next ⟩

Abstract

The paper is a study to develop and verify a blood glucose level prediction model based on biosignals obtained from photoplethysmography (PPG) sensors, ICT technology and data. Blood glucose prediction used the MLP architecture of machine learning. The input layer of the machine learning model consists of 10 input nodes and 5 hidden layers: heart rate, heart rate variability, age, gender, VLF, LF, HF, SDNN, RMSSD, and PNN50. The results of the predictive model are MSE=0.0724, MAE=1.1022 and RMSE=1.0285, and the coefficient of determination (R2) is 0.9985. A blood glucose prediction model using bio-signal data collected from digital devices and machine learning was established and verified. If research to standardize and increase accuracy of machine learning datasets for various digital devices continues, it could be an alternative method for individual blood glucose management.

논문은 광용적맥파(photoplethysmography, PPG) 센서에서 획득한 생체 신호, ICT 기술 및 데이터 기반의 혈당수치 예측 모델을 개발하고 검증하는 연구이다. 혈당 예측은 기계학습의 MLP 아키텍처를 이용하였다. 기계학습 모델의 입력층은 심박수, 심박변이도, 나이, 성별, VLF, LF, HF, SDNN, RMSSD, PNN50의 10개의 입력노드와 은닉층은 5개로 구성된다. 예측모델의 결과는 MSE=0.0724, MAE=1.1022 및 RMSE=1.0285이며, 결정계수(R2)는 0.9985이다. 비채혈방식으로 디지털기기에서 수집한 생체신호 데이터와 기계학습을 활용한 혈당 예측 모델을 수립하고 검증하였다. 일상에 적용하기 위해 다양한 디지털 기기의 기계학습 데이터셋 표준화와 정확성을 높이는 연구가 이어진다면 개인의 혈당 관리에 대안적 방법이 될 수 있을 것이다.

Keywords

References

  1. H. Wu et al. (2022) Diabetes and Its Cardiovascular Complications: Comprehensive Network and Systematic Analyses. Front. Cardiovasc. Med. 9:841928. DOI: 10.3389/fcvm.2022.841928
  2. J. E. Shaw, R. A. Sicree & P. Z. Zimmet. (2010). Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes research and clinical practice, 87(1), 4-14. DOI : 10.1016/j.diabres.2009.10.007
  3. WHO. (2022.9.16.). Diabetes(Online). https://www.who.int/news-room/fact-sheets/detail/diabetes
  4. UK Prospective Diabetes Study (UKPDS). (1991). VIII. Study design, progress and performance. Diabetologia, 34(12), 877-890. https://doi.org/10.1007/BF00400195
  5. S. Genuth et al. (2003). Implications of the United kingdom prospective diabetes study. Diabetes care, 26, S28-S32. DOI : 10.2337/diacare.26.2007.s28
  6. Korea Disease Control and Prevention Agency. (2021). Trends in prevalence of diabetes among Korean adults aged 30 years and over, 2011-2019. Division of Chronic Disease Control, Korea Disease Control and Prevention Agency. Weekly Health and Disease, 14(46).
  7. J. S. Krinsley & J. C. Preiser. (2015). Time in blood glucose range 70 to 140 mg/dl >80% is strongly associated with increased survival in non-diabetic critically ill adults. Critical care (London, England), 19(1), 179. DOI : 10.1186/s13054-015-0908-7
  8. S. Habbu, M. Dale & R. Ghongade. (2019). Estimation of blood glucose by non-invasive method using photoplethysmography. Sadhana, 44(6), 1-14. DOI : 10.1007/s12046-019-1118-9.
  9. Ministry of Food and Drug Safety. (2019.09.23). Blood glucose monitoring systems, self-testing (Online). https://udiportal.mfds.go.kr/brd/view/P03_01?ntceSn=14
  10. CareSensPRO. (2021. 11. 25). User's Manual; Blood Glucose Monitoring System(Online). https://i-sens.com/pdf/ko/2021/PGUAA-0000038-REV1-CS-PRO-M-Manual-KR_Positive-LCD.pdf
  11. S. K. Choi & C. G. Park. (2021). The study of blood glucose level prediction model using ballistocardiogram and artificial intelligence. Journal of Digital Convergence, 19(9), 257-269. DOI : 10.14400/JDC.2021.19.9.257
  12. W. L. Clarke et al. (1987). Evaluating clinical accuracy of systems for self-monitoring of blood glucose. Diabetes care, 10(5), 622-628. DOI : 10.2337/diacare.10.5.622
  13. W. L. Clarke. (2005). The original Clarke Error Grid Analysis (EGA). Diabetes technology & therapeutics, 7(5), 776-779. DOI : 10.1089/dia.2005.7.776