Research Interests:
The post-pandemic world requires a renewed focus from service providers in ensuring that all customer segments receive the essential services (food, healthcare, housing, education, etc.) they need. Philanthropic service providers are... more
The post-pandemic world requires a renewed focus from service providers in ensuring that all customer segments receive the essential services (food, healthcare, housing, education, etc.) they need. Philanthropic service providers are unable to cope with the increased demands caused by the social, economic, and operational challenges induced by the pandemic. Customer self-selecting no-pay service strategies are becoming popular in various settings. Obtaining insights into how they can efficiently balance societal and financial goals is critical for a for-profit service provider. We develop and analyze a quantitative model of customer utilities, vertically-differentiated product assortment, pricing, and market size to understand how service providers can effectively use customer segmentation and serve the poor at the bottom of the pyramid. We identify conditions under which designing the service delivery to be accessible to the poor can simultaneously benefit the for-profit service provider, customers, and the entire society. Our work provides a framework to obtain operational, economic, and strategic insights into socially responsible service delivery strategies.
Research Interests:
Research Interests:
Research Interests:
We study an integrated airline schedule design and fleet assignment model for constructing schedules by simultaneously selecting from a pool of optional flights and assigning fleet types to these scheduled flights. This is a crucial... more
We study an integrated airline schedule design and fleet assignment model for constructing schedules by simultaneously selecting from a pool of optional flights and assigning fleet types to these scheduled flights. This is a crucial tactical decision that greatly influences airline profits. As passenger demand is often substitutable among available fare products (defined as a combination of an itinerary and a fare class) between the same origin–destination pair, we present an optimization approach that includes a passenger choice model for fare product selections. To tackle the formidable computational challenge of solving this large-scale network design problem, we propose a decomposition approach based on partitioning the flight network into smaller subnetworks by exploiting weak dependencies in network structure. The decomposition relies on a series of approximation analyses and a novel fare split problem to allocate optimally the fares of products that are shared by flights in different subnetworks. We present several reformulations that represent fleet assignment and schedule decisions and formally characterize their relative strengths. This gives rise to a new reformulation that is able to trade off strength and size flexibly. We conduct detailed computational experiments using two realistically sized airline instances to demonstrate the effectiveness of our approach. Under a simulated passenger booking environment with both perfect and imperfect forecasts, we show that the fleeting and scheduling decisions informed by our approach deliver significant and robust profit improvement over all benchmark implementations and previous models in the literature.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Introduction/Hypothesis: Occult ongoing hemorrhage without change in standard clinical parameters remains one of the most important diagnostic challenges in injured patients. We undertook development of a porcine model of subclinical... more
Introduction/Hypothesis: Occult ongoing hemorrhage without change in standard clinical parameters remains one of the most important diagnostic challenges in injured patients. We undertook development of a porcine model of subclinical hemorrhage in which there is a significant period of unchanged vital signs before progression to hypotension. Such a model may be used to study the performance of non-invasive measurement technologies and development of machine learning-based diagnostic algorithms. Methods: A prospective porcine study with alternating arterial and venous hemorrhage. Twentytwo female swine underwent hemorrhage from an intravascular catheter. Target hypotension was defined at a 15% decrease in mean arterial blood pressure (MAP), or a 20% increase in heart rate (HR). Standard physiological monitoring and enhanced non-invasive monitoring, including impedance spectroscopy/tomography/cardiography, and near-infrared spectroscopy, were utilized. Results: Bleeding in the range 0.20±0.13 cc/Kg min. resulted in 58±46 min. of hemorrhage before target. The mean volume of blood before target blood pressure or heart rate was 308 + 229 cc. The mean right atrial pressure was 9.8±5.5 mmHg at the start of the bleed, 7.4±5.9 mmHg at the end of the bleed, with a mean change of -2.35±2.40 mmHg after the bleed. The common pattern was a consistent decrease in right atrial pressure followed by a decrease in MAP and an increase in heart rate. Conclusions: It is possible to hemorrhage pigs at a rate which creates an initial period of occult subclinical progression. This period may be used to evaluate non-invasive measurement technologies and develop machine learning-based diagnostic algorithms. There is significant phenotypic variability in the total volume of blood loss associated with a defined change in MAP, and in the pattern of change for individual measurements. Multivariate machine learning-based approaches would be required for accurate diagnostic performance int his model.
