Trustworthy Composition of Dynamic App-Centric Architectures for Medical Application Platforms

Abstract:

Medical devices are typically developed as stand-alone units. Current industrial Verification and Validation (V&V) techniques primarily target stand-alone systems. Moreover, the US Food and Drug Administration's (FDA) regulatory clearance processes are designed to approve such devices that are integrated by a single manufacturer with complete control over all components. From a systems engineering perspective, it is well understood that the current state of medical practice uses non-integrated devices and health information systems (HIS) cooperatively according to informal manual protocols to deliver clinical solutions. Providing the IT infrastructure, safety/security framework, and trustworthy composition techniques to integrate devices and HIS, and automatically coordinate their actions as a system of systems (SoS), would provide opportunities to implement, e.g., multi-device smart alarms and safety interlocks, enable better clinical decision support, automate clinical workflows, and implement closed loop control. A Medical Application Platform (MAP) is an emerging vision that would provide device and HIS interoperability, safety critical network middleware, and an execution environment for clinical applications (apps) that encode the envisioned medical SoS behaviors described above. Realizing this vision can bring about a paradigm shift that can dramatically increase effectiveness and reduce costs of health care delivery. Numerous challenges exist that are preventing this vision of deeply integrated and highly beneficial cyber-physical medical systems from being realized. These include: (a) lack of domain knowledge and infrastructure on the part of academic researchers as they seek to develop appropriate V&V technologies, safety-critical system components, and programming models, (b) lack of awareness in industry of formal modeling and verification technologies that could tackle the problems of compositional construc-tion of highly interactive safety-critical systems, and (c) lack of realistic applications of cutting edge V&V and programming technologies in the device integration space that might provide science-based inputs for guiding the formation of new regulatory policies. Only a coordinated, broad-based effort of academics, industry, and regulatory officials can likely solve these interrelated challenges. Working with a synergistic and multi-disciplinary team of academic researchers, clinicians, device manufacturers, device integrators, health IT companies, regulatory agencies, standards/certification or-ganizations, this project is developing nextgeneration medical device coordination infrastructure to support a broad-based community effort in this domain and to seed the infrastructure with key technological advances that achieve coverage across the areas that believed to be especially important for realizing the vision of safe and effective medical application platforms. The key components of this project include:

• A next generation open-source MAP prototype medical device coordination framework (MDCF) to enable interdisciplinary research on technical topics needed to achieve the vision of construction, verification, certification, and clinical deployment of safe and secure MAPs.
• Formal MAP architectural specifications defining both medical device and infrastructure interfaces necessary to support 3rd party certification, explicit capture of component properties and assumptions, and interface extensibility.
• Real-time middleware with novel capabilities supporting compositional dynamic construction of regulatory-approved safe, secure, and correct MAP-based medical systems by clinicians at the point of care.
• An integrated development, verification, and certification environment for MAP apps and device interfaces based on industry standard modeling languages, development processes, and safety standards.
• Evidence-based V&V techniques, compositional contract-based approaches for reasoning about key functional and non-functional properties, e.g., timeliness, reliability, end-to-end communication latency.
• Device technology for futuristic medical devices that are designed to be deeply integrated with MAPs, allowing them to have significantly altered form-factors, to be reconfigurable, and adaptive, and to support easy care-context migration of patients.
• Definition and illustration of evidence-based safety-ecosphere for MAP systems including 3rd party certification guidelines, tools, and example artifacts, along with descriptions and examples of how regulatory oversight of MAP systems would be handled within the FDA regulatory regime.

Our team is working closely with Medical Device Plug-and-Play interoperability program led by Dr. Julian Goldman at the Center for Integration of Medicine and Innovative Technology (CIMIT), with engineers at the FDA who are collecting science-based input to shape regulatory policy, and the engineers at Underwriters Laboratories working on safety standards for medical device interoperability.