Ariel Oleksiak

Background/Objectives: The study aimed to evaluate the most significant factors that impact arterial blood gas parameters: pH, pO₂, pCO_2, and concentration of lactates. Methods: The study was a retrospective analysis of clinical data obtained from the patients’ records hospitalized at the Department of Pediatric Anesthesiology and Intensive Care. A total of 71 patients were enrolled in the study. A total of 479 measurements were performed for arterial blood, 41 were excluded. The analysis was performed for 438 results. The artificial neural network (ANN) regression models were applied, and the Least Absolute Shrinkage and Selection Operator (LASSO) regression was used. ANNs were built considering the following activation functions: hyperbolic tangent, linear, exponential, and logistic. The following three sets were separated: training, testing, and validation. In the case of LASSO regression, the regularization was applied, excluding insignificant variables from the model. Besides the machine learning techniques, the correlation between the variables was calculated. Results: The correlation coefficients for regression ANN models exceeded the value for testing set of 0.92. According to the sensitivity analysis, the most significant variable for pH was cCl⁻, for pO₂ it was pO₂/FiO₂, for pCO₂ it was Fshunt, and for concentration of lactates it was pH. In the case of LASSO regression for pH, the most significant factor was pCO₂, for pO₂ it was pO₂/FiO₂, for pCO₂ it was cCl⁻, and for concentration of lactates it was pCO₂. Conclusions: The results show the usefulness of machine learning methods in analyzing complex physiological relationships. Such techniques can help improve diagnostic accuracy and optimize therapeutic management in pediatric patients.

The global share of energy consumption by data centers has reached the level of several percent and continues to rise. For this reason, ensuring optimal energy efficiency of the cooling process in data centers is an essential matter. As most data centers are air-cooled, it is required to investigate an airflow pattern inside as well as the accompanying heat transfer phenomena to assess the efficiency of the cooling process of the object. What is more, the thermal state of a server room as well as its power consumption often has a transient nature not only due to turbulent airflow inside but also because of dynamically changing servers’ load and time-varying output conditions of the coolers as a consequence of their control algorithms. That is why it is necessary to use transient modeling to get a deep and complete insight into the nature of the airflow and heat transfer phenomena in a server room and the resulting energy usage. A technique which allows modeling of the airflow and heat transfer in a detailed, volumetric way is computational fluid dynamics (CFD). Unfortunately, employing CFD for complex models especially in a transient manner can become a tedious process for an engineer and a challenging task for computational resources. In this paper we present a framework which is designed for automatic numerical heat transfer analysis of data center objects, using in the backend OpenFOAM, a well-known open-source CFD toolkit. Besides the automation process, the presented framework includes the implementation of transient models of active server room equipment and modeling of their energy usage as well as support for using high performance computing (HPC) systems equipped with SLURM queuing system. The software description is followed by a presentation of its application for a study of a hypothetical server room. A transient scenario of rapid application of servers’ load is simulated, including an energy consumption analysis of cooling devices.

The growing demand for more computing resources has increased the overall energy consumption of computer systems. To support this increasing demand, power and energy consumption must be considered as a constraint on software execution. Modern architectures provide tools for managing the power constraints of a system directly. The Intel Power Cap is a relatively new tool developed to give users fine-grained control over power usage at the central processing unit (CPU) level. The complexity of these tools, in addition to the high variety of modern heterogeneous architectures, hinders predictions of the energy consumption and the performance of any target software. The application of power capping technologies usually leads to the bi-objective optimization problem for energy efficiency and execution time but optimal power constraints could also produce exceeding performance losses. Thus, methods and tools are needed to calculate the proper parameters for power capping technologies, and to optimize energy efficiency. We propose a methodology to analyze the performance and the energy efficiency trade-offs using this power cap technology for a given application. A Pareto front is extracted for the multi-objective performance and energy problem, which represents multiple feasible configurations for both objectives. An extensive experimentation is carried out to categorize the different applications to determine the overall optimal power cap configurations. We propose the use of machine learning (ML) clustering techniques to categorize each application in the target architecture. The use of ML allows us to automate the process and simplifies the effort required to solve the optimization problem. A practical case is presented where we categorize the applications using ML techniques, with the possibility of adding a new application into an existing categorization.

In this paper, we address the management of Data Centers (DCs) by considering their optimal integration with the electrical, thermal, and IT (Information Technology) networks helping them to meet sustainability objectives and gain primary energy savings. Innovative scenarios are defined for exploiting the DCs electrical, thermal, and workload flexibility as a commodity and Information and Communication Technologies (ICT) are proposed and used as enablers for the scenarios’ implementation. The technology and scenarios were evaluated in the context of two operational DCs: a micro DC in Poznan which has on-site renewable sources and a DC in Point Saint Martin. The test cases’ results validate the possibility of using renewable energy sources (RES) for exploiting DCs’ energy flexibility and the potential of combining IT load migration with the availability of RES to increase the amount of energy flexibility by finding a trade-off between the flexibility level, IT load Quality of Service (QoS), and the RES production level. Moreover, the experiments conducted show that the DCs can successfully adapt their thermal energy profile for heat re-use as well as the combined electrical and thermal energy profiles to match specific flexibility requests.

Rapid growth of demand for remote computational power, along with high energy costs and infrastructure limits, has led to treating power usage as a primary constraint in data centers. Especially, recent challenges related to development of exascale systems or autonomous edge systems require tools that will limit power usage and energy consumption. This paper presents a power capping method that allows operators to quickly adjust the power usage to external conditions and, at the same time, to reduce energy consumption and negative impact on performance of applications. We propose an optimization model and both heuristic and exact methods to solve this problem. We present an evaluation of power capping approaches supported by results of application benchmarks and experiments performed on new heterogeneous servers.

This paper addresses the problem of data centers’ cost efficiency considering the potential of reusing the generated heat in district heating networks. We started by analyzing the requirements and heat reuse potential of a high performance computing data center and then we had defined a heat reuse model which simulates the thermodynamic processes from the server room. This allows estimating by means of Computational Fluid Dynamics simulations the temperature of the hot air recovered by the heat pumps from the server room allowing them to operate more efficiently. To address the time and space complexity at run-time we have defined a Multi-Layer Perceptron neural network infrastructure to predict the hot air temperature distribution in the server room from the training data generated by means of simulations. For testing purposes, we have modeled a virtual server room having a volume of 48 m³ and two typical 42U racks. The results show that using our model the heat distribution in the server room can be predicted with an error less than 1 °C allowing data centers to accurately estimate in advance the amount of waste heat to be reused and the efficiency of heat pump operation.