1,721,002 research outputs found
Advancing Sustainability Research Using Mathematical Programming Techniques
No full textThe central thesis of this dissertation is that mathematical programming techniques can be successfully applied to gain novel insight into problems in the energy sector related to building assessment systems and wind farms. We focus on the novel application of mathematical programming to the Leadership in Energy and Environmental Design (LEED) rating system, and wind farm layout optimization (WFLO).
In the first part of this dissertation, we use an inverse optimization technique to assess and propose improvements to the LEED rating system for buildings. Due to the large dimensionality of the inverse optimization problem, we develop an approximation to improve tractability, and provide numerical evidence to validate the approximation method. Based on the results from our inverse model, we perform a statistical analysis and determine that some of the valuation of LEED credits by building designers may be based on specific building attributes not previously considered.
Second, we develop a new mathematical programming approach for wind farm layout optimization. We use Jensen's wake decay model to represent multi-turbine wake effect, develop mixed-integer linear and quadratic optimization formulations, and apply our formulations to several example layouts cases. Compared to previous approaches, our models generate layouts that are more symmetric and produce more power. We also develop a heuristic bounding policy for a special class of quadratic integer programs to speed up computational times, useful in this case, and potentially other applications.
Finally, in the last part of this dissertation, we develop a comprehensive WFLO framework that simultaneously takes into account wake effect, sound regulations, turbine infrastructure, and landowner compensation. A financial analysis is performed to determine a common measure of comparison for these different factors, and we develop various mixed-integer linear formulations by combinations of specific factors. We perform six different experiments using our formulations to demonstrate the value of our framework, and from the results we determine the factors impact on the optimal positioning of turbines, infrastructure, and landowner compensation.Ph.D
Robust Optimization on Healthcare Referral Networks: Engaging Private Providers in Low- and Middle-income Countries
Full text linkPrivate practitioners are a ubiquitous part of healthcare systems in low- and middle-income countries. In order to make any systemic improvements to the healthcare system, the private sector must be included through engagement and incentivization. We develop a methodology that can be leveraged by private practitioner engagement programs to determine the optimal set of practitioners to engage with while considering uncertainty in the network of patient referrals, where patients are first diagnosed by a general practitioner (GP) and then referred to a specialist for additional diagnoses. We demonstrate our modelling framework and analyze policies for practitioner selection using a real-world dataset from an organization that works to improve tuberculosis diagnosis in Mumbai. We find that the optimal policy is to use a two-phase heuristic that involves first selecting GPs with the highest patient loads and then using the remaining budget to select specialists in decreasing order of patient load.M.A.S
Optimal Defibrillator Placement for In-Hospital Cardiac Arrest
Full text linkIn-hospital cardiac arrest (IHCA) affects over 300,000 adults across North America every year. Rapid treatment via defibrillation is an effective treatment for many IHCA victims, but the placement of in-hospital defibrillators is typically arbitrary, which may negatively impact response and patient outcomes. We conducted a descriptive analysis of IHCA incidents at St. Michael's Hospital in Toronto, Ontario, and then developed two mathematical optimization models that determined optimal placement locations for defibrillators based on historical IHCA locations such that the mean and worst-case IHCA-to-defibrillator distances were minimized respectively. We found that optimized defibrillator locations were able to significantly reduce mean IHCA-to-defibrillator distances, while worst-case distances were not significantly different compared to current placement configurations. We conducted predictive analysis on IHCA response metrics and patient outcomes but were generally unable to find any significant associations. We end with suggestions for future studies and policy implications to improve IHCA response.M.A.S
Sensitivity and Stability Analysis for Inverse Optimization with Applications in Intensity-modulated Radiation Therapy
Full text linkInverse optimization aims to use observed data to understand the dynamics of a system by estimating the parameters of the underlying optimization problem. In this dissertation, we first develop an inverse model and investigate its properties for the cases where single and multiple observed data points are available. Next, we study the sensitivity and stability of the inverse model with respect to changes in the model parameters and perturbations in data. In particular, we illustrate the sensitivity of the inverse model with respect to the type of penalty functions used in the objective. Furthermore, we show that perturbations in the observed data can dramatically alter the outcome of the inverse model. We subsequently generalize our model to improve its stability with respect to perturbations in the input. Finally, we perform extensive simulations on synthetic and Intensity-Modulated Radiation Therapy (IMRT) data to validate the effectiveness of the generalized inverse model.M.A.S.2016-11-25 00:00:0
Sensitivity and Stability Analysis for Inverse Optimization with Applications in Intensity-modulated Radiation Therapy