Research Interests:
Dynamic contrast-enhanced fluorescence imaging (DCE-FI) classification of tissue viability in twelve adult patients undergoing below knee leg amputation is presented. During amputation and with the distal bone exposed, indocyanine green... more
Dynamic contrast-enhanced fluorescence imaging (DCE-FI) classification of tissue viability in twelve adult patients undergoing below knee leg amputation is presented. During amputation and with the distal bone exposed, indocyanine green contrast-enhanced images were acquired sequentially during baseline, following transverse osteotomy and following periosteal stripping, offering a uniquely well-controlled fluorescence dataset. An unsupervised classification machine leveraging 21 different spatiotemporal features was trained and evaluated by cross-validation in 3.5 million regions-of-interest obtained from 9 patients, demonstrating accurate stratification into normal, suspicious, and compromised regions. The machine learning (ML) approach also outperformed the standard method of using fluorescence intensity only to evaluate tissue perfusion by a two-fold increase in accuracy. The generalizability of the machine was evaluated in image series acquired in an additional three patients, confirming the stability of the model and ability to sort future patient image-sets into viability categories.
Research Interests:
Research Interests:
Problem definition: Electric vertical-takeoff-and-landing (eVTOL) vehicles enable urban aerial mobility (UAM). This paper optimizes the number, locations, and capacities of vertiports in UAM systems while capturing interdependencies... more
Problem definition: Electric vertical-takeoff-and-landing (eVTOL) vehicles enable urban aerial mobility (UAM). This paper optimizes the number, locations, and capacities of vertiports in UAM systems while capturing interdependencies between strategic vertiport deployment, tactical operations, and passenger demand. Academic/practical relevance: The model includes a “tractable part” (based on mixed-integer second-order conic optimization) and also a nonconvex demand function. Methodology: We develop an exact algorithm that approximates nonconvex functions with piecewise constant segments, iterating between a conservative model (which yields a feasible solution) and a relaxed model (which yields a solution guarantee). We propose an adaptive discretization scheme that converges to a global optimum—because of the relaxed model. Results: Our algorithm converges to a 1% optimality gap, dominating static discretization benchmarks in terms of solution quality, runtimes, and solution guarante...
Research Interests:
Accurate and reliable non-invasive monitoring of early systemic disease—such as ongoing hemorrhage, sepsis, and acute respiratory disease like COVID-19—is one of the largest unmet needs in biomedicine. An early alert to progression with... more
Accurate and reliable non-invasive monitoring of early systemic disease—such as ongoing hemorrhage, sepsis, and acute respiratory disease like COVID-19—is one of the largest unmet needs in biomedicine. An early alert to progression with high sensitivity and an acceptable false-positive rate would allow medical staff to risk-stratify patients, saving resources, lives, and in the context of pandemic disease, minimize staff exposure. Noninvasive technologies have thus far failed to produce a reliable early detection system, reflecting the limitation of uniplex approaches to describe complex pathophysiology. Our team, in collaboration with an STTR start-up, have developed an optico-impedance system combining near-infrared spectroscopy and electrical impedance tomography measured at three locations (thorax, abdomen, limb) together with machine learning methods to provide exceptional diagnostic performance in systemic disease. The optical portion consists of 6 pairs of time-multiplexed red and IR LEDs embedded in custom 3D-printed probes, which are each connected to the leg of a trifurcated fiber bundle, allowing measurement of three-location, two-distance broadband 550-950 nm spectra using a single commercial spectrometer. Data is demultiplexed and analyzed using derivative spectroscopy to quantify oxy/deoxyhemoglobin. Additional diagnostic signal was obtained from: impedance tomography and spectroscopy, ECG and plethysmography. In one of the largest porcine hemorrhage studies to date (n = 60), we demonstrate an 85% accuracy to detect a 2-3% blood volume loss. Preliminary results from 11 healthy human subjects undergoing lower body negative pressure (LBNP) challenge show a 95% accuracy in detecting 15-mmHg changes in pressure—an excellent surrogate for occult hemorrhage. Our system fills a critical need, including in the current pandemic, where clinicians struggle to predict which patients will deteriorate.
