Recursive and Rolling Windows for Medical Time Series Forecasting: A Comparative Study

Medical data accuracy is crucial for providing reliable remote healthcare services for patients. To this end, data accuracy assessment mechanisms must be developed. Prediction algorithms may be used in order to address this issue. This paper extends previous studies by investigating the relevance of different statistical concepts in modeling and predicting the variance instability of medical data. Doing this, we refer to the standard auto-regressive model as a benchmark for prediction purpose and two competing specifications namely the rolling and the recursive windows. We evaluate their forecasting adequacy for medical time series in terms of prediction errors and the Theil Inequality Coefficient. Results show on the one hand that the rolling window concept seems to be an efficient technique for forecasting medical series with instability variances. On the other hand, the recursive window performs better when forecasting medical time series with constant variances.

[1]  Zucchini,et al.  An Introduction to Model Selection. , 2000, Journal of mathematical psychology.

[2]  Sandeep Sharma,et al.  Anomaly Detection in Medical Wireless Sensor Networks using Machine Learning Algorithms , 2015 .

[3]  Yaning Liu,et al.  Anomaly Detection in Medical Wireless Sensor Networks , 2013, J. Comput. Sci. Eng..

[4]  Schahram Dustdar,et al.  Quality of Context: models and applications for context-aware systems in pervasive environments , 2014, The Knowledge Engineering Review.

[5]  Syed Mahfuzul Aziz,et al.  Sensor Anomaly Detection in Wireless Sensor Networks for Healthcare , 2015, Sensors.

[6]  Ionut Cardei,et al.  A Mobile Device Prototype Application for the Detection and Prediction of Node Faults in Wireless Sensor Networks , 2014, ArXiv.

[7]  Joon Lee,et al.  Accessing the public MIMIC-II intensive care relational database for clinical research , 2013, BMC Medical Informatics and Decision Making.

[8]  K. Knight.  Consistency of Akaike's Information Criterion for Infinite Variance Autoregressive Processes , 1989 .

[9]  Duc Khuong Nguyen,et al.  Forecasting the conditional volatility of oil spot and futures prices with structural breaks and long memory models , 2012 .

[10]  Nasser L. Azad,et al.  Vehicle speed prediction via a sliding-window time series analysis and an evolutionary least learning machine: A case study on San Francisco urban roads , 2015 .

[11]  Borko Furht,et al.  Anomaly Detection Scheme for Medical Wireless Sensor Networks , 2013, Handbook of Medical and Healthcare Technologies.

[12]  Jeffrey M. Hausdorff,et al.  Physionet: Components of a New Research Resource for Complex Physiologic Signals". Circu-lation Vol , 2000 .

[13]  Nigel H. Lovell,et al.  Effect of Home Telehealth Data Quality on Decision Support System Performance , 2015, CENTERIS/ProjMAN/HCist.

[14]  Borko Furht,et al.  Sensor fault and patient anomaly detection and classification in medical wireless sensor networks , 2013, 2013 IEEE International Conference on Communications (ICC).

[15]  Osman Hasan,et al.  Formal reliability analysis of Device Interoperability Middleware (DIM) based E-health system using PRISM , 2015, 2015 17th International Conference on E-health Networking, Application & Services (HealthCom).

[16]  Rance Cleaveland,et al.  Formal verification of software-based medical devices considering medical guidelines , 2013, International Journal of Computer Assisted Radiology and Surgery.

[17]  Borko Furht,et al.  Anomaly Detection in Medical Wireless Sensor Networks using SVM and Linear Regression Models , 2014, Int. J. E Health Medical Commun..

[18]  Marc A. Rosen,et al.  Application of sliding window technique for prediction of wind velocity time series , 2014 .