ABSTRACT
A key challenge in closed-loop chronic biomedical systems is the ability to detect complex physiological states from patient signals within a constrained power budget. Data-driven machine-learning techniques are major enablers for the modeling and interpretation of such states. Their computational energy, however, scales with the complexity of the required models. In this paper, we propose a low-energy, biomedical computation platform optimized through the use of an accelerator for data-driven classification. The accelerator retains selective flexibility through hardware reconfiguration and exploits voltage scaling and parallelism to operate at a sub-threshold minimum-energy point. Using cardiac arrhythmia detection algorithms with patient data from the MIT-BIH database, classification is achieved in 2.96 μJ (at Vdd = 0.4 V), over four orders of magnitude smaller than that on a low-power general-purpose processor. The energy of feature extraction is 148 μJ while retaining flexibility for a range of possible biomarkers.
- I. S. Abu-Khater, A. Bellaouar, and M. I. Elmasry. Circuit techniques for CMOS low-power high-performance multipliers. IEEE J. Solid-State Circuits, 31(10):1535--1546, Oct. 1996.Google ScholarCross Ref
- A. L. Benabid. Deep brain stimulation for Parkinson's disease. Current Op. in Neurobiology, 13:696--706, Dec. 03.Google ScholarCross Ref
- A. Csavoy, G. Molnar, and T. Denison. Creating support circuits for the nervous system: Considerations for brain-machine interfacing. In Proc. Int. Symp. VLSI Circuits, pages 4--7, Jun. 2009.Google Scholar
- P. de Chazal, M. O'Dwyer, and R. B. Reilly. Automatic classification of heartbeats using ECG morphology and heartbeat interval features. IEEE Trans. Biomedical Engineering, 51(7):1196--1206, Jul. 2004.Google ScholarCross Ref
- E. Dishman. Inventing wellness systems for aging in place. IEEE Computer, 37(5):34--41, 2004. Google ScholarDigital Library
- D. Hau and E. Coiera. Learning qualitative models from physiological signals. In Proc. AAAI Symp. Artificial Intelligence in Medicine, pages 67--71, 1994.Google Scholar
- A. S. Jaffe, L. Babuin, and F. S. Apple. Biomarkers in acute cardiac disease: The present and the future. J. American College of Cardiology, 48:1--11, 2006.Google ScholarCross Ref
- T. Jaochims. SVM-Light, support vector machine. http://svmlight/jaochims.org.Google Scholar
- F. M. Khan, M. G. Arnold, and W. M. Pottenger. Hardware-based support vector machine classification in logarithmic number systems. In Proc. IEEE Int. Symp. Circuits and Systems, pages 23--26, May 2005.Google ScholarCross Ref
- M. A. Lebedev and M. A. L. Nicolelis. Brain-machine interfaces: Past, present and future. Elsevier Trends in Neurosciences, 29(9):536--546, 2006.Google ScholarCross Ref
- G. Meyfroidt, F. Guiza, J. Ramon, and M. Bruynooghe. Machine learning techniques to examine large patient databases. Best Practice & Research Clinical Anaesthesiology, 23(1):127--143, Mar. 2009.Google ScholarCross Ref
- Physionet. MIT-BIH Physionet database. http://www.physionet.org/physiobank/database.Google Scholar
- S. Cadambi et al. A massively parallel FPGA-based coprocessor for support vector machines. In Proc. Int. Symp. Field Programmable Custom Computing Machines, pages 115--122, Apr. 2009. Google ScholarDigital Library
- A. Shoeb, B. Bourgeois, S. T. Treves, S. C. Schachter, and J. Guttag. Impact of patient-specificity on seizure onset detection performance. In Proc. Int. Conf. IEEE EMBS, pages 4110--4114, Aug. 2007.Google ScholarCross Ref
- A. Shoeb, D. Carlson, E. Panken, and T. Denison. A micropower support vector machine based seizure detection architecture for embedded medical devices. In Proc. IEEE Int. Conf. EMBS, pages 4202--4205, 2005.Google Scholar
- A. Shoeb and J. Guttag. Application of machine learning to seizure detection. In Proc. Conf. Machine Learning, Jun. 2010.Google Scholar
- A. H. Shoeb. Application of Machine Learning to Epileptic Seizure Onset Detection and Treatment. Electrical and Medical Engineering, Massachusetts Institute of Technology, Boson, Massachusetts, Sep. 2009.Google Scholar
- V. Sze and A. P. Chandrakasan. A 0.4-V UWB baseband processor. In Proc. IEEE Int. Symp. Low Power Electronics and Design, pages 262--267, Aug. 2007. Google ScholarDigital Library
- Tensilica Inc. The Xtensa processor. http://www.tensilica.com.Google Scholar
- E. D. Ubeyli. ECG beats classification using multiclass support vector machines with error correcting output codes. DSP, 17(3):675--684, May 2007. Google ScholarDigital Library
- N. Verma, A. Shoeb, J. Guttag, and A. Chandrakasan. A micro-power EEG acquisition SoC with integrated seizure detection processor for continuous patient monitoring. In Proc. Symp. VLSI Circuits, pages 62--63, Jun. 2009.Google Scholar
- S. A. Vitale, P. W. Wyatt, N. Checka, J. Kedzierski, and C. L. Keast. FD-SOI process technology for subthreshold-operation ultralow-power electronics. Proc. IEEE, 98(2):333--342, Feb. 2010.Google ScholarCross Ref
- A. Wang and A. P. Chandrakasan. A 180-mV subthreshold FFT processor using a minimum energy design methodology. J. Solid-State Circuits, 40(1):310--319, Jan. 2005.Google ScholarCross Ref
Index Terms
- A low-energy computation platform for data-driven biomedical monitoring algorithms
Recommendations
Hybrid classification engine for cardiac arrhythmia cloud service in elderly healthcare management
The self-regulation ability of the elderly is largely degenerated with the age increases, and the elderly often expose to great potential hazards of heart disorders. In practice, the electrocardiography (ECG) is one of the well-known non-invasive ...
Design and implementation of a personal health monitoring system with an effective SVM-based PVC detection algorithm in cardiology
SAC '14: Proceedings of the 29th Annual ACM Symposium on Applied ComputingIn this paper, we present a bio-health monitoring system prototype specialized in capturing the Premature Ventricular Contraction (PVC) event, one of the major cardiac disorder events. The proposed bio-health system comprises three parts: (1) the ...
An improved cardiac arrhythmia classification using stationary wavelet transform decomposed short duration QRS segment and Bi-LSTM network
AbstractArrhythmia is a kind of cardiac conduction disorder those result in irregular heartbeats. The electrocardiograph (ECG) signal may identify conduction system abnormalities. However, its visual analysis is challenging and time-consuming. An ...
Comments