Research Interests: Engineering and Medicine
Introduction The ability to accurately detect hypotension in trauma patients at the earliest possible time is important in improving trauma outcomes. The earlier an accurate detection can be made, the more time is available to take... more
Introduction The ability to accurately detect hypotension in trauma patients at the earliest possible time is important in improving trauma outcomes. The earlier an accurate detection can be made, the more time is available to take corrective action. Currently, there is limited research on combining multiple physiological signals for an early detection of hemorrhagic shock. We studied the viability of early detection of hypotension based on multiple physiologic signals and machine learning methods. We explored proof of concept with a small (5 minutes) prediction window for application of machine learning tools and multiple physiologic signals to detecting hypotension. Materials and Methods Multivariate physiological signals from a preexisting dataset generated by an experimental hemorrhage model were employed. These experiments were conducted previously by another research group and the data made available publicly through a web portal. This dataset is among the few publicly availab...
Research Interests:
Research Interests:
Research Interests:
Abstract To schedule aircraft, assignments of fleet types to flights and of aircraft to routes must be determined. The former is known as the fleet assignment problem while the latter is known as the aircraft routing problem in the... more
Abstract To schedule aircraft, assignments of fleet types to flights and of aircraft to routes must be determined. The former is known as the fleet assignment problem while the latter is known as the aircraft routing problem in the literature. Aircraft routing is usually addressed as a feasibility problem whose solution is needed for constructing crew schedules. These problems are usually solved from 4 to 6 months before the day of operations. Therefore, there is limited information regarding each aircraft’s operational condition. The tail assignment problem, which has received limited attention in air transportation literature, is solved when additional information regarding operational conditions is revealed aiming at determining each aircraft’s route for the day of operations accounting for the originally planned aircraft routes and crew schedules. Therefore, it is a problem to be solved the day before operations. We propose a mathematical programming approach based on sequencing that captures all operational constraints and maintenance requirements while assignment costs are minimized. Computational experiments are based on realistic cases drawn from a Spanish airline, which features a network with more than 1000 flights and more than 100 aircraft.
Research Interests:
Airlines are known to compete for passengers, and airline profitability heavily depends on the ability to estimate passenger demand, which in turn depends on flight schedules, fares, and the number of seats available at each fare, across... more
Airlines are known to compete for passengers, and airline profitability heavily depends on the ability to estimate passenger demand, which in turn depends on flight schedules, fares, and the number of seats available at each fare, across all airlines. Interestingly, such competitive interactions and passenger substitution effects may not be limited to the planning stages. Existing regulations in some countries and regions impose monetary compensations to passengers in case of disruptions, altering the way they perceive the utility of other travel alternatives after the disruption starts. These passenger rights regulations may act as catalysts of passengers’ response to recovered schedules. Ignoring such passenger response behavior under operational disruptions may lead airlines to develop subpar recovery schedules. We develop a passenger response model and embed it into a novel integrated optimization approach that recovers airline schedules, aircraft, and passenger itineraries whil...
Research Interests:
Research Interests:
Federal, state and local aviation planners rely on air traffic forecasts for workforce staff planning (particularly for air traffic controllers), evaluation of current and future technological improvements at airports, planning of airport... more
Federal, state and local aviation planners rely on air traffic forecasts for workforce staff planning (particularly for air traffic controllers), evaluation of current and future technological improvements at airports, planning of airport capacity expansion, and evaluation of federal funding requests for airport infrastructure improvements. While most existing forecasting models are econometric or statistical in nature, incorporating a more behavioral understanding of airline competition into the forecasts and planning process represents a significant opportunity to improve their efficacy. With these possibilities in mind, we develop a two-stage game-theoretic model of airline capacity allocation decisions under competition. We first demonstrate desirable theoretical properties and computational tractability of our model, and then exploit them to develop solution algorithms with very fast convergence properties to enable rapid generation of forecasts, requiring only a few seconds of...
