The Comprehensive Policy for the Prevention and Care of Psychoactive Substance Use in Colombia aims to improve the care provided to people, families, and communities at risk or struggling with psychoactive substance use through prevention and mitigation programs. The effectiveness of these programs depends on both population participation and access to intervention centers. This study proposes a bi-objective integer programming model within a location-allocation framework to support policy decisions under budget constraints. To estimate drug-related risk, we integrate sentiment analysis from social media data (X, formerly Twitter) as a key input into the optimization model. Specifically, negative sentiment derived from posts is used to inform the spatial distribution of risk between locations. The model simultaneously minimizes population-level risk and distance to services, while ensuring equitable coverage based on multidimensional poverty and rurality. The proposed approach was applied to real-world data from Atlántico. The results demonstrated that the bi-objective model achieved an average coverage of 24.67% of the population within a 40 km radius, effectively balancing service accessibility between high-risk urban areas and underserved rural zones. Compared to a population-based heuristic, which achieved only 8.85% coverage and excluded 22 of 23 locations, the proposed model significantly improved equity in service distribution. Furthermore, complementary topic modeling using Latent Dirichlet Allocation (LDA) revealed key themes in public discourse, including drug cartels, addiction risks, and social impacts, providing valuable information to support tailored communication and community engagement strategies for new intervention centers.

While prior research on inpatient care costs has primarily focused on patient- and clinical-level factors, limited empirical attention has been given to how physician collaboration shapes cost outcomes. Few studies have examined this relationship using social network analysis at the micro level. This study investigates how collaboration networks influence care costs, the mechanisms through which these effects occur, and the moderating role of attending physicians' workload. The structure of collaboration networks determines how efficiently information is shared and decisions are made, which in turn influences healthcare costs. Physicians' centrality within the network impacts their ability to access information and facilitate knowledge transfer, with higher centrality promoting better collaboration, reducing redundancies, and improving decision-making. Using digital trace data from a hospital in China, we employed social network analysis to identify collaborative networks and fitted a log-linear model to examine the association between these networks and healthcare costs. The results demonstrate that degree and closeness centrality of the attending physicians are negatively correlated with hospitalization cost. In contrast, betweenness centrality was found positively correlated with hospitalization cost. Additionally, we find that centrality metrics help reduce diagnostic and treatment costs by enhancing information exchange and clinical decision-making. Furthermore, the workload of attending physicians significantly impacted the relationship between collaboration network centrality and care costs. Specifically, the combined effect of an attending physician's degree and workload has an additional negative impact on hospitalization costs. The interaction between betweenness centrality and workload was found to be positively correlated with hospitalization costs. As the healthcare industry continues to evolve towards more collaborative and integrated models, these findings contribute to guiding effective and cost-efficient healthcare delivery.

Pandemics pose significant challenges, particularly in the early stages when vertical resupply chains are overwhelmed. To mitigate the impact of medical resource shortages, we develop a multi-period optimization model incorporating lateral transshipment and hospital admission to minimize the total number of infected individuals by strategically allocating regional resources in the face of complex dynamics, including endogenous hospital admission rates and pandemic spread. To capture the temporal-spatial nature of pandemics, we extend the Susceptible-Exposed-Infected-Hospitalized-Recovered (SEIHR) model by accounting for population migration. Additionally, we derive threshold-type structures for optimal resource transfers, considering factors such as pandemic dynamics, patient length of stay, and budget constraints. We also demonstrate the effectiveness of our models via numerical experiments. Our research identifies three main findings: 1) Pooling medical resources effectively reduces infections and alleviates shortages in outbreak areas. This strategy is particularly beneficial during pandemics due to self-reinforcing infection dynamics and surging demand. 2) Regions adjacent to the epicenter should exercise caution in contributing resources to avoid exacerbating infections through population migration. 3) While effective in localized outbreaks, widespread resource scarcity can limit the viability of pooling strategies, potentially leading to increased infections and fluctuating resource levels in transferring regions.

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.

Harnessing the synergy between artificial intelligence (AI) and operations research (OR) helps drive efficiency, safety, and innovation in biomanufacturing. AI offers predictive capabilities, while OR represents the pinnacle of prescriptive analytics. AI and OR complement each other by offering structured, interpretable, and verifiable solutions to complex operational challenges. In this commentary, we reflect on how to realize the full potential of AI-OR implementations in biomanufacturing. We elaborate on recent university-industry partnerships demonstrating these benefits and propose a roadmap for AI-OR integration in biomanufacturing.

