Camila C. S. Caiado

Dynamic prediction modeling approaches for cardiac surgery

Background— The calibration of several cardiac clinical prediction models has deteriorated over time. We compare different model fitting approaches for in-hospital mortality after cardiac surgery that adjust for cross-sectional case mix in a heterogeneous patient population. Methods and Results— Data from >300 000 consecutive cardiac surgery procedures performed at all National Health Service and some private hospitals in England and Wales between April 2001 and March 2011 were extracted from the National Institute for Cardiovascular Outcomes Research clinical registry. The study outcome was in-hospital mortality. Model approaches included not updating, periodic refitting, rolling window, and dynamic logistic regression. Covariate adjustment was made in each model using variables included in the logistic European System for Cardiac Operative Risk Evaluation model. The association between in-hospital mortality and some variables changed with time. Notably, the intercept coefficient has been steadily decreasing during the study period, consistent with decreasing observed mortality. Some risk factors, such as operative urgency and postinfarct ventricular septal defect, have been relatively stable over time, whereas other risk factors, such as left ventricular function and surgery on the thoracic aorta, have been associated with lower risk relative to the static model. Conclusions— Dynamic models or periodic model refitting is necessary to counteract calibration drift. A dynamic modeling framework that uses contemporary and available historic data can provide a continuously smooth update mechanism that also allows for inferences to be made on individual risk factors. Better models that withstand the effects of time give advantages for governance, quality improvement, and patient-level decision making.

Social tipping points and Earth systems dynamics

Recently, Early Warning Signals (EWS) have been developed to predict tipping points in Earth Systems. This discussion highlights the potential to apply EWS to human social and economic systems, which may also undergo similar critical transitions. Social tipping points are particularly difficult to predict, however, and the current formulation of EWS, based on a physical system analogy, may be insufficient. As an alternative set of EWS for social systems, we join with other authors encouraging a focus on heterogeneity, connectivity through social networks and individual thresholds to change.

Fitness landscapes among many options under social influence

Cultural learning represents a novel problem in that an optimal decision depends not only on intrinsic utility of the decision/behavior but also on transparency of costs and benefits, the degree of social versus individual learning, and the relative popularity of each possible choice in a population. In terms of a fitness-landscape function, this recursive relationship means that multiple equilibria can exist. Here we use discrete-choice theory to construct a fitness-landscape function for a bi-axial decision-making map that plots the magnitude of social influence in the learning process against the costs and payoffs of decisions. Specifically, we use econometric and statistical methods to estimate not only the fitness function but also movements along the map axes. To search for these equilibria, we employ a hill-climbing algorithm that leads to the expected values of optimal decisions, which we define as peaks on the fitness landscape. We illustrate how estimation of a measure of transparency, a measure of social influence, and the associated fitness landscape can be accomplished using panel data sets.

Markov Chain Monte Carlo inversion of temperature and salinity structure of an internal solitary wave packet from marine seismic data.

Marine seismic reflection technique is used to observe the strong ocean dynamic process of nonlinear internal solitary waves (ISWs or solitons) in the near-surface water. Analysis of ISWs is problematical because of their transient nature and limitations of classical physical oceanography methods. This work explores a Markov Chain Monte Carlo (MCMC) approach to recover the temperature and salinity of ISW field using the seismic reflectivity data and in situ hydrographic data. The MCMC approach is designed to directly sample the posterior probability distributions of temperature and salinity which are the solutions of the system under investigation. The principle improvement is the capability of incorporating uncertainties in observations and prior models which then provide quantified uncertainties in the output model parameters. We tested the MCMC approach on two acoustic reflectivity data sets one synthesized from a CTD cast and the other derived from multichannel seismic reflections. This method finds the solutions faithfully within the significantly narrowed confidence intervals from the provided priors. Combined with a low frequency initial model interpreted from seismic horizons of ISWs, the MCMC method is used to compute the finescale temperature, salinity, acoustic velocity, and density of ISW field. The statistically derived results are equivalent to the conventional linearized inversion method. However, the former provides us the quantified uncertainties of the temperature and salinity along the whole section whilst the latter does not. These results are the first time ISWs have been mapped with sufficient detail for further analysis of their dynamic properties.

