Skilful precipitation nowcasting using deep generative models of radar.

Precipitation nowcasting, the high-resolution forecasting of precipitation up to two hours ahead, supports the real-world socioeconomic needs of many sectors reliant on weather-dependent decision-making1,2. State-of-the-art operational nowcasting methods typically advect precipitation fields with radar-based wind estimates, and struggle to capture important non-linear events such as convective initiations3,4. Recently introduced deep learning methods use radar to directly predict future rain rates, free of physical constraints5,6. While they accurately predict low-intensity rainfall, their operational utility is limited because their lack of constraints produces blurry nowcasts at longer lead times, yielding poor performance on rarer medium-to-heavy rain events. Here we present a deep generative model for the probabilistic nowcasting of precipitation from radar that addresses these challenges. Using statistical, economic and cognitive measures, we show that our method provides improved forecast quality, forecast consistency and forecast value. Our model produces realistic and spatiotemporally consistent predictions over regions up to 1,536 km × 1,280 km and with lead times from 5-90 min ahead. Using a systematic evaluation by more than 50 expert meteorologists, we show that our generative model ranked first for its accuracy and usefulness in 89% of cases against two competitive methods. When verified quantitatively, these nowcasts are skillful without resorting to blurring. We show that generative nowcasting can provide probabilistic predictions that improve forecast value and support operational utility, and at resolutions and lead times where alternative methods struggle.

A clinically applicable approach to continuous prediction of future acute kidney injury.

The early prediction of deterioration could have an important role in supporting healthcare professionals, as an estimated 11% of deaths in hospital follow a failure to promptly recognize and treat deteriorating patients1. To achieve this goal requires predictions of patient risk that are continuously updated and accurate, and delivered at an individual level with sufficient context and enough time to act. Here we develop a deep learning approach for the continuous risk prediction of future deterioration in patients, building on recent work that models adverse events from electronic health records2-17 and using acute kidney injury-a common and potentially life-threatening condition18-as an exemplar. Our model was developed on a large, longitudinal dataset of electronic health records that cover diverse clinical environments, comprising 703,782 adult patients across 172 inpatient and 1,062 outpatient sites. Our model predicts 55.8% of all inpatient episodes of acute kidney injury, and 90.2% of all acute kidney injuries that required subsequent administration of dialysis, with a lead time of up to 48 h and a ratio of 2 false alerts for every true alert. In addition to predicting future acute kidney injury, our model provides confidence assessments and a list of the clinical features that are most salient to each prediction, alongside predicted future trajectories for clinically relevant blood tests9. Although the recognition and prompt treatment of acute kidney injury is known to be challenging, our approach may offer opportunities for identifying patients at risk within a time window that enables early treatment.

Building machines that learn and think for themselves.

We agree with Lake and colleagues on their list of "key ingredients" for building human-like intelligence, including the idea that model-based reasoning is essential. However, we favor an approach that centers on one additional ingredient: autonomy. In particular, we aim toward agents that can both build and exploit their own internal models, with minimal human hand engineering. We believe an approach centered on autonomous learning has the greatest chance of success as we scale toward real-world complexity, tackling domains for which ready-made formal models are not available. Here, we survey several important examples of the progress that has been made toward building autonomous agents with human-like abilities, and highlight some outstanding challenges.

STELA: a community-centred approach to norm elicitation for AI alignment.

Value alignment, the process of ensuring that artificial intelligence (AI) systems are aligned with human values and goals, is a critical issue in AI research. Existing scholarship has mainly studied how to encode moral values into agents to guide their behaviour. Less attention has been given to the normative questions of whose values and norms AI systems should be aligned with, and how these choices should be made. To tackle these questions, this paper presents the STELA process (SocioTEchnical Language agent Alignment), a methodology resting on sociotechnical traditions of participatory, inclusive, and community-centred processes. For STELA, we conduct a series of deliberative discussions with four historically underrepresented groups in the United States in order to understand their diverse priorities and concerns when interacting with AI systems. The results of our research suggest that community-centred deliberation on the outputs of large language models is a valuable tool for eliciting latent normative perspectives directly from differently situated groups. In addition to having the potential to engender an inclusive process that is robust to the needs of communities, this methodology can provide rich contextual insights for AI alignment.

Hyperglycemia as a risk factor for the development of retinopathy of prematurity.