Introduction/Hypothesis: Ongoing hemorrhage without change in standard clinical parameters remains one of the most important diagnostic challenges in injured patients. We have developed a porcine model of occult hemorrhage in which there... more
Introduction/Hypothesis: Ongoing hemorrhage without change in standard clinical parameters remains one of the most important diagnostic challenges in injured patients. We have developed a porcine model of occult hemorrhage in which there is a significant period of unchanged vital signs before progression to frank hypotension/tachycardia, defined as a 15% drop in mean arterial pressure (MAP) or a 20% increase in heart rate. Such a model may be used to develop and evaluate the performance of innovative diagnostic modalities. We used this model to compare the cardiopulmonary reserve, defined as time/volume to hypotension/tachycardia, during primary versus secondary bleeding. Phenotypically, the physiologic response to recurrent hemorrhage may be blunted or enhanced as a function of processes such as preconditioning or two-hit injury models. Methods: A prospective porcine study with recurrent low volume hemorrhage. Ten swine underwent 30 controlled lowrate hemorrhages from an intravascu...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
This study is concerned with the determination of an optimal appointment schedule in an outpatient-inpatient hospital system where the inpatient exams can be cancelled based on certain rules while the outpatient exams cannot be cancelled.... more
This study is concerned with the determination of an optimal appointment schedule in an outpatient-inpatient hospital system where the inpatient exams can be cancelled based on certain rules while the outpatient exams cannot be cancelled. Stochastic programming models were formulated and solved to tackle the stochasticity in the procedure durations and patient arrival patterns. The first model, a two-stage stochastic programming model, is formulated to optimize the slot size. The second model further optimizes the inpatient block (IPB) placement and slot size simultaneously. A computational method is developed to solve the second optimization problem. A case study is conducted using the data from Magnetic Resonance Imaging (MRI) centers of Lahey Hospital and Medical Center (LHMC). The current schedule and the schedules obtained from the optimization models are evaluated and compared using simulation based on FlexSim Healthcare. Results indicate that the overall weighted cost can be reduced by 11.6% by optimizing the slot size and can be further reduced by an additional 12.6% by optimizing slot size and IPB placement simultaneously. Three commonly used sequencing rules (IPBEG, OPBEG, and a variant of ALTER rule) were also evaluated. The results showed that when optimization tools are not available, ALTER variant which evenly distributes the IPBs across the day has the best performance. Sensitivity analysis of weights for patient waiting time, machine idle time and exam cancellations further supports the superiority of ALTER variant sequencing rules compared to the other sequencing methods. A Pareto frontier was also developed and presented between patient waiting time and machine idle time to enable medical centers with different priorities to obtain solutions that accurately reflect their respective optimal tradeoffs. An extended optimization model was also developed to incorporate the emergency patient arrivals. The optimal schedules from the extended model show only minor differences compared to those from the original model, thus proving the robustness of the scheduling solutions obtained from our optimal models against the impacts of emergency patient arrivals. Timestamped operational data was analyzed to identify sources of uncertainty and delays. Stochastic programming models were developed to optimize slot size and inpatient block placement. A case study showed that the optimized schedules can reduce overall costs by 23%. Distributing inpatient and outpatient slots evenly throughout the day provides the best performance. A Pareto frontier was developed to allow practitioners to choose their own best tradeoffs between multiple objectives.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Supplemental material, AppendixB_online_supp for Simulation Analysis and Comparison of Point of Care Testing and Central Laboratory Testing by Reed Harder, Keji Wei, Vikrant Vaze and James E. Stahl in MDM Policy & Practice
With the soaring popularity of ride-hailing, the interdependence between transit ridership, ride-hailing ridership, and urban congestion motivates the following question: can public transit and ride-hailing coexist and thrive in a way... more
With the soaring popularity of ride-hailing, the interdependence between transit ridership, ride-hailing ridership, and urban congestion motivates the following question: can public transit and ride-hailing coexist and thrive in a way that enhances the urban transportation ecosystem as a whole? To answer this question, we develop a mathematical and computational framework that optimizes transit schedules while explicitly accounting for their impacts on road congestion and passengers’ mode choice between transit and ride-hailing. The problem is formulated as a mixed integer nonlinear program and solved using a bilevel decomposition algorithm. Based on computational case study experiments in New York City, our optimized transit schedules consistently lead to 0.4%–3% system-wide cost reduction. This amounts to rush-hour savings of millions of dollars per day while simultaneously reducing the costs to passengers and transportation service providers. These benefits are driven by a better...
