arXiv subjects

Varun Ramamohan

Publications and source records attributed to Varun Ramamohan.

12 recordsLinked to original sources

arXiv2026

Healthcare Facility Assignment Using Real-Time Length-of-Stay Predictions: Queuing-Theoretic and Simulation-driven Machine Learning Approaches

Longer stays at healthcare facilities, driven by uncertain patient load, inefficient patient flow, and lack of real-time information about medical care, pose significant challenges for patients and healthcare providers. Providing patients with estimates of their expected real-time length of stay (RT-LOS), generated as a function of the operational state of the healthcare facility at their anticipated time of arrival (as opposed to estimates of average LOS), can help them make informed decisions regarding which facility to visit within a network. In this study, we develop a healthcare facility assignment (HFA) algorithm that assigns healthcare facilities to patients using RT-LOS predictions at facilities within the network of interest. We describe the generation of RT-LOS predictions via two methodologies: (a) an analytical queuing-theoretic approach, and (b) a hybrid simulation-driven machine learning approach. Because RT-LOS predictors are highly specific to the queuing system in question, we illustrate the development of RT-LOS predictors using both approaches by considering the outpatient experience at primary health centers. Via computational experiments, we compare outcomes from the implementation of the RT-HFA algorithm with both RT-LOS predictors to the case where patients visit the facility of their choice. Computational experiments also indicated that the RT-HFA algorithm substantially reduced patient wait times and LOS at congested facilities and led to more equitable utilization of medical resources at facilities across the network. Finally, we show numerically that the effectiveness of the RT-HFA algorithm in improving outcomes is contingent on the level of compliance with the assignment decision.

eess.SY↗

arXiv2025

A Framework for Predicting Runtime Savings from Discrete-Event Simulation Model Simplification Operations

Abstraction or substitution and aggregation are the most widely used simulation model simplification operations. Abstraction involves replacing subsystems within a discrete-event simulation (DES) with one or more quantities - typically random variables - representing the lengths of stay in the subsystems(s) in question to create a `simplified' system comprising only of subsystems of interest to the analysis at hand. Aggregation involves replacing more than one subsystem of the original `parent' simulation with a single subsystem. However, the model simplification process itself can be expensive, in terms of the computational runtime and effort required to collect the data required to estimate the distributions of the length of stay variables, the distribution-fitting process, and testing and validation of the simplified model. Moreover, the savings in simulation runtime that the simplification process yields is \textit{a priori} unknown to the modeller. In this context, a method that predicts the runtime savings (RS) from DES model simplification operations before their execution - at the conceptualisation stage of the simplified model development process - may help judge whether its development is indeed worth undertaking. In this paper, we present a queueing-theoretic framework for the prediction of RS from model simplification operations. Our framework is applicable for DES models comprising $M/M/, M/G/ \text{ and } G/G/$ subsystems. The performance of the RS prediction framework is demonstrated using multiple computational experiments. Our proposed framework contributes to the literature around DES model complexity and more broadly to DES runtime prediction.

eess.SY↗

arXiv2024

Impact of Transmission Dynamics and Treatment Uptake, Frequency and Timing on the Cost-effectiveness of Directly Acting Antivirals for Hepatitis C Virus Infection

Cost-effectiveness analyses, based on decision-analytic models of disease progression and treatment, are routinely used to assess the economic value of a new intervention and consequently inform reimbursement decisions for the intervention. Many decision-analytic models developed to assess the economic value of highly effective directly acting antiviral (DAA) treatments for the hepatitis C virus (HCV) infection do not incorporate the transmission dynamics of HCV, accounting for which is required to estimate the number of downstream infections prevented by curing an infection. In this study, we develop and validate a comprehensive agent-based simulation (ABS) model of HCV transmission dynamics in the Indian context and use it to: (a) quantify the extent to which the cost-effectiveness of a DAA is underestimated - as a function of its uptake rate - if disease transmission dynamics are not considered in a cost-effectiveness analysis model; and (b) quantify the impact of the frequency and timing of treatment with DAAs, also as a function of their uptake rate, within a disease surveillance period on its cost-effectiveness.

eess.SY↗

arXiv2022

Discrete Simulation Optimization for Tuning Machine Learning Method Hyperparameters

Machine learning (ML) methods are used in most technical areas such as image recognition, product recommendation, financial analysis, medical diagnosis, and predictive maintenance. An important aspect of implementing ML methods involves controlling the learning process for the ML method so as to maximize the performance of the method under consideration. Hyperparameter tuning is the process of selecting a suitable set of ML method parameters that control its learning process. In this work, we demonstrate the use of discrete simulation optimization methods such as ranking and selection (R&S) and random search for identifying a hyperparameter set that maximizes the performance of a ML method. Specifically, we use the KN R&S method and the stochastic ruler random search method and one of its variations for this purpose. We also construct the theoretical basis for applying the KN method, which determines the optimal solution with a statistical guarantee via solution space enumeration. In comparison, the stochastic ruler method asymptotically converges to global optima and incurs smaller computational overheads. We demonstrate the application of these methods to a wide variety of machine learning models, including deep neural network models used for time series prediction and image classification. We benchmark our application of these methods with state-of-the-art hyperparameter optimization libraries such as $hyperopt$ and $mango$. The KN method consistently outperforms $hyperopt$'s random search (RS) and Tree of Parzen Estimators (TPE) methods. The stochastic ruler method outperforms the $hyperopt$ RS method and offers statistically comparable performance with respect to $hyperopt$'s TPE method and the $mango$ algorithm.

