Oracle (Belgium)

Introduction: ICU admission occurs for active treatment (ventilation, vasopressors) and to monitor patients at risk. The Acute Physiology and Chronic Health Evaluation (APACHE) IVb defines Low Risk Monitor (LRM) as not actively treated on ICU day 1 and < 10% prospective risk of ever needing active treatment. LRM patients potentially fill ICU beds required by acutely ill patients. We investigated if unprecedented ICU demand during the COVID-19 pandemic decreased LRM admissions during COVID surges. Methods: Retrospective analysis of hospitals tracking COVID-19 status and consistently contributing to the APACHE database March 23, 2020 to December 31, 2021. Baseline pre-pandemic data was also assessed. Patients with primary surgical and trauma diagnoses were removed to eliminate incidental COVID diagnoses. Pearson’s correlation coefficient (r) assessed the weekly relationship between %COVID and LRM patients. Results: 117,004 patients were admitted to ICU at 43 hospitals. Baseline LRM averaged 28.6% pre-COVID. During successive COVID peaks in April, July and December 2020 and April, August and December 2021, there was high inverse correlation (r=-0.90) between COVID census and LRM percentage. For example, in September 2020 COVID% was 7.81 and LRM was 28.2%. In December 2020, COVID surged to 31.1% and LRM dropped to 21.3%. These percentages returned to COVID 9.5% and LRM 28.2% during the March 2021 trough. Hospital mortality was 10.9% pre-pandemic, and 14.69% actual/13.66 predicted (SMR=1.08) from April 2020 through December 2021. Mean ventilator days were 4.08 pre- and 5.58 pandemic. ICU LOS increased from 3.52 to 4.16 days (ratio 1.11). Conclusions: LRM admissions decreased dramatically during successive COVID-19 surges, demonstrating considerable elasticity in ICU triage decisions. Mortality, ventilator days and ICU LOS all increased during the pandemic compared to baseline. Consistent measurement of % LRM may be helpful in recognizing opportunities to reduce inappropriate ICU bed utilization and as a marker of strained capacity.

No brand owner wants their ad turning up next to unsavoury online content. With the automated daily delivery of trillions of ad impressions across the open web, brands and agencies rely on systems and tools to ensure their ad is being placed in the most appropriate contexts. Unfortunately, in the last 25 years of programmatic advertising, the tools that we rely on to protect brands have not always kept up with our desire and ability to safely spread our brand message far and wide. Tools such as keyword blocklists were once the de facto method for managing correct placement. Recently, however, media coverage on the blocking of COVID-19, progressive content and even small-scale scandals have shown how inefficient these antiquated tools can be. Blocklists use single terms to define suitability, which in turn generates broad-brushstroke stereotyping, impacting scale and accuracy of ad placement. As well as missed opportunities and poor ad placement decisioning, this type of binary approach to safety disproportionately impacts progressive content. By featuring them and the tools that act on them, brands can limit consumer exposure to valued content by defunding the publishers giving voice to these potentially progressive topics. Thankfully, there are now better ways to manage brand safety online. Contextual brand suitability tools disregard single terms and instead ascertain relevance on the basis of an understanding and appreciation of more nuanced contexts. There are several steps any brand or agency can take to begin to evolve their brand safety strategy. These steps are an important part of the journey to increase online safety while also ensuring we are not our own blocker to safer, smarter and progressive approaches to online ad placement decisions.

Abstract This paper describes a new approach to designing and building missiles by Raytheon Missile and Defense Systems. Key aspects of the approach include (1) modular common components connected through identified standards, (2) modular open systems approaches and standards based interfaces, (3) Feature‐based Product Line Engineering (FbPLE) practices for identifying commonality and managing variation, and (4) implementing digital transformation through digital engineering capabilities to begin a missile's digital twin. Much more ambitious than simply reusing existing component designs from previously built missiles, this approach involves automatic generation, exploration, and pruning of an automatically generated trade space of possible missile designs that satisfy a given set of requirements. The goal is to radically lower development and production costs by rapidly settling on a viable design that can be taken to design validation, then complete design and production, all in a digital ecosystem. The paper focuses on the FbPLE aspect of the engineering approach and shows how its technology is used to create and manage the trade space, and then create a digital twin of a chosen design.

Dissolved gas analysis (DGA) is among the most essential techniques for diagnosis of incipient faults in power transformers. Here, a novel graphical DGA technique is proposed in which fault zones are distinguished based on certainty of prediction. The Duval Pentagon 1 and gas ratio combination methods are two most recent techniques with high prediction accuracy. In the Duval Pentagon 1, the rigidly separated distinct fault zones reduce the flexibility of analysis because the fault distributions themselves are not that strictly separated. This also prevents the full utilisation of the information available from the distribution patterns of the graphical representation. This problem has been addressed by overlapping individual fault zones and overlapping them using a multi‐layer perceptron (MLP) network with fuzzy class boundaries. Then, in the regions, where multiple fault zones overlap, a modified gas ratio combination method is applied. Finally, a fuzzy decision‐making system is developed for predicting the fault using information from both graphical distribution and gas ratios. The combined accuracy of the regions of certainty has been found exceptionally high (98.36%) compared to the regions of uncertainty (58.97%), whereas the overall prediction accuracy of the proposed technique is found comparatively higher (83%) than both the existing methods.

With the proliferation of global positioning systems (GPS) enabled devices, a growing number of database systems are capable of storing and querying different spatial objects including points, polylines and polygons. In this paper, we present our experience with supporting one important class of spatial queries in these database systems: distance queries. For example, a traveler may want to find hotels within 500 meters of a nearby beach. In addition, this paper presents new techniques implemented in Oracle Spatial for some distance-related problems, such as the maximum distance between complex spatial objects, and the diameter, the convex hull and the minimum bounding circle of complex spatial objects. We conduct our experiments by utilizing real-world data sets and demonstrate that these distance and distance-related queries can be significantly improved.