Health Care Management Science最新文献

The COVID-19 pandemic shed light on the fragility of today's public health systems and failure to sufficiently invest in preparedness. These shortcomings are observed in delays achieving timely, equitable, and sufficient access to life-saving vaccines when faced with erratic demand. This Current Opinion describes vaccine supply networks (VSNs) from a complex adaptive systems (CAS) lens, highlighting interactions between system elements and co-evolution with the environment in which they operate. More specifically, it shows how broadening the boundaries of VSNs reveals the high degree of complexity that leads to unexpected and emergent system behavior, especially when disease threats evolve over time and across geographies. A CAS lens allows for the design of improved management strategies to ensure continued performance of VSNs during both outbreak and inter-epidemic periods, thus contributing to sustained disease management. It points to ample opportunities for more integrated modeling across disciplines to capture inherent feedback loops that influence both VSNs and disease dynamics. Furthermore, it reveals how pandemic preparedness relies on a broader understanding of the mechanisms that drive outbreak prevention and control, beyond vaccines and their direct supply chains. Finally, it highlights the value of adaptive management to navigate inevitable future disruptions and associated uncertainties, overcoming limitations of typical risk-mitigation strategies based on prediction and control.

In home healthcare service systems, each healthcare service provider (HSP) is assigned a list of patients to be visited at their homes. We focus on generating a daily patient visit plan that selects the patients to be visited according to their priorities and locations, and determines the route of each HSP. Additionally, we address unexpected urgent patients by solving an optimization problem involving all HSPs cooperating when an urgent patient visit request arises. This problem is formulated with multiple objectives in a lexicographic optimization framework. Two approaches have been implemented: a mixed integer programming model solved within a time limit (TL-MIP) and a Greedy Randomized Adaptive Search Procedure followed by Variable Neighborhood Search (GRASP+VNS). These approaches are compared in a case study that considers serving patients, with several performance metrics analyzed through extensive simulation experiments. The results indicate that the heuristic approach (GRASP+VNS) significantly reduces run times (by approximately 85% on the average overall instances) compared to the TL-MIP approach, while providing solutions that are not far from the TL-MIP approach in terms of the total priority of visited patients, the heuristic deviates at most 2% over different types of instances. Centralized planning with cooperation among two or three service providers reduced the total travel time by 30% and 45%, respectively, and decreased the number of postponed visits by 50% compared to the non-cooperation model.

In positive and unlabeled (PU) learning problems, only positive examples are labeled. Unlabeled data contain both positive and negative examples. Studies show that positive examples of (secondary) diagnoses, and clinical conditions, such as sepsis, are present in unlabeled hospital administrative data, potentially distorting hospital reimbursement systems, and negatively affecting hospitals' revenue and profitability. We investigate whether PU learning is suitable for improving the quality of hospital administrative data. We train three models on 313,434 hospital cases using hospital cost features: two based on the two-step "spy" approach and one using a robust PU learning method. For model evaluation, we rely exclusively on positive examples due to the PU setting. To further assess model performance, we perform an external validity check: We relabel unlabeled sepsis cases, derive new sepsis rates, and compare them to those reported in medical record review studies. All models identify true positives well in unseen data. External validity checks show, however, that only the robust PU learner effectively discriminates between positives and negatives in the unlabeled data, yielding new sepsis rates within the range of sepsis rates reported in medical record review studies. PU learning can improve the quality of hospital administrative data, but its effectiveness depends strongly on the choice of learning approach and classifier. The output of a PU learner can potentially improve hospital reimbursement systems, hospital revenue and profitability management, and sensitivity analyses in healthcare management science, health economics, health services research, and disease surveillance.

This paper studies the effects of quality oversight in the context of assessing kidney transplantation-related outcomes and possible unintended consequences (e.g., cherry-picking of organs and selection of healthier transplant candidates). In this context, we propose a stochastic economic model that identifies socially optimal kidney transplant choices given the inherent trade-off between the expected wait time and the quality of the received donor kidney for a given patient. Socially optimal decisions seek to maximize the utilitarian welfare function defined as the sum of all patients' post-transplant expected utilities. To determine the social loss, we compare the socially optimal decisions to those taken by a transplant program that maximizes its utility. We derive the optimal quality oversight policy that minimizes social loss and examine how decisions are impacted due to the changes introduced by the new Kidney Allocation System. Our empirical analysis using data from the Scientific Registry of Transplant Recipients and United States Renal Data System indicates that current quality oversight imposed through Conditions of Participation results in inefficient transplant decisions for 56% of recipients, and the performance is inconsistent across different regions and parameters. We propose that the risk-adjusted post-transplant performance assessment policy considers the factors impacting demand-supply parameters such as organ availability in the 11 US transplant regions, candidates' blood type, and the newly introduced Kidney Allocation System. Policymakers and providers can utilize insights from our findings to design effective oversight mechanisms and make informed decisions regarding transplant and waitlist management that yield desired outcomes.