cs.LG↗

arXiv2021

Outpatient Diversion using Real-Time Length-of-Stay Predictions

In this work, we show how real-time length-of-stay (LOS) predictions can be used to divert outpatients from their assigned facility to other facilities with lesser congestion. We illustrate the implementation of this diversion mechanism for two primary health centers (PHCs), wherein we divert patients from their assigned PHC to the other PHC based on their predicted LOSs in both facilities. We develop a discrete-event simulation model of patient flow operations at these two PHCs in an Indian district and observe significantly longer LOSs at one of the PHCs due to disparities in the patient loads across both PHCs. We first determine the expected LOS of the patient at the point in time at which they are expected to arrive at a PHC using system state information recorded at the current time at the PHC in question. The real-time LOS predictions are generated by estimating patient wait times on a real-time basis at the queueing subsystems within the PHC. We then divert the patient to the appropriate PHC on the basis of the predicted LOS estimates at both PHCs, and show through simulation that the proposed framework leads to more equitable utilization of resources involved in provision of outpatient care.

stat.AP↗

arXiv2021

A decision support framework for optimal vaccine distribution across a multi-tier cold chain network

In this paper, we present a decision support framework for optimizing multiple aspects of vaccine distribution across a multitier cold chain network. We propose two multi-period optimization formulations within this framework: first to minimize inventory, ordering, transportation, personnel and shortage costs associated with a single vaccine; the second being an extension of the first for the case when multiple vaccines with differing efficacies and costs are available for the same disease. Vaccine transportation and administration lead times are also incorporated within the models. We also develop robust optimization versions of the single vaccine model to account for the impact of uncertainty in model parameters on the optimal vaccine distribution solution. We use the case of the Indian state of Bihar and COVID-19 vaccines to illustrate the implementation of the framework. We present computational experiments to demonstrate: (a) the organization of the model outputs; (b) how the models can be used to assess the impact of cold chain point storage capacities, transportation vehicle capacities, and manufacturer capacities on the optimal vaccine distribution pattern; and (c) the impact of vaccine efficacies and associated costs such as ordering and transportation costs on the vaccine selection decision informed by the model. We then consider the computational expense of the framework for realistic problem instances, and suggest multiple preprocessing techniques to reduce their computational burden. Finally, we also demonstrate how the robust versions of the single vaccine model outperform the deterministic version under multiple levels of uncertainty in key model parameters. Our study presents public health authorities and other stakeholders with a vaccine distribution and capacity planning tool for multi-tier cold chain networks.

physics.soc-ph↗

arXiv2021

A Discrete Simulation Optimization Approach Towards Calibration of an Agent-based Simulation Model of Hepatitis C Virus Transmission

This study demonstrates the implementation of the stochastic ruler discrete simulation optimization method for calibrating an agent-based model (ABM) developed to simulate hepatitis C virus (HCV) transmission. The ABM simulates HCV transmission between agents interacting in multiple environments relevant for HCV transmission in the Indian context. Key outcomes of the ABM are HCV and injecting drug user (IDU) prevalences among the simulated cohort. Certain input parameters of the ABM need to be calibrated so that simulation outcomes attain values as close as possible to real-world HCV and IDU prevalences. We conceptualize the calibration process as a discrete simulation optimization problem by discretizing the calibration parameter ranges, defining an appropriate objective function, and then applying the stochastic ruler random search method to solve this problem. We also present a method that exploits the monotonic relationship between the simulation outcomes and calibration parameters to yield improved calibration solutions with lesser computational effort.

eess.SY↗

arXiv2021

Simulation Modelling and Analysis of Primary Health Centre Operations

We present discrete-event simulation models of the operations of primary health centres (PHCs) in the Indian context. Our PHC simulation models incorporate four types of patients seeking medical care: outpatients, inpatients, childbirth cases, and patients seeking antenatal care. A generic modelling approach was adopted to develop simulation models of PHC operations. This involved developing an archetype PHC simulation, which was then adapted to represent two other PHC configurations, differing in numbers of resources and types of services provided, encountered during PHC visits. A model representing a benchmark configuration conforming to government-mandated operational guidelines, with demand estimated from disease burden data and service times closer to international estimates (higher than observed), was also developed. Simulation outcomes for the three observed configurations indicate negligible patient waiting times and low resource utilisation values at observed patient demand estimates. However, simulation outcomes for the benchmark configuration indicated significantly higher resource utilisation. Simulation experiments to evaluate the effect of potential changes in operational patterns on reducing the utilisation of stressed resources for the benchmark case were performed. Our analysis also motivated the development of simple analytical approximations of the average utilisation of a server in a queueing system with characteristics similar to the PHC doctor/patient system. Our study represents the first step in an ongoing effort to establish the computational infrastructure required to analyse public health operations in India, and can provide researchers in other settings with hierarchical health systems a template for the development of simulation models of their primary healthcare facilities.

