Resumen de Diseño e implementación de una arquitectura de capas enfocada a microservicios en el contexto EHEALTH

Huriviades Calderón

• español

  Esta memoria de tesis doctoral tiene el objetivo primordial de proponer una arquitectura multicapa dirigida a soportar la interoperabilidad entre las aplicaciones de salud tradicionales involucradas en las instituciones médicas (p. ej. ficha clínica digital), comúnmente desarrolladas bajo el estilo arquitectónico Service-Oriented Architecture ¿SOA¿, y en conjunto con aplicaciones modernas bajo el estilo arquitectónico Microservice Architecture ¿MSA¿ adaptados a variantes Basado en aprendizaje automático mediante el uso de metadatos estandarizados proveniente de diversas fuentes de datos dinámicos y heterogéneos. Sin embargo, estos rápidos cambios han tenido abrumadores impactos de cómo se relaciona la salud con el mundo digital causado por la tendencia de la digitalización e impulsado por políticas de un gobierno electrónico.

  A diferencia de otras propuestas de investigación relacionadas a la arquitectura de microservicios, esta propuesta se centra en la replicabilidad, escalabilidad e interoperabilidad de los servicios especializados que conforman a una propia característica de software basada en la versatilidad de alto rendimiento. Para demostrar la viabilidad del ecosistema de salud digital fue necesario adaptar esta propuesta a un caso de uso real, específicamente al proyecto SPIDEP (Diseño e implementación de un sistema inteligente de bajo coste para el prediagnóstico y la teleasistencia de enfermedades infecciosas en personas de edad avanzada) perteneciente a la convocatoria internacional ERANET LAC 2015-FP7, cuya finalidad del proyecto fue construir un sistema inteligente basado en las tecnologías de la información y las comunicaciones para apoyar el diagnóstico temprano de enfermedades infecciosas respiratorias y urinarias en personas mayores a través de la recolección y seguimiento remoto de los pacientes en residencias, lo que permitió la detección de anomalías en los signos vitales.

  Teniendo en cuenta lo anterior, mediante esta memoria se han realizado cincos contribuciones principales: (i) demostrar las implicaciones y desafíos que conlleva implementar esta arquitectura multicapa orientada a microservicios en un ecosistema de salud digital; (ii) proveer los pasos necesarios para el diseño, implementación y despliegue de esta propuesta adaptado a un caso de uso exitoso mediante una plataforma SPIDEP; (iii) señalar que esta propuesta es respaldada por una gran base de datos que integra fuentes de información diversas y heterogéneas obtenida en la duración de SPIDEP; (iv) definir las característica de softwares centrada en solventar las necesidades de la televigilancia médica enfocado a arquitectura multicapa; (v) que la implementación de SOA y MSA depende de la naturaleza y las necesidades de las organizaciones médicas (p. ej. rendimiento, interoperabilidad u otros). Sin embargo, los patrones arquitectónicos SOA y MSA pueden considerarse aliados complementarios para una arquitectura interempresarial o intergeneracional que confiere un conjunto de diferentes servicios, en lugar de ser competidores. Cabe destacar, que esta propuesta se creó con miras de ser adaptado a otras áreas eHealth (p. ej. diálisis, diabetes, cáncer de colon u otros).

  Por otro lado, esta memoria es presentada bajo la modalidad de compendio basándose en tres contribuciones científicas publicadas en prestigiosas revistas que corresponden a sendos trabajos de investigación que reflejan mediante un sólido hilo conductor la aportación original de la tesis para proporcionar servicios inteligentes avanzados en el ámbito de un ecosistema de salud digital desarrollando a tal efecto una propuesta de arquitectura de varias capas que soporta novedoso y avanzados microservicios especializados.

