MedPath: A process-based modeling language for designing care pathways

Highlights

• Doctors can benefit from simple DSLs that are less verbose and closer to the sketches they make manually when modeling care pathways.

• Specifying the types of activities in a medical action flow improve organization and disposition of medical activities as a process-based flow.

• Model-Based Engineering can fully translate the concepts of care pathways into a programmable artifact (a Domain Specific Language).

• MedPath is currently employed in both hospital and clinical care, having practical use one of its validation metrics.

Abstract

Context

Medical professionals and hospitals promote solutions like care pathways and Health Information Systems (HIS) to support medical conduct and improve the quality of medical care.

Purpose

This study proposes MedPath: a Domain Specific Language (DSL) for modeling care pathways based on the paradigms of Model-Based Engineering (MBE) that can be integrated into software solutions.

Procedures

We have developed MedPath's abstract syntax with the Eclipse Modeling Framework by employing Ecore technology and concrete syntax with the Eclipse Sirius.

Findings

We have modeled over 85 care pathways that are in use in 45 hospitals in Brazil. MedPath-originated pathways have been used over 3.2 million times since October 2017. We conducted a survey among the professionals who used MedPath to evaluate user satisfaction.

Conclusions

We believe MedPath can translate any care pathway into an action flow with its current abstractions. MedPath makes care pathways more easily integrated into HIS and electronic patient records, as it enables programmatic modeling and generates consumable artifacts.

Introduction

Medical care comprehends critical processes that concern patient welfare. Medical errors could be critical and cost patients their lives as well as substantial amounts in damages to hospitals and healthcare providers. In the early 1980s, American hospitals started implementing step by step definitions of common medical practice called care pathways [1], [2]. They were designed to be patient-focused workflows to aid medical professionals during patient care. Care pathways increased in use, amount, and quality over the years [3], [4], improving the quality of medical care and reducing the possibility of errors. They work as a boilerplate that translates medical care into a sequence of steps, similar to business process action flows [5], [6], [7], facilitating conduct standardization.

Health Information Systems (HIS) [8] are informatized systems designed to manage medical care data. However, they focus on static forms to register patient treatment data rather than proactive workflow-based systems. As a result, systems are often complicated for medical professionals and take much time of the patient care process, without guiding the professionals to follow conduct standards. The lack of proper tools to manage their execution makes the care process almost entirely dependent on the doctor's experience, making it difficult to enforce conduct standardization and to know whether doctors are following the best practices or not.

Some studies propose systems, ontologies (formal representations of a concept or domain) [9] or approaches to modeling care pathways with Business Process Modeling Notation (BPMN), declarative languages [10], [11], [12], and Universal Modeling Language (UML) [13], [14], [15], [16]. These approaches sometimes lack on infrastructure for integration and execution, or rely on languages that are not customized for medical context.

We propose MedPath, a language to define care pathways based on the paradigms of Model-Based Engineering (MBE) [17], [18]. MedPath is a Domain Specific Language (DSL) [19], [20], [21] that encapsulates the concepts of care pathway creation, and was born directly from medical practice in the context of hospital emergency care. That way, healthcare professionals can immediately identify with the concepts contained in it, which makes using the language a more natural process. The DSL can be easily embedded into programmatic applications for building and executing care pathways organically, in real environments.

As MedPath was born from practical use, we validate its applicability by observing statistics from its employment in real-life scenarios and evaluate its efficiency by measuring user satisfaction data collected in a survey applied to MedPath's users.