cs.CY↗

arXiv2021

Patient Diversion Across Primary Health Centers Using Real Time Delay Predictors

In the current work, we consider diversion of childbirth patients who arrive seeking emergency admission to public primary health centers (PHCs). PHCs are the first point of contact for an Indian patient with formal medical care, and offer medical care on an outpatient basis, and limited inpatient and childbirth care. In this context, real-time prediction of the wait time of the arriving patient becomes important in order to determine whether the patient must be diverted to another PHC or not. We study this problem using a discrete event simulation that we develop of medical care operations in two PHCs in India. We approximate the labour room service at each PHC as an M/G/1 queueing system and show how the accuracy of real-time delay predictors impacts the extent of the change in operational outcomes at each PHC. We simulate patient diversion using actual delays as well as the delay estimates generated by various delay predictors based on the state of the system such as queue-length, elapsed service time, and observed delay histories. The simulation of the diversion process also incorporates travel time between the PHCs. We also propose a new delay predictor that incorporates information regarding the system state as well as the service time distribution. We compare the operational outcomes at both PHCs without diversion and with diversion using the above delay predictors. We show numerically that more accurate delay predictors lead to more equitable distribution of resources involved in provision of childbirth care across both PHCs.

eess.SY↗

arXiv2020

Optimal minimal-contact routing of randomly arriving agents through connected networks

Collision-free or contact-free routing through connected networks has been actively studied in the industrial automation and manufacturing context. Contact-free routing of personnel through connected networks (e.g., factories, retail warehouses) may also be required in the COVID-19 context. In this context, we present an optimization framework for identifying routes through a connected network that eliminate or minimize contacts between randomly arriving agents needing to visit a subset of nodes in the network in minimal time. We simulate the agent arrival and network traversal process, and introduce stochasticity in travel speeds, node dwell times, and compliance with assigned routes. We present two optimization formulations for generating optimal routes - no-contact and minimal-contact - on a real-time basis for each agent arriving to the network given the route information of other agents already in the network. We generate results for the time-average number of contacts and normalized time spent in the network.

math.OC↗

arXiv2020

A Note on the Stochastic Ruler Method for Discrete Simulation Optimization

In this paper, we propose a relaxation to the stochastic ruler method originally described by Yan and Mukai in 1992 for asymptotically determining the global optima of discrete simulation optimization problems. The `original' version of the stochastic ruler and its variants require that a candidate for the next estimate of the optimal solution pass a certain number of tests with respect to the stochastic ruler to be selected as the next estimate of the optimal solution. This requirement - that all tests need to be passed - can lead to promising candidate solutions being rejected and can slow down the convergence of the algorithm. Our proposed modification to the stochastic ruler algorithm relaxes this requirement, and we show analytically that our proposed variant of the stochastic ruler method incurs lesser computational overhead when a new solution in the neighborhood of the current solution is a `successful' candidate for the next estimate of the current solution. We then show numerically that this can yield accelerated convergence to the optimal solution via multiple numerical examples. We also provide the theoretical grounding for the asymptotic convergence in probability of the variant to the global optimal solution under the same set of assumptions as those underlying the original stochastic ruler method.

math.OC↗

arXiv2020

Discriminatory Price Mechanism for Smart Grid

We consider a scenario where a retailer can set different prices for different consumers in a smart grid. The retailer's objective is to maximize the revenue, minimize the operating cost, and maximize the consumer's welfare. The retailer wants to optimize a convex combination of the above objectives using price signals specific to each consumer. However, variability in unit prices across consumers is bounded by a parameter $\eta$, hence limiting the discrimination. We formulate the pricing problem as a Stackelberg game where the retailer is the leader and consumers are followers. Since the retailer's optimization problem turns out to be non-convex, we convexify it via relaxations. We provide performance guarantees for the relaxations in the asymptotic sense (when number of consumers tends to $\infty$). Further, we show that despite the variability in pricing, the pricing scheme proposed by our model is fair as higher prices are charged to consumers who have higher willingness for demand. We extend our analysis to the scenario where consumers can feed energy back to the grid via net-metering. We show that our pricing policy promotes fairness even in this scenario as prosumers who contribute more to the grid, are given large cuts on buying rates. The policy is also found to incentivize more prosumers to invest in renewable energy, thus encouraging sustainability.

math.OC↗