Bayesian Strategies to Assess Uncertainty in Velocity Models

Abstract. Quantifying uncertainty in models derived from observed seismic datais a major issue. In this research we examine the geological structure of the sub-surface using controlledsourceseismology which gives the data in time and thedistance between the acoustic source and thereceiver. Inversion tools exist tomap these data into a depth model, but a full exploration of the uncertainty ofthe model is rarely done because robust strategies do not exist for large non-linearcomplex systems. There are two principal sources of uncertainty: the rst comesfrom the input data which is noisy andband-limited; the second is from the modelparameterisation and forward algorithm which approximate the physics to makethe problem tractable. To address these issues we propose a Bayesian approachusing the Metropolis-Hastings algorithm.Keywords: Gaussian Processes, Metropolis-Hastings algorithm, Seismology, Veloc-ity Modelling 1 Introduction Seismic reection surveying is the principal means of investigating the geological struc-ture of the Earth to depths of about 30km. Acoustic energy from a source on thesurface propagates into the Earth and is partially reected due to a change ofacousticimpedancebetween di erent layers of rock. The amplitude and phase of the reectedenergy is dependent on the change of the elastic parameters at the interface, namelyp-wavevelocity,s-wavevelocity and density (see Glossary after Section 8).In the eld, a series of acoustic sources are red into the sub-surface at knownpoints along the seismic pro le to be investigated and each source is recorded by anarray ofreceivers. Knowledge of the position of the source andreceiverlocations isused during processing to collect together recordedtraceswith the same geometricalmid-point between the source andreceivercalled acommon midpoint (CMP)gather.A CMP gather consists of a number of traces with varying source-receiver o sets. InFigure1, we can see a simple example of a CMP gather with only two traces: therecorded peak on the rst trace of Figure1ashows how long it took for the signal totravel from source S

Evaluating reproductive decisions as discrete choices under social influence.

Discrete choice, coupled with social influence, plays a significant role in evolutionary studies of human fertility, as investigators explore how and why reproductive decisions are made. We have previously proposed that the relative magnitude of social influence can be compared against the transparency of pay-off, also known as the transparency of a decision, through a heuristic diagram that maps decision-making along two axes. The horizontal axis represents the degree to which an agent makes a decision individually versus one that is socially influenced, and the vertical axis represents the degree to which there is transparency in the pay-offs and risks associated with the decision the agent makes. Having previously parametrized the functions that underlie the diagram, we detail here how our estimation methods can be applied to real-world datasets concerning sexual health and contraception.

Validation of Three Postoperative Risk Prediction Models for Intensive Care Unit Mortality after Cardiac Surgery

Background Several cardiac surgery risk prediction models based on postoperative data have been developed. However, unlike preoperative cardiac surgery risk prediction models, postoperative models are rarely externally validated or utilized by clinicians. The objective of this study was to externally validate three postoperative risk prediction models for intensive care unit (ICU) mortality after cardiac surgery. Methods The logistic Cardiac Surgery Scores (logCASUS), Rapid Clinical Evaluation (RACE), and Sequential Organ Failure Assessment (SOFA) scores were calculated over the first 7 postoperative days for consecutive adult cardiac surgery patients between January 2013 and May 2015. Model discrimination was assessed using receiver operating characteristic curve analyses. Calibration was assessed using the Hosmer‐Lemeshow (HL) test, calibration plots, and observed to expected ratios. Recalibration of the models was performed. Results A total of 2255 patients were included with an ICU mortality rate of 1.8%. Discrimination for all three models on each postoperative day was good with areas under the receiver operating characteristic curve of >0.8. Generally, RACE and logCASUS had better discrimination than SOFA. Calibration of the RACE score was better than logCASUS, but ratios of observed to expected mortality for both were generally <0.65. Locally recalibrated SOFA, logCASUS and RACE models all performed well. Conclusion All three models demonstrated good discrimination for the first 7 days after cardiac surgery. After recalibration, logCASUS and RACE scores appear to be most useful for daily risk prediction after cardiac surgery. If appropriately calibrated, postoperative cardiac surgery risk prediction models have the potential to be useful tools after cardiac surgery.