• English

  This doctoral thesis aims to primarily propose a multi-layer architecture aimed at supporting interoperability between traditional health applications involved in medical institutions (e.g., digital medical record), commonly developed under the Service-Oriented Architecture (SOA) architectural style, and in conjunction with modern applications under the Microservice Architecture (MSA) adapted to variations based on machine learning through the use of standardized metadata from various sources of dynamic and heterogeneous data. However, these rapid changes have had overwhelming impacts on how health relates to the digital world, caused by the digitization trend and driven by e-government policies.

  Unlike other research proposals related to microservices architecture, this proposal focuses on the replicability, scalability, and interoperability of specialized services that make up a high-performance versatility-based software feature. To demonstrate the feasibility of the digital health ecosystem, it was necessary to adapt this proposal to a real-world use case, specifically the SPIDEP project (Design and implementation of a low-cost intelligent system for the pre-diagnosis and teleassistance of infectious diseases in elderly people) belonging to the international ERANET LAC 2015-FP7 call, whose goal was to build an intelligent system based on information and communication technologies to support early diagnosis of respiratory and urinary infectious diseases in older people through the remote collection and monitoring of patients in nursing homes, which allowed for the detection of anomalies in vital signs.

  On this basis, this thesis has made five main contributions: (i) demonstrate the implications and challenges involved in implementing this microservices-oriented multi-layer architecture in a digital health ecosystem; (ii) provide the necessary steps for the design, implementation, and deployment of this proposal adapted to a successful use case through the SPIDEP platform; (iii) indicate that this proposal is supported by a large database that integrates various and heterogeneous sources of information obtained during the SPIDEP project; (iv) define the software features focused on solving the needs of medical telemonitoring with a multi-layer architecture; (v) and that the implementation of SOA and MSA depends on the nature and needs of medical organizations (e.g.

  performance, interoperability, or others) and that the implementation of SOA and MSA depends on the nature and needs of medical organizations (e. v. However, the SOA and MSA architectural patterns can be considered complementary allies for an inter-enterprise or inter-generational architecture that confers a set of different services rather than being competitors. It's worth mentioning that this proposal was created with the intention of being adapted to other eHealth areas (e.g., dialysis, diabetes, colon cancer, or others).

  On the other hand, this thesis is presented in the form of a compendium based on three scientific contributions published in prestigious journals, which correspond to respective research works that reflect, through a solid thread, the original contribution of the thesis to provide advanced intelligent services in the field of a digital health ecosystem, by developing in this way a proposal of multilayer architecture that supports advanced specialized microservices, whose publications are: (i) Medical Prognosis of Infectious Diseases in Nursing Homes Using Machine Learning on Clinical Data Collected in Cloud Microservices, the first article focuses on the initial proposal of a flexible microservices architecture that provides access and functionality to the system oriented to a digital health ecosystem in which it can be made to treat infectious diseases in elderly people because these patients tend to arrive at medical appointments with advanced symptoms. (ii) Telemonitoring System for Infectious Disease Prediction in Elderly People Based on a novel microservice architecture, this second article describes the design, development, and implementation of new services focused on a microservices architecture that allows the detection and clinically assisted diagnosis of infectious diseases in elderly patients based on the use of telemonitoring. Unlike the first article, in this one we focus more on the aspect of software engineering, as we propose a more mature workflow for the continuous deployment and automation of the developed microservices. (iii) Evaluating Service-Oriented and Microservice Architecture Patterns to Deploy eHealth Applications in a Cloud Computing Environment. This third article proposes a new framework for the conception of an eHealth platform focused on cloud computing environments, given current and emerging approaches regarding the development of telemonitoring-based recommendation systems and access to digital clinical history for different health areas. Unlike the previous two articles, this one evaluates and compares the performance of the most used architectural patterns for developing health applications, both in their SOA and MSA variants, using as a reference the quantitative values obtained from the various performance tests and its ability to adapt to the software characteristics required by SPIDEP. As a result, it was determined that MSA presents better performance in terms of the performance quality attribute (54.21%), in the same way that when processing multiple requests from different services, the response time was lower in comparison with SOA (7.34%), but the bandwidth consumption in MSA was more significant than SOA (73.80%)