No full textInverse optimization aims to use observed data to understand the dynamics of a system by estimating the parameters of the underlying optimization problem. In this dissertation, we first develop an inverse model and investigate its properties for the cases where single and multiple observed data points are available. Next, we study the sensitivity and stability of the inverse model with respect to changes in the model parameters and perturbations in data. In particular, we illustrate the sensitivity of the inverse model with respect to the type of penalty functions used in the objective. Furthermore, we show that perturbations in the observed data can dramatically alter the outcome of the inverse model. We subsequently generalize our model to improve its stability with respect to perturbations in the input. Finally, we perform extensive simulations on synthetic and Intensity-Modulated Radiation Therapy (IMRT) data to validate the effectiveness of the generalized inverse model.M.A.S.2016-11-25 00:00:0
Advancing Sustainability Research Using Mathematical Programming Techniques
Full text linkThe central thesis of this dissertation is that mathematical programming techniques can be successfully applied to gain novel insight into problems in the energy sector related to building assessment systems and wind farms. We focus on the novel application of mathematical programming to the Leadership in Energy and Environmental Design (LEED) rating system, and wind farm layout optimization (WFLO).
In the first part of this dissertation, we use an inverse optimization technique to assess and propose improvements to the LEED rating system for buildings. Due to the large dimensionality of the inverse optimization problem, we develop an approximation to improve tractability, and provide numerical evidence to validate the approximation method. Based on the results from our inverse model, we perform a statistical analysis and determine that some of the valuation of LEED credits by building designers may be based on specific building attributes not previously considered.
Second, we develop a new mathematical programming approach for wind farm layout optimization. We use Jensen's wake decay model to represent multi-turbine wake effect, develop mixed-integer linear and quadratic optimization formulations, and apply our formulations to several example layouts cases. Compared to previous approaches, our models generate layouts that are more symmetric and produce more power. We also develop a heuristic bounding policy for a special class of quadratic integer programs to speed up computational times, useful in this case, and potentially other applications.
Finally, in the last part of this dissertation, we develop a comprehensive WFLO framework that simultaneously takes into account wake effect, sound regulations, turbine infrastructure, and landowner compensation. A financial analysis is performed to determine a common measure of comparison for these different factors, and we develop various mixed-integer linear formulations by combinations of specific factors. We perform six different experiments using our formulations to demonstrate the value of our framework, and from the results we determine the factors impact on the optimal positioning of turbines, infrastructure, and landowner compensation.Ph.D
Inverse linear optimization for the recovery of constraint parameters in robust and non-robust problems
No full textMost inverse optimization models impute unspecified parameters of an objective function to make an observed solution optimal for a given optimization problem. In this thesis, we propose two approaches to impute unspecified left-hand-side constraint coefficients in addition to a cost vector for a given linear optimization problem. The ďŹ rst approach minimally perturbs prior estimates of the unspecified parameters to satisfy strong duality, while the second identifies parameters minimizing the duality gap, if it is not possible to satisfy the optimality conditions exactly. We apply these two approaches to the general linear optimization problem. We also use them to impute unspecified parameters of the uncertainty set for robust linear optimization problems under interval and cardinality constrained uncertainty. Each inverse optimization model we propose is nonconvex, but we show that a globally optimal solution can be obtained by solving a ďŹ nite number of tractable optimization problems.M.A.S
A Physiology-based Mathematical Model for the Selection of Appropriate Ventilator Controls for Lung and Diaphragm Protection
Full text linkMechanical ventilation is used to sustain respiratory function in patients with acute respiratory failure. To aid clinicians in consistently selecting lung- and diaphragm-protective ventilation settings, a physiology-based decision support system is needed. To form the foundation of such a system, the Lung- and Diaphragm-Protective Ventilation (LDPV) model has been developed. It centers around respiratory drive and uses patient-specific parameters as inputs and outputs predictions of a patient's transpulmonary and esophageal driving pressures as well as their blood pH. Ventilation and sedation parameters were demonstrated to modulate the model outputs in accordance with what is currently known in literature. The model was also proven to be able to provide robust predictions of model outputs for patient parameters with realistic variability. The usability of the model was improved by applying machine learning methods to the system inputs, allowing several patient-specific input parameters to be inferred.M.A.S
Geographic Determinants of Atrial Fibrillation Outcomes
No full textAtrial fibrillation management requires longitudinal care and living further from hospitals may be associated with worse outcomes. We examined associations between atrial fibrillation outcomes and hospital distance.