The KDIGO acute kidney injury guidelines for cardiac surgery patients in critical care: a validation study

BackgroundThe Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury (AKI) guidelines assign the same stage of AKI to patients whether they fulfil urine output criteria, serum creatinine criteria or both criteria for that stage. This study explores the validity of the KDIGO guidelines as a tool to stratify the risk of adverse outcomes in cardiac surgery patients.MethodsProspective data from consecutive adult patients admitted to the cardiac intensive care unit (CICU) following cardiac surgery between January 2013 and May 2015 were analysed. Patients were assigned to groups based on the criteria they met for each stage of AKI according to the KDIGO guidelines. Short and mid-term outcomes were compared between these groups.ResultsA total of 2267 patients were included with 772 meeting criteria for AKI-1 and 222 meeting criteria for AKI-2. After multivariable adjustment, patients meeting both urine output and creatinine criteria for AKI-1 were more likely to experience prolonged CICU stay (OR 4.9, 95%CI 3.3–7.4, p < 0.01) and more likely to require renal replacement therapy (OR 10.5, 95%CI 5.5–21.9, p < 0.01) than those meeting only the AKI-1 urine output criterion. Patients meeting both urine output and creatinine criteria for AKI-1 were at an increased risk of mid-term mortality compared to those diagnosed with AKI-1 by urine output alone (HR 2.8, 95%CI 1.6–4.8, p < 0.01). Patients meeting both urine output and creatinine criteria for AKI-2 were more likely to experience prolonged CICU stay (OR 16.0, 95%CI 3.2–292.0, p < 0.01) or require RRT (OR 11.0, 95%CI 4.2–30.9, p < 0.01) than those meeting only the urine output criterion. Patients meeting both urine output and creatinine criteria for AKI-2 were at a significantly increased risk of mid-term mortality compared to those diagnosed with AKI-2 by urine output alone (HR 3.6, 95%CI 1.4–9.3, p < 0.01).ConclusionsPatients diagnosed with the same stage of AKI by different KDIGO criteria following cardiac surgery have significantly different short and mid-term outcomes. The KDIGO criteria need to be revisited before they can be used to stratify reliably the severity of AKI in cardiac surgery patients. The utility of the criteria also needs to be explored in other settings.

Improved linear programming methods for checking avoiding sure loss

Abstract We review the simplex method and two interior-point methods (the affine scaling and the primal-dual) for solving linear programming problems for checking avoiding sure loss, and propose novel improvements. We exploit the structure of these problems to reduce their size. We also present an extra stopping criterion, and direct ways to calculate feasible starting points in almost all cases. For benchmarking, we present algorithms for generating random sets of desirable gambles that either avoid or do not avoid sure loss. We test our improvements on these linear programming methods by measuring the computational time on these generated sets. We assess the relative performance of the three methods as a function of the number of desirable gambles and the number of outcomes. Overall, the affine scaling and primal-dual methods benefit from the improvements, and they both outperform the simplex method in most scenarios. We conclude that the simplex method is not a good choice for checking avoiding sure loss. If problems are small, then there is no tangible difference in performance between all methods. For large problems, our improved primal-dual method performs at least three times faster than any of the other methods.

Market Structure with Interacting Consumers

Economic theory has developed a typology of markets which depends upon the number of firms which are present. Much of the literature, however, is set in the context of a given market structure, with the consequences of the structure being explored. Considerably less attention is paid to the process by which any particular structure emerges. In this paper, we examine the process of how different types of market structure emerge in new product markets, and in particular on markets which are primarily web-based. A wide range of outcome is possible. But the uncertainty of outcome of the evolution of market shares in such markets is based, not on the various strategies of the firms. Instead, it is inherent in the behavioral rule of choice used by consumers. We examine the consequences, for the market structure which emerges, of a realistic behavioral rule for consumer choice in new product markets. The rule has been applied in a range of different empirical contexts. It is essentially based on the model of genetic drift pioneered by Sewall Wright in the inter-war period. We identify the parameter ranges in the model in which the Herfindahl-Hirschman Index is likely to fall within the ranges identified by the US Department of Justice - unconcentrated markets; moderately concentrated markets and highly concentrated markets.