Search

IoT environments are continuously changing. Changes may come from the service, connectivity, or physical layers of the IoT architecture. Therefore, to function appropriately, the system needs to dynamically adapt to its environment. In previous work, we posited eight challenges to foster adaptation through all architecture layers of IoT systems. In this paper, we address the challenges to manage the inclusion of new devices and devices' transient connection , by means of dynamic adaptations incorporated into our proposed software architecture for adaptive IoT systems. To manage dynamic adaptations, we extend the reference IoT architecture with our specialized components. In particular, we use (1) ontologies and instances to represent the domain knowledge; (2) a matching algorithm to pair services and IoT devices, taking into account their functional requirements, quality attributes and sensors properties; and (3) a match update algorithm used whenever sensors become (un)available. We evaluate the effectiveness of our solution with respect to the accuracy of matching services and IoT devices, and the response to environment changes. • This paper presents a new adaptive architecture for IoT Systems. • Ontology matching algorithm-based architecture to manage transient connections. • Functional, non-functional, and QoS matching algorithm between IoT components. • Matching algorithm with better precision and recall compared with the gold-standard. • Architecture deployed in a real case study for a SUDS system in the city of Bogotá. [ABSTRACT FROM AUTHOR]

Reliable, timely and detailed information on lung cancer prevalence, mortality and costs from middle-income countries is essential to policy design. Thus, we aimed to develop an electronic algorithm to identify lung cancer prevalent patients in Colombia by using administrative claims databases, as well as to estimate prevalence rates by age, sex and geographic region. We performed a cross-sectional study based on national claim databases in Colombia (Base de datos de suficiencia de la Unidad de Pago por Capitación and Base de Datos Única de Afiliados) to identify lung cancer prevalent patients in 2017, 2018 and 2019. Several algorithms based on the presence or absence of oncological procedures (chemotherapy, radiotherapy and surgery) and a minimum number of months that each individual had lung cancer ICD-10 codes were developed. After testing 16 algorithms, those with the closest prevalence rates to those rates reported by aggregated official sources (Global Cancer Observatory and Cuenta de Alto Costo) were selected. We estimated prevalence rates by age, sex and geographic region. Two algorithms were selected: i) one algorithm that was defined as the presence of ICD-10 codes for 4 months or more (the sensitive algorithm); and ii) one algorithm that was defined by adding the presence of at least one oncological procedure (the specific algorithm). The estimated prevalence rates per 100,000 inhabitants ranged between 11.14 and 18.05 for both, the contributory and subsidized regimes over years 2017, 2018 and 2019. These rates in the contributory regime were higher in women (15.43, 15.61 and 17.03 per 100,000 for years 2017, 2018 and 2019), over 65-years-old (63.45, 56.92 and 61.79 per 100,000 for years 2017, 2018 and 2019) who lived in Central, Bogota and Pacific regions. Selected algorithms showed similar aggregated prevalence estimations to those rates reported by official sources and allowed us to estimate prevalence rates in specific aging, regional and gender groups for Colombia by using national claims databases. These findings could be useful to identify clinical and economical outcomes related to lung cancer patients by using national individual-level databases. [ABSTRACT FROM AUTHOR]

This paper proposes an iterative mathematical optimization framework to solve the layout and hydraulic design problems of sewer networks. The layout selection model determines the flow rate and direction per pipe using mixed-integer programming, which results in a tree-like structured network. This network layout parametrizes a second model that determines hydraulic features including the diameter and the upstream and downstream invert elevations of pipes using a shortest path algorithm. These models are embedded in an iterative scheme that refines a cost function approximation for the first model upon learning the actual design cost from the second model. The framework was successfully tested on two sewer network benchmarks from the literature and a real sewer network located in Bogotá, Colombia, that is proposed as a new instance. For both benchmarks, the proposed methodology found a better solution with up to 42% cost reduction compared to the best methodologies reported in the literature. These are near-optimal solutions with respect to construction cost that satisfy all hydraulic and pipe connectivity constraints of a sewer system. [ABSTRACT FROM AUTHOR]