Search

Background:Postpancreaticoduodenectomy (PD) hemorrhage (PPH) is a life-threatening complication after PD. The main objective of this study was to evaluate incidence and factors associated with late PPH as well as the management strategy and outcomes. Methods: Between May 2017 and March 2020, clinical data from 192 patients undergoing PD were collected prospectively in the CHIRPAN Database (NCT02871336) and retrospectively analyzed. In our institution, all patients scheduled for a PD are routinely admitted for monitoring and management in intensive/intermediate care unit (ICU/IMC). Results: The incidence of late PPH was 17% (32 of 192), whereas the 90-day mortality rate of late PPH was 19% (6 of 32). Late PPH was associated with 90-day mortality (P = 0.001). Using multivariate analysis, independent risk factors for late PPH were postoperative sepsis (P = 0.036), and on day 3, creatinine (P = 0.025), drain fluid amylase concentration (P = 0.023), lipase concentration (P0.001), and C-reactive protein (CRP) concentration (P0.001). We developed two predictive scores for PPH occurrence, the PANCRHEMO scores. Score 1 was associated with 68.8% sensitivity, 85.6% specificity, 48.8% predictive positive value, 93.2% negative predictive value, and an area under the receiver operating characteristic curves of 0.841. Score 2 was associated with 81.2% sensitivity, 76.9% specificity, 41.3% predictive positive value, 95.3% negative predictive value, and an area under the receiver operating characteristic curve of 0.859. Conclusions: Routine ICU/IMC monitoring might contribute to a better management of these complications. Some predicting factors such as postoperative sepsis and biological markers on day 3 should help physicians to determine patients requiring a prolonged ICU/IMC monitoring.

The potential for the contribution of the built environment towards sustainable development is recognized around the world. The need to achieve thermal comfort has proven to be the prime source of energy consumption in buildings, with mechanical ventilation and air-conditioning known to represent more than half of the energy bill. The effect of climate change has exacerbated the problem, leading to a vicious cycle of emitting more greenhouse gases in bringing comfortable indoor environments, while contributing further to climate change with warmer summers and colder winters. A very effective way to decouple economic growth and urbanization with increasing carbon footprint of our building stock is through the integration of passive measures, which hold huge potential for climate zones characterized as hot and warm. Moreover, the variability of climate means that permanent passive measures do not represent the optimum configuration for harnessing the natural resources in the form of daylight, natural ventilation and solar radiation, calling for building controls to regulate these passive elements. Furthermore, the need to set suitable control strategies for modulating these passive measures require a knowledge base to understand their influence on the indoor environment with respect to the external climatic conditions. The complexity of the interaction between the external and internal environments through the building envelope has led to renewed interest in adopting an AI approach to the problem. This paper presents a methodology developed to assess and quantify the efficacy of passive measures with associated controls for regulating specific parameters pertaining to the indoor environment, and presents simulation results for the automation of window shading with respect to indoor temperature and illumination level as an example of the proposed framework.