Research Interests:
Introduction Early prediction of the acute hypotensive episode (AHE) in critically ill patients has the potential to improve outcomes. In this study, we apply different machine learning algorithms to the MIMIC III Physionet dataset,... more
Introduction Early prediction of the acute hypotensive episode (AHE) in critically ill patients has the potential to improve outcomes. In this study, we apply different machine learning algorithms to the MIMIC III Physionet dataset, containing more than 60,000 real-world intensive care unit records, to test commonly used machine learning technologies and compare their performances. Materials and Methods Five classification methods including K-nearest neighbor, logistic regression, support vector machine, random forest, and a deep learning method called long short-term memory are applied to predict an AHE 30 minutes in advance. An analysis comparing model performance when including versus excluding invasive features was conducted. To further study the pattern of the underlying mean arterial pressure (MAP), we apply a regression method to predict the continuous MAP values using linear regression over the next 60 minutes. Results Support vector machine yields the best performance in te...
Research Interests:
Introduction/Hypothesis: Occult ongoing hemorrhage below the threshold detectable clinically remains one of the most important diagnostic challenges in injured patients. Standard non-invasive technologies have been unable to detect... more
Introduction/Hypothesis: Occult ongoing hemorrhage below the threshold detectable clinically remains one of the most important diagnostic challenges in injured patients. Standard non-invasive technologies have been unable to detect ongoing hemorrhage before deterioration in vital signs.Using a porcine model of subclinical hemorrhage, we evaluated the performance of multi-location near-infrared spectroscopy (NIRS) to detect the presence or absence of low volume hemorrhage. Methods: A prospective porcine study with varying arterial and venous low-volume hemorrhage in the range of 0.125 – 0.165 cc/Kg min. 9 female swine underwent 21 separate bleeds from an intravascular catheter. Epochs of hemorrhage and no-hemorrhage were compared using standard physiological monitoring and near-infrared spectroscopy (NIRS) (FLAME-NIR, Ocean Optics, Largo, FL). NIRS was performed in the transcranial, trans-truncal, and trans-extremity locations. We first detected hemorrhages on a single subject by tra...
Research Interests:
The postpandemic world requires a renewed focus from service providers on ensuring that all customer segments receive the essential services (food, healthcare, housing, education, etc.) that they need. Philanthropic service providers are... more
The postpandemic world requires a renewed focus from service providers on ensuring that all customer segments receive the essential services (food, healthcare, housing, education, etc.) that they need. Philanthropic service providers are unable to cope with the increased demand caused by the social, economic, and operational challenges induced by the pandemic. For-profit service providers offering no-pay services to customers, allowing them to self-select a service option, is becoming a popular strategy in various settings. Obtaining insights into how to efficiently balance societal and financial goals is critical for a for-profit service provider. We develop and analyze a quantitative model of customer utilities, vertically differentiated product assortment, pricing, and market size to understand how service providers can effectively use customer segmentation and serve the poor in the lowest economic strata. We identify conditions under which designing the service delivery to be ac...
Research Interests:
The second UN Sustainable Development Goal establishes food security as a priority for governments, multilateral organizations, and NGOs. These institutions track national-level food security performance with an array of metrics and weigh... more
The second UN Sustainable Development Goal establishes food security as a priority for governments, multilateral organizations, and NGOs. These institutions track national-level food security performance with an array of metrics and weigh intervention options considering the leverage of many possible drivers. We studied the relationships between several candidate drivers and two response variables based on prominent measures of national food security: the 2019 Global Food Security Index (GFSI) and the Food Insecurity Experience Scale’s (FIES) estimate of the percentage of a nation’s population experiencing food security or mild food insecurity (FI
Research Interests:
Research Interests:
Research Interests:
Flight timetabling can greatly impact an airline’s operating profit, yet data-driven or model-based solutions to support it remain limited. Timetabling optimization is significantly complicated by two factors. First, it exhibits strong... more
Flight timetabling can greatly impact an airline’s operating profit, yet data-driven or model-based solutions to support it remain limited. Timetabling optimization is significantly complicated by two factors. First, it exhibits strong interdependencies with subsequent fleet assignment decisions of the airlines. Second, flights’ departure and arrival times are important determinants of passenger connection opportunities, of the attractiveness of each (nonstop or connecting) itinerary, and, in turn, of passengers’ booking decisions. Because of these complicating factors, most existing approaches rely on incremental timetabling. This paper introduces an original integrated optimization approach to comprehensive flight timetabling and fleet assignment under endogenous passenger choice. Passenger choice is captured by a discrete-choice generalized attraction model. The resulting optimization model is formulated as a mixed-integer linear program. This paper also proposes an original mult...