Operating room (OR) scheduling is a critical challenge in healthcare, directly impacting patient outcomes and hospital efficiency. Traditional methods often struggle with the complex, multi-level constraints and uncertainties inherent in OR scheduling, such as resource limitations, variable surgery durations, and emergency cases. This study aims to develop a novel hybrid framework that optimizes OR scheduling by integrating multi-level optimization with reinforcement learning and column generation techniques. The proposed framework decomposes the OR scheduling problem into strategic, tactical, and operational levels, enabling focused optimization at each layer while ensuring cohesive decision-making across the hierarchy. Reinforcement learning guides the column generation process, learning policies that identify promising scheduling options to enhance solution quality and computational efficiency. Robust uncertainty handling mechanisms are incorporated to manage variability in surgery durations and resource availability without compromising tractability. Experiments were conducted using three years of real-world data from Shanxi Provincial People's Hospital, complemented by large-scale synthetic datasets to evaluate scalability and robustness of the framework. The framework demonstrates meaningful improvements in key operational metrics compared to traditional approaches. Analysis of three years of implementation shows consistent enhancements in operational efficiency, including a reduction in average patient waiting time by 15.8% (from 10.1 to 8.5 days), an increase in OR utilization by 5.4 percentage points (from 73.8% to 79.2%), and improved workload balance among surgeons. The framework maintains robust performance under uncertainty, achieving a 92.5% feasibility rate and reducing schedule disruptions by 26.2%. The proposed hybrid framework offers a practical and scalable solution for optimizing OR scheduling, demonstrating improvements in healthcare delivery and operational performance in real hospital environments. By effectively balancing multiple operational objectives while handling practical constraints and uncertainties, the framework provides a viable approach for healthcare systems seeking incremental yet sustainable improvements in efficiency and patient care.

Cardiovascular diseases (CVDs) are one of the primary reasons for death worldwide. These diseases often occur due to the occlusion of coronary arteries, thereby leading to insufficient blood and oxygen supply that damages cardiac muscle cells. Electrocardiogram (ECG) signals which reflect heart electrical activity are being used for diagnosing various cardiac diseases. Typically, a standard ECG consists of 12 channels referred to as leads which enable practitioners to monitor heartbeats through different channels where each heartbeat lasts approximately 600 ms. The majority of studies focus on the classification and early diagnosis of arrhythmias. Although the current studies on change-point methods have acquired massive accuracy in detecting potential shifts during a multi-channel process, they lack flexibility in manually assigning more weights to the channels, which are of more importance for experts. This could be addressed by implementing the weighted multivariate functional principal component analysis (WMFPCA). The objective of this study is to develop a novel change-point detection method to monitor long-term cardiovascular treatment. A third-order tensor structure was employed to represent the 12-lead ECG data in three dimensions (beats × samples × leads). Exploiting intra-beat, inter-beat, and inter-lead correlations along with channel significance in the third-order tensor, the WMFPCA is incorporated into Hotelling's statistic to construct monitoring schemes. Simulation results show that the proposed approach outperforms the existing methods in monitoring multi-channel processes. Finally, applying the suggested model on a real-world dataset containing Myocardial Infarction (MI) subjects verifies the model.

During the initial phase of a pandemic outbreak, the rapid increase in the number of infected patients leads to shortages of medical resources for both pandemic-related and non-pandemic (ordinary) patients. It is crucial to efficiently utilize limited existing resources and strike a balance between controlling the pandemic and sustaining regular healthcare system operations. To tackle this challenge, we introduce and investigate the problem of optimizing the location of designated hospitals, reallocating beds within these hospitals, and assigning different types of patients to these hospitals. Designated hospitals isolate pandemic-related patients from ordinary patients to prevent cross-infection. Moreover, isolation beds can be converted into ordinary beds and vice versa. Considering the stochasticity and evolving nature of the pandemic, we formulate this problem as a multi-stage stochastic programming model, integrating a compartmental model with time-varying random parameters to enable dynamic resource allocation as the pandemic progresses. The model is then solved by a data-driven rolling horizon solution approach. We illustrate the effectiveness of our model using real data from the COVID-19 pandemic. Compared with two other approaches, our model demonstrates superior performance in controlling the spread of the pandemic while addressing the needs of both pandemic-related and ordinary patients. We also conduct a series of experiments to uncover managerial insights for policymakers to better utilize existing resources in response to pandemic outbreaks. Results indicate that admitting as many pandemic-related patients as possible during the initial phases of the outbreak can effectively flatten the pandemic peaks and alleviate strain on the healthcare system.