TL;DR
This paper introduces Protocure, a project that applies formal methods to improve the quality and usability of medical guidelines. By treating protocols as programs and using formal verification, they demonstrate that complex medical guidelines can be analyzed for ambiguities and incompleteness, paving the way for "living guidelines" that are continuously updated and more reliable.
What Problem Does This Paper Address?
Medical guidelines are crucial for improving patient care quality and reducing healthcare costs. However, a significant challenge is that these guidelines are often static documents, revised only every few years. This lag means they can become outdated quickly as medical knowledge rapidly advances, leading to applicability issues and potential errors. Furthermore, current guidelines are frequently written in an unstructured, narrative format, making them prone to ambiguity and incompleteness. The Protocure project aims to address these issues by developing methods to ensure the quality and facilitate the continuous updating of medical protocols, moving towards the concept of "living guidelines."
Methods
The Protocure project employs a multi-stage approach to formalize and verify medical protocols.
- Selection of Reference Protocols: Two real-life medical protocols with diverse characteristics were chosen for analysis.
- Modelling in Asbru: The selected protocols were first translated into the Asbru language, an intermediate representation designed for time-oriented clinical guidelines. This step helps structure the informal knowledge.
- Formalization in KIV: The Asbru models were then translated into a formal representation using the KIV interactive verification system. This formalization is crucial for enabling rigorous analysis.
- Formal Semantics: A formal semantics was defined for the core constructs of the Asbru language to ensure precise interpretation for verification.
- Property Identification and Verification: Interesting properties of the protocols were identified, and formal verification techniques were used within KIV to check these properties. This involved proving that the formalized protocols adhere to desired behaviors.
- Evaluation by Medical Experts: The results of the formalization and verification process, including identified anomalies, were presented to medical professionals to assess the practical utility and impact of the approach.
The Protocure II project extends this by integrating formal methods into the entire life-cycle of guideline development, from initial knowledge gathering to maintenance, viewing guideline development as analogous to software engineering.
Datasets
The paper describes the use of two real-life medical protocols as case studies:
- A jaundice protocol (likely the "Practice parameter: management of hyperbilirubinemia in the healthy term newborn" by AAP [1]).
- A diabetes protocol (likely the "NHG-Standaard Diabetes Mellitus Type 2" [6]).
These protocols were chosen to cover a wide range of characteristics and served as the basis for the formalization and verification process.
Key Insights and Ideas
- Formal Methods for Clinical Quality: Applying formal methods, typically used in software engineering, can significantly improve the quality and reduce ambiguities in medical guidelines.
- "Living Guidelines" are Achievable: The static nature of current guidelines is a major bottleneck. Formal methods offer a path towards dynamic, continuously updated "living guidelines."
- Protocol ≈ Program Metaphor: Treating medical protocols as programs allows for the application of established software engineering techniques, including formal verification, to ensure correctness and identify errors.
- Identifying Anomalies: The formalization process revealed various anomalies in real-life protocols, such as ambiguities and incompleteness, which are difficult to detect through informal review alone.
- Feasibility of Formal Verification: The project demonstrated that formal verification of complex medical protocols is feasible, even if it requires adding assumptions to make properties true.
Key Results
- Consolidated Formal Language: The project led to the identification of core constructs for modeling medical protocols, termed "Asbru-Light," with clarified and formally specified semantics.
- Formalized Protocols: Two real-life medical protocols were successfully formalized in both Asbru (intermediate representation) and KIV (formal representation). For instance, the jaundice protocol, originally 10 pages, was represented in 18 pages of Asbru and 30 pages of KIV code.
- Anomaly Detection: The formalization process uncovered a significant number of anomalies:
- Jaundice protocol: 1 ambiguity, 10 incompleteness anomalies, 6 inconsistencies.
- Diabetes protocol: 4 ambiguities, 38 incompleteness anomalies, 2 redundancies, no inconsistencies.
- Verification Proofs: Formal verification proofs were successfully completed for identified protocol properties, demonstrating the practical application of the KIV system. These proofs could be extensive, sometimes containing 1000 steps.
- Expert Validation: Medical experts acknowledged the utility of formal methods in spotting protocol problems and suggested that such issues should be avoided by guideline developers.
Limitations and Future Work
- Interpretation of Anomalies: The paper notes that the interpretation of the identified anomalies and their absolute numbers requires the expertise of medical practitioners.
- Protocure II Focus: The Protocure II project aims to extend the work by integrating formal methods into the entire guideline life-cycle, including development, deployment, and maintenance, moving beyond just analyzing existing protocols.
- "Living Guidelines" Development: Future work will focus on techniques and tools to support the whole guideline development process, enabling the creation and maintenance of "living guidelines." This includes developing process models, representation levels, transformation tools, visualization aids, design pattern libraries, and validation strategies.
- Maintenance of Evolving Guidelines: For "living guidelines," techniques for validating and verifying evolving guidelines, such as reusing proofs and managing changes, will be a key area of focus.
Summary updated