We conducted a retrospective study of patients diagnosed with atrial fibrillation in an emergency department between 2007-2016 in Ontario, Canada using datasets from Ontario Health. Distance between patients’ residence and hospitals was compared to various outcomes including death, stroke, bleeding, and ablation. For ablations, patients living near specific electrophysiology centers were identified to explore physician practice patterns. Multivariate logistic regression was utilized.
There was substantial regional heterogeneity in atrial fibrillation outcomes. Distance to hospitals was generally not associated with outcomes. There were significant differences in ablation rates for patients living around certain electrophysiology centers.
Greater distance to hospitals was not strongly associated with increased risk of adverse outcomes. However, electrophysiology center practice patterns may significantly affect a patient’s likelihood of receiving ablation.M.A.S
Geospatial Analytics and Mathematical Optimization in Pre-Hospital Resuscitation
Full text linkResuscitation is the process of treating critically ill patients who are at imminent risk of death; as such, treatment of resuscitation patients is time-critical and must begin as soon as possible. This thesis aims to use data analytics and mathematical optimization approaches to quantify the benefit of proposed interventions aimed at improving resuscitation response in the pre-hospital setting.The first part of this thesis considers the analysis and modelling of static resources that can be retrieved by laypersons to the patient’s location. In Chapter 2, we present a narrative-driven review of the literature surrounding public access automated
external defibrillators (AEDs) as an intervention for out-of-hospital cardiac arrest (OHCA) and provide commentary on the state of public access defibrillation in Scotland. In Chapter 3, we analyze the spatial coverage and socioeconomic equity
of public access AEDs for nearby OHCAs in Scotland. Compared to existing AED locations, which are neither efficiently nor equitably placed, optimization-driven AED placement can lead to effective and socioeconomically equitable coverage of OHCAs. In Chapter 4, we consider various placement strategies for publicly accessible naloxone kits for nearby opioid poisonings in Vancouver and find that an optimization-driven placement strategy leads to the most efficient coverage of opioid poisonings.
The second part of this thesis considers the modelling of mobile resources which are actively dispatched to the patient’s location. In Chapter 5, we analyze the effect of varying base locations for a proposed drone-based AED delivery network for OHCAs in southern Vancouver Island and find modest differences in an otherwise substantial improvement to OHCA response times. In Chapter 6, we model the potential addition of vertical takeoff-and-landing air ambulances dedicated to OHCA response in Paris and Vancouver, and find dramatic improvements in OHCA response times, particularly at the advanced life support level. In Chapter 7, we quantify the accessibility of various strategies to deploy crews administering extracorporeal cardiopulmonary resuscitation for certain OHCA patients in Scotland and estimate the resulting increase
in survival.
Overall, our findings show the importance and benefit of centralized, data-driven decision making when developing systems-based interventions aimed at optimizing response to the patient.Ph.D
- …
