A Preliminary Study of a Probabilistic Risk-based Approach for Ambient Intelligence Healthcare Systems

The Ambient Intelligence (AmI) paradigm applied to the healthcare sector is a promising solution to develop software-based systems capable of supporting medical procedures and activities carried out in a close, high-regulated, and complex healthcare environment. An AmI Healthcare System (AmI-HS) which may impact on the health and life of its users (i.e. doctors, caregivers, patients, etc.) is considered as a Medical Device (MDs), and thus subject to pass through a cumbersome risk-based regulatory process which evaluates and certifies the system safety before it is put on the market. Thus, a human-centred risk analysis is of paramount importance to establish the safety level of an AmI-HS. In this paper, we propose a dynamic probabilistic risk assessment (DPRA) approach for AmI-HS which allows the quantitative assessment of risk in different hazard scenarios in order both to support the design and development of AmI-HSs and to provide those objective evidences needed during the regulatory process. In addition, to support our risk-based methodology we define a probabilistic risk model (PRM), based on an extension of a Markov Decision Process (MDP), capable of taking into account two main peculiarities of AmI-HSs: context-awareness and personalisation. Some preliminary results show the feasibility of our approach and the capability of our model to assess risk of context-aware hazard scenarios.