We study the scheduling of appointments for population-based breast cancer screening, considering different patient types in view of their stochastic no-show behaviour and service duration. The associated multi-level problem under study comprises both tactical planning decisions, assigning patients in advance to a mammography machine at a dispersed unit and appointment day, and operational scheduling decisions, stipulating the appointment time for patients. To mitigate the impact of operational variability, performance is safeguarded by optimising the minimum performance associated with defined chance constraints relative to the minimum number of performed screenings and the maximum patient wait time, resource idle time and overtime. We develop a decomposition method that iterates between tactical and operational decision levels with feedback loops. The tactical problem is reformulated as a deterministic mixed-integer quadratic-constrained programming problem and solved via a heuristic that defines a promising solution region based on problem-specific estimates. The operational problem is solved via Sample Average Approximation and decomposition of patient sequencing and appointment time assignment decisions. Computational results show that the developed decomposition-based procedure with feedback and the phase-specific methodologies are superior in terms of time and solution quality compared to alternative methods.

This current opinion explores the transformative potential of explainable artificial intelligence (XAI) for health care management systems. While AI has already demonstrated substantial benefits in clinical decision-making, operational efficiency and patient outcomes, its adoption is often hindered by the lack of transparency in AI-driven decision-making. XAI bridges this gap by providing interpretability, thereby increasing trust between policy-makers, clinicians, administrators and patients. However, despite promising examples, the explicit integration of XAI remains underexplored in health care management research. This current opinion therefore aims to emphasize the crucial role of XAI in improving health care management and to position it as an important topic for advancing the field, with Health Care Management Science (HCMS) playing a leadership role in fostering this development.

The ability to accurately forecast unscheduled care needs is of paramount importance for decision making in healthcare operations, ensuring a continuous and high-quality level of care. In this work, we provide a literature review of 156 research articles of forecasting applications with special focus on care services that are not scheduled in advance such as emergency departments. Our paper presents two key contributions. Firstly, we propose a novel framework designed to characterize the application of forecasting process across various unplanned healthcare services. Our taxonomy facilitates the detection, decomposition, and categorization of forecasting processes, enhancing the understanding of their deployment in different unscheduled care settings. Secondly, we conduct a comprehensive literature review based on a systematic search, critically analyzing the state of forecasting research in unscheduled care services and identifying key research gaps. We explore forecasting problems in depth, examining their purpose, the various methodologies used, the rigor used in generating and evaluating forecasts, and the reproducibility of results, all within the context of the proposed framework. By consolidating the current state of the art, this paper provides valuable insights to both healthcare professionals and academics regarding the effective application of forecasting in unscheduled care services. Finally, it serves as a roadmap for identifying major research gaps and outlines an agenda for future investigations.

Ensuring widespread accessibility of healthcare services is a crucial policy objective. Accordingly, the Italian National Recovery and Resilience Plan (NRRP) has prioritized territorial medicine, channeling post-pandemic investments toward the restructuring of primary care services. A notable change is the establishment of Community Healthcare Centers (CHCs). This paper investigates how CHCs contribute to the accessibility of healthcare in urban and rural areas. By leveraging a comprehensive dataset of general practitioners' availability and estimating future demand-and-supply scenarios, we examine the impact of CHCs under two different capacity allocation strategies. Strategy 1-Capacity expansion-involves allocating additional service hours of general practitioners to CHCs in order to maximize accessibility. Strategy 2-Capacity redistribution-accounts for the persistent shortage of healthcare professionals faced by Italy in the recent years by reallocating a portion of general practitioners' current services from their existing workplace locations to CHCs. Our results indicate that CHCs have the potential to maintain current accessibility levels and also enhance them in the years to come. Moreover, we demonstrate that simply redistributing the current capacity can improve future accessibility. Finally, we show that a mix of the capacity expansion and redistribution strategies (Strategy 3) can maximize accessibility in the future, limiting the need for new professional staff.

In this research, we present an interpretable AutoML approach for the early diagnosis of hypertension and hyperinsulinemia among adolescents, conditions that are critical to identify during these formative years due to their requirement for lifelong care and monitoring. The dataset, collected from 2019 to 2022 by Serbia's Healthcare Center through an observational cross-sectional study, posed challenges common to medical datasets, including imbalances, data scarcity, and a need for transparent, explainable predictive models. To counter these issues, we utilized three AutoML frameworks - AutoGluon, H2O, and MLJAR - in conjunction with a Tabular Variational Autoencoder (TVAE) to synthetically augment the data points, Prinicipal Component Analysis (PCA) for dimensionality reduction, and SHapley Additive exPlanations (SHAP) and Permutation feature importance analyses to extract insights from the results. AutoGluon outperformed the others on the original dataset, delivering better results with weighted ensemble models for both conditions under a 12-minute budget-time constraint and maintaining all evaluation metrics below a 4% threshold, all without the need for further scaling or calibration in the experimental setup. Our research underscores the broad applicability of the current AutoML paradigm, highlighting its particular benefits for the healthcare domain and diagnostics, where such advanced tools can enhance patient care.