**7. Conclusion and future work**

In this chapter we addressed the problem of behavioral pattern recognition, behavior change detection and anomaly detection based on IoT data in smart city environment. We proposed a framework for behavioral change detection that will be utilized in context of mild cognitive impairment (MCI) and frailty risk assessment and detection in the City4Age project. Behavioral modeling and risk assessment for MCI and Frailty are very challenging tasks because of the large variations between each specific personal case, and the practical lack of universally agreed and adopted criteria in geriatric practice (in real-life environment, not controlled "lab" settings) on the referent thresholds or ranges of quantified risk factors or geriatric domain variables that actually denote certain MCI/frailty risk or potential onset.

[2] Van Poucke S, Thomeer M, Heath J, Vukicevic M. Are randomized controlled trials the (G) old Standard? From clinical intelligence to prescriptive analytics. Journal of Medical Internet Research. 2016;**18**(7):e185. DOI: 10.2196/jmir.5549. PMID: 27383622, PMCID:

Temporal Clustering for Behavior Variation and Anomaly Detection from Data Acquired…

http://dx.doi.org/10.5772/intechopen.75203

133

[3] Kaddachi F, Aloulou H, Abdulrazak B, Fraisse P, Mokhtari M. Unobtrusive Technological Approach for Continuous Behavior Change Detection Toward Better Adaptation of Clinical Assessments and Interventions for Elderly People. In: International Conference on Smart Homes and Health Telematics. Cham: Springer; 2017, August. pp. 21-33

[4] Sprint G, Cook DJ, Schmitter-Edgecombe M. Unsupervised detection and analysis of changes in everyday physical activity data. Journal of Biomedical Informatics. 2016

[5] Gupta M, Gao J, Aggarwal CC, Han J. Outlier detection for temporal data: A survey. IEEE Transactions on Knowledge and Data Engineering. 2014;**26**(9):2250-2267

[6] Silva JA, Faria ER, Barros RC, Hruschka ER, de Carvalho AC, Gama J. Data stream clus-

[7] Aghabozorgi S, Shirkhorshidi AS, Wah TY. Time-series clustering–A decade review.

[8] Copelli S, Mercalli M, Ricevuti G, Venturini L. City4Age frailty and MCI risk model, v2.

[9] Azkune G, Almeida A, López-de-Ipiña D, Chen L. Extending knowledge-driven activity models through data-driven learning techniques. Expert Systems with Applications.

[10] Bilmes JA. A gentle tutorial of the EM algorithm and its application to parameter estimation for Gaussian mixture and hidden Markov models. International Computer Science

tering: A survey. ACM Computing Surveys (CSUR). 2013;**46**(1):13

City4Age Project Public Deliverable D. 2017 October;**2**(06)

4954919. http://www.jmir.org/2016/7/e185

July;**63**:54-65. ISSN: 1532-0464

Information Systems. 2015;**53**:16-38

Institute. 1998;**4**(510):126

2015 April;**42**(6):3115-3128. ISSN: 0957-4174

Thus we developed data driven models based on HMMs that exploit IoT sensory data and allow automated behavior recognition, change and anomaly detection. Models are used for characterization of data that serves as an input for exploratory analytics through interactive dashboarding and/or enrichment of modeled Geriatric factors that quantify the specific behavior characterizations and risk levels for MCI and Frailty.

In future work, we will integrate results from this research in City4Age interactive monitoring dashboards and thus enable geriatricians to gain additional insights into care recipients behavior and potential risk. This will open the space for supervised behavioral scoring and risk prediction. Further, we will develop data driven behavioral models for multivariate IoT data series and explore mutual influence between series. Finally, we will evaluate more unsupervised models for behavioral modeling including deep learning models (e.g., recurrent neural networks) in the analyses.
