LifeTrack: An Integrated Statistical and Probabilistic Framework for Anomaly Detection and Behavioral Prediction in Wearable Health Data

This paper presented LifeTrack, an integrated statistical and probabilistic framework for anomaly detection and behavioral prediction using heterogeneous wearable-health data. The system connects data preprocessing, Gaussian z-score analysis, Shannon entropy, a Behavioral Consistency Index, behavioral-state classification, Markov prediction, and optional ARIMA forecasting. The central contribution is the combination of complementary analytical perspectives in a single interpretable pipeline. Gaussian analysis provides point-wise deviation information, entropy measures broader variability, BCI summarizes multiple behavioral dimensions, and Markov modeling adds sequential prediction. This creates a structured pathway from raw wearable observations to understandable behavioral information. The framework provides a practical foundation for research and personalized lifestyle analytics. Its modular architecture can be extended with robust anomaly detection, adaptive personalization, explainable machine learning, real-time streaming, multimodal sensing, and privacy-preserving computation. The system should remain an analytical and research tool unless its predictions are independently validated for a specific health or clinical application.

[1] Y. Wu, Y. Feng, and Z. Lin, “Multivariate time-series classification usingGaussian models and KL-divergence metrics,” Proc. AAAI Conf. Artif. Intell., vol. 33, no. 1, pp. 6402–6409, 2019. [2] Y. Wang, S. Li, and G. Zhao, “Differential evolution with entropy-basedadaptive mutation strategy guided by a Markov chain,” IEEE Access, vol. 9, pp. 116825–116838, 2021. [3] W. Huang, W. Zhang, G. Zeng, and G. Zhou, “A short-term load forecastingmodel of power system based on ARMA and Kalman filter,” Energy Rep., vol. 3, pp. 51–56, 2017. [4] C. E. Shannon, “A mathematical theory of communication,” Bell Syst. Tech.J., vol. 27, pp. 379–423, 623–656, 1948. [5] L. Rabiner, “A tutorial on hidden Markov models and selected applications inspeech recognition,” Proc. IEEE, vol. 77, no. 2, pp. 257–286, 1989. [6] J. D. Hamilton, Time Series Analysis. Princeton, NJ, USA: Princeton Univ.Press, 1994. [7] G. E. P. Box, G. M. Jenkins, G. C. Reinsel, and G. M. Ljung, Time SeriesAnalysis: Forecasting and Control, 4th ed. Hoboken, NJ, USA: Wiley, 2008. [8] T. Hastie, R. Tibshirani, and J. Friedman, The Elements of StatisticalLearning, 2nd ed. New York, NY, USA: Springer, 2009. [9] L. Breiman, “Random forests,” Mach. Learn., vol. 45, pp. 5–32, 2001. [10] A. Hyvärinen, “Independent component analysis: Recent advances,” Philos. Trans. Roy. Soc. A, vol. 371, 2013.