Research Interests:
Background. In response to demand for fast and efficient clinical testing, the use of point-of-care testing (POCT) has become increasingly common in the United States. However, studies of POCT implementation have found that adopting POCT... more
Background. In response to demand for fast and efficient clinical testing, the use of point-of-care testing (POCT) has become increasingly common in the United States. However, studies of POCT implementation have found that adopting POCT may not always be advantageous relative to centralized laboratory testing. Methods. We construct a simulation model of patient flow in an outpatient care setting to evaluate tradeoffs involved in POCT implementation across multiple dimensions, comparing measures of patient outcomes in varying clinical scenarios, testing regimes, and patient conditions. Results. We find that POCT can significantly reduce clinical time for patients, as compared to traditional testing regimes, in settings where clinic and central testing areas are far apart. However, as distance from clinic to central testing area decreased, POCT advantage over central laboratory testing also decreased, in terms of time in the clinical system and estimated subsequent productivity loss....
Research Interests:
Research Interests:
In the absence of opportunities for capacity expansion or operational enhancements, air traffic congestion mitigation may require scheduling interventions to control overcapacity scheduling at busy airports. Previous research has shown... more
In the absence of opportunities for capacity expansion or operational enhancements, air traffic congestion mitigation may require scheduling interventions to control overcapacity scheduling at busy airports. Previous research has shown that large delay reductions could be achieved through comparatively small changes in the schedule of flights. While existing approaches have focused on minimizing the overall impact across the airlines, this paper designs, optimizes, and assesses a novel approach for airport scheduling interventions that incorporates interairline equity objectives. It relies on a multilevel modeling architecture based on on-time performance (i.e., mitigating airport congestion), efficiency (i.e., meeting airline scheduling preferences), and equity (i.e., balancing scheduling adjustments fairly among the airlines) objectives, subject to scheduling and network connectivity constraints. Theoretical results show that, under some scheduling conditions, equity and efficienc...
Research Interests:
We propose, optimize, and validate a methodological framework for estimating the extent of the crew-propagated delays and disruptions (CPDDs). We identify the factors that influence the extent of the CPDDs and incorporate them into a... more
We propose, optimize, and validate a methodological framework for estimating the extent of the crew-propagated delays and disruptions (CPDDs). We identify the factors that influence the extent of the CPDDs and incorporate them into a robust crew-scheduling model. We develop a fast heuristic approach for solving the inverse of this robust crew-scheduling problem to generate crew schedules that are similar to real-world crew-scheduling samples. We develop a sequence of exact and heuristic techniques to quickly solve the forward problem within a small optimality gap for network sizes that are among the largest in robust crew-scheduling literature. Computational results using four large real-world airline networks demonstrate that the crew schedules produced by our approach generate propagation patterns similar to those observed in the real world. Extensive out-of-sample validation tests indicate that the parameters calibrated for one network perform reasonably well for other networks. ...
Research Interests:
Airlines make capacity and fare decisions in a competitive environment. Capacity decisions, encompassing decisions about frequency of service and seats-per-flight, affect both the operating costs and revenues of airlines. These decisions... more
Airlines make capacity and fare decisions in a competitive environment. Capacity decisions, encompassing decisions about frequency of service and seats-per-flight, affect both the operating costs and revenues of airlines. These decisions have significant implications for the performance of the air transportation system as a whole. Capacity and fare decisions of different airlines are interdependent, both serving as tools in an airlines competitive arsenal. This interdependency motivates a game theoretic approach to modeling the decision process. Capacity (especially frequency) decisions are typically made months in advance of flight departure, with only an approximate knowledge of what fares will be, while fare decisions are made weeks to minutes ahead of flight departure. Several studies have stressed the need to develop two-stage game theoretic models to account for the sequential nature of these decisions, but there are very few analytical, computational, or empirical results ava...