BACKGROUND: Hyperglycemia has recently been described as a risk factor for the development of retinopathy of prematurity (ROP), a proliferative vascular disease of the retina that primarily affects premature infants. This study was to evaluate the relationship of hyperglycemia and the development of ROP in premature infants less than 32 weeks gestation. METHODS: This was a retrospective cohort study of all infants less than 32 weeks gestation from 2003-2007 who survived to discharge in our NICU. Demographic data including birthweight, gestational age, Apgar scores, method of delivery, antenatal steroid use, neonatal steroid use, and size for gestational age was collected for each infant. Episodes of sepsis, grade of intraventricular hemorrhage, presence of a patent ductus arteriosus, number of days on the ventilator, and stage of necrotizing enterocolitis were assessed as well as days of hyperglycemia, defined as number of days with whole blood glucose > 150 mg/dl. In addition, the highest stage of ROP was recorded for each infant. A Student's two tailed t-test or Fisher's exact test was performed to identify significant clinical risk factors associated with the development of ROP. From this univariate analysis, a multiple logistic regression was performed to determine the effect of hyperglycemia on the development of ROP, adjusting for significant clinical risk factors. Statistical analysis was performed using SAS v.9.2. RESULTS: Univariate analysis demonstrated that infants with ROP were of lower birthweight and gestational age, and were affected by a patent ductus arteriosus, neonatal sepsis, intraventricular hemorrhage, have significant lung disease and received postnatal glucocorticoid therapy. Infants with ROP experienced more days with hyperglycemia (7 vs. 2, p = < 0.0001). Using multiple logistic regression analysis to compare no ROP vs. all stages of ROP, gestational age (OR 0.745, 95% CI [0.634, 0.877], p = 0.0004), mean days of hyperglycemia (OR 1.073, 95% CI [1.004, 1.146], p = 0.04), and mean days receiving mechanical ventilation (OR 1.012, 95% CI [1.000, 1.025], p = 0.05) remained significantly associated with ROP after adjusting for other risk factors. CONCLUSION: Our data suggests that hyperglycemia is associated with the development of ROP in premature infants.

Learning skillful medium-range global weather forecasting.

Global medium-range weather forecasting is critical to decision-making across many social and economic domains. Traditional numerical weather prediction uses increased compute resources to improve forecast accuracy but does not directly use historical weather data to improve the underlying model. Here, we introduce GraphCast, a machine learning-based method trained directly from reanalysis data. It predicts hundreds of weather variables for the next 10 days at 0.25° resolution globally in under 1 minute. GraphCast significantly outperforms the most accurate operational deterministic systems on 90% of 1380 verification targets, and its forecasts support better severe event prediction, including tropical cyclone tracking, atmospheric rivers, and extreme temperatures. GraphCast is a key advance in accurate and efficient weather forecasting and helps realize the promise of machine learning for modeling complex dynamical systems.

Role of Near-infrared Spectroscopy in the Diagnosis and Assessment of Necrotizing Enterocolitis.

Near-infrared spectroscopy (NIRS) is a noninvasive, bedside diagnostic tool that could assist in the early diagnosis of necrotizing enterocolitis (NEC) in preterm neonates. NIRS is a safe and effective clinical tool in the neonatal intensive care unit to detect abnormal alterations in tissue perfusion and oxygenation. In addition, NIRS could also detect the complications of NEC, such as bowel necrosis and perforation. NEC is the most common gastrointestinal complication associated with preterm birth and critically ill infants. It is observed in 6-10% of preterm neonates, weighing below 1500 g, leading to considerable morbidity, mortality, and healthcare cost burden. The mortality rate ranges from 20 to 30%, highest in NEC infants undergoing surgery. NIRS is a promising diagnostic modality that could facilitate the early diagnosis of NEC and early detection of complications alone or with the imaging modalities.

Use of deep learning to develop continuous-risk models for adverse event prediction from electronic health records.

Early prediction of patient outcomes is important for targeting preventive care. This protocol describes a practical workflow for developing deep-learning risk models that can predict various clinical and operational outcomes from structured electronic health record (EHR) data. The protocol comprises five main stages: formal problem definition, data pre-processing, architecture selection, calibration and uncertainty, and generalizability evaluation. We have applied the workflow to four endpoints (acute kidney injury, mortality, length of stay and 30-day hospital readmission). The workflow can enable continuous (e.g., triggered every 6 h) and static (e.g., triggered at 24 h after admission) predictions. We also provide an open-source codebase that illustrates some key principles in EHR modeling. This protocol can be used by interdisciplinary teams with programming and clinical expertise to build deep-learning prediction models with alternate data sources and prediction tasks.

AI for social good: unlocking the opportunity for positive impact.

Advances in machine learning (ML) and artificial intelligence (AI) present an opportunity to build better tools and solutions to help address some of the world's most pressing challenges, and deliver positive social impact in accordance with the priorities outlined in the United Nations' 17 Sustainable Development Goals (SDGs). The AI for Social Good (AI4SG) movement aims to establish interdisciplinary partnerships centred around AI applications towards SDGs. We provide a set of guidelines for establishing successful long-term collaborations between AI researchers and application-domain experts, relate them to existing AI4SG projects and identify key opportunities for future AI applications targeted towards social good.

Genetic contributions to the development of retinopathy of prematurity.

