This chapter is from the book
This chapter is from the book
This chapter is from the book
1.5 Summary and Open Challenges
This chapter presented a broad overview of trends in MCPS and the design challenges that these trends present. It also discussed possible approaches to address these challenges, based on recent results in MCPS research.
The first challenge is related to the prevalence of software-enabled functionality in modern MCPS, which makes assurance of patient safety a much harder task. Model-based development techniques provide one way to ensure the safety of a system. Increasingly, model-based development is embraced by the medical devices industry. Even so, the numerous recalls of medical devices that have occurred in recent years demonstrate that the problem of device safety is far from being solved.
The next-level challenge arises from the need to organize individual devices into a system of interconnected devices that collectively treat the patient in a complex clinical scenario. Such multi-device MCPS can provide new modes of treatment, give enhanced feedback to the clinician, and improve patient safety. At the same time, additional hazards can arise from communication failures and lack of interoperability between devices. Reasoning about safety of such on-demand MCPS, which are assembled at the bedside from available devices, creates new regulatory challenges and requires medical application platforms—that is, trusted middleware that can ensure correct interactions between the devices. Research prototypes of such middleware are currently being developed, but their effectiveness needs to be further evaluated. Furthermore, interoperability standards for on-demand MCPS need to be further improved and gain wider acceptance.
To fully utilize the promise inherent in multi-device MCPS, new algorithms need to be developed to process and integrate patient data from multiple sensors, provide better decision support for clinicians, produce more accurate and informative alarms, and so on. This need gives rise to two kinds of open challenges. On the one hand, additional clinical research as well as data analysis needs to be performed to determine the best ways of using the new information made available through combining multiple rich data sources. On the other hand, new software tools are needed to facilitate fast prototyping and deployment of new decision support and visualization algorithms.
MCPS promises to enable a wide array of physiological closed-loop systems, in which information about the patient’s condition, collected from multiple sensors, can be used to adjust the treatment process or its parameters. Research on such closed-loop control algorithms is gaining prominence, especially as means to improve glycemic control in patients with diabetes. However, much research needs to be performed to better understand patient physiology and develop adaptive control algorithms that can deliver personalized treatment to each patient.
In all of these applications, patient safety and effectiveness of treatment are the two paramount concerns. MCPS manufacturers need to convince regulators that systems they build are safe and effective. The growing complexity of MCPS, the high connectivity, and the prevalence of software-enabled functionality make evaluation of such systems’ safety quite difficult. Construction of effective assurance cases for MCPS, as well as for CPS in general, remains a challenge in need of further research.