There is growing support for the role of genetic factors in the development of retinopathy of prematurity (ROP), a serious visual morbidity resulting from preterm birth. We used both candidate gene and data-mining approaches to investigate the role of genetic polymorphisms in the development of ROP. Our study population consisted of 330 infants, less than 32 wk gestation, and their parents. We initially studied 24 single nucleotide polymorphisms (SNPs) in 11 candidate genes. Using a family-based analysis strategy, we found an association between SNPs in the EPAS1 gene and the development of ROP (p = 0.007). Logistic regression analysis showed three SNPs associated with development of ROP, two in the CFH gene (p = 0.01) and one in the EPAS1 gene (p = 0.001). Extending this analysis to include genotyping data from a larger genetic study of prematurity (455 SNPs in 153 genes), we found SNPs in five genes associated with the development of ROP: IHH (p = 0.003), AGTR1 (p = 0.005), TBX5 (p = 0.003), CETP (p = 0.004), and GP1BA (p = 0.005). Our data suggest that genetic risk factors contribute to the development of ROP.

Phosphoinositide turnover in human platelets is stimulated by albumin.

The effect of human serum albumin (HSA) on the hydrolysis of phosphatidylinositides in human platelets labeled with myo(3H)inositol was studied. Incubation of platelets with HSA (4 gm/dl) for 10 seconds increased IP2, and IP3, by 169% and 217% respectively. 93% of IP3 accumulated within the first 10 seconds. This effect was also shared by bovine serum albumin, although no changes in IP3 levels occurred with ovalbumin. All albumin species used induced 45Ca+2 release from platelets irrespective of its effect on IP3 accumulation. These findings indicate that albumin may function in biological systems by inducing intracellular signaling.

Catecholamine-induced cardiomyopathy.

OBJECTIVE: To review the pathogenesis as well as the clinical and laboratory features of catecholamine-induced cardiomyopathy associated with pheochromocytoma and other disorders and discuss the various treatment options available. METHODS: Materials used for this article were identified through MEDLINE, PubMed, and Google Scholar searches of the relevant literature from 1955 to the present. RESULTS: Catecholamines and their oxidation products cause a direct toxic effect on the myocardium. Catecholamines also exert a receptor-mediated effect on the myocardium. Catecholamine-mediated myocardial stunning has been implicated in the pathogenesis of stress-induced cardiomyopathy. Biopsy of the myocardium in patients with pheochromocytoma or those with stress-induced cardiomyopathy shows similar pathologic findings. The clinical features in pheochromocytoma-related cardiomyopathy include hypertension, dilated or hypertrophic cardiomyopathy, pulmonary edema due to cardiogenic and noncardiogenic factors, cardiac arrhythmias, and even cardiac arrest. Stress-related cardiomyopathy such as takotsubo cardiomyopathy occurs primarily in postmenopausal women. These patients may present with clinical features suggestive of an acute myocardial infarction or a hemodynamically compromised state. The definitive management of cardiomyopathy associated with pheochromocytoma includes medical treatment with alpha-adrenergic blockade, possibly along with angiotensin converting enzyme blockers and beta1-adrenergic receptor blockers, followed by excision of the tumor. Stress-induced cardiomyopathy is usually self-limiting; patients may require support with nonadrenergic inotropes. CONCLUSION: Recognition of catecholamine-induced cardiomyopathy, especially in patients with pheochromocytoma, before surgical treatment is important to minimize morbidity and mortality.

Pituitary and peripheral hormone responses to T3 administration during Antarctic residence.

Very little is known regarding hormonal adaptation in human subjects who are exposed to the extremes of temperature and light that are found in polar latitudes. We have previously reported a 50% elevation in the serum thyrotropin (TSH) response to thyrotropin-releasing hormone (TRH), a fall in serum total triiodothyronine (T3) and free T3 (fT3), and no change in serum total thyroxine (T4) or free T4 (fT4) after 42 wk of Antarctic cold exposure. To differentiate between central and peripheral mechanisms that may lead to these changes, we report the effect of sequentially increasing oral doses of T3 (Cytomel) on serum T3 and fT3 levels and on the resultant attenuation of the TSH response to TRH in nine men before, during, and after 42 wk residence in Antarctica. Serum T3 values basally and following the administration of 25, 50, and 75 micrograms/day of T3 were lower after 42 wk of cold exposure (151 +/- 4, 160 +/- 8, 189 +/- 10, and 222 +/- 14 ng/dl, respectively, compared with control values of 160 +/- 7, 178 +/- 7, 202 +/- 9, and 251 +/- 19 ng/dl, respectively, P less than 0.05). Likewise, the fT3 values measured after these three increasing T3 doses were also lower after 42 wk of cold exposure. The pituitary response to TRH was attenuated by each T3 regimen (48 +/- 6, 68 +/- 4, and 77 +/- 4% decreases in the control period), and this suppression was not different after 20 and 42 wk of Antarctic residence.(ABSTRACT TRUNCATED AT 250 WORDS)