
Leishmaniasis poses a significant global health concern due to the absence of vaccines for humans and high infection rates in some countries. It is classified as a neglected tropical disease. In 2022, roughly 85% of global visceral leishmaniasis cases were reported in seven countries: Brazil, Ethiopia, India, Kenya, Somalia, South Sudan, and Sudan. Despite Brazil's advanced medical capabilities compared to other affected regions, certain areas still witness a significant number of cases, prompting increased attention from researchers and raising concerns within the healthcare system. This study explores the application of artificial intelligence algorithms, particularly machine learning (ML) models to predict leishmaniasis outbreaks in selected Brazilian cities based on accumulated cases from 2007 to 2022, leveraging available meteorological data to enhance model accuracy. Our investigation concentrated on the following cities in Brazil: Fortaleza/CE, Teresina/PI, and S & atilde;o Luis/MA were chosen for the study of visceral leishmaniasis, whereas Manaus/AM, Rio Branco/AC, and Macap & aacute;/AP were selected for the study of tegumentary leishmaniasis, encompassing both cutaneous and mucocutaneous forms. Several Artificial Neural Network (ANN) architectures were evaluated, including a Simple Feedforward Neural Network (SFNN), a Deep Feedforward Neural Network (DFNN), and a Long Short-Term Memory (LSTM) recurrent neural network. Additionally, the Support Vector Machine (SVM), specifically the Support Vector Regression (SVR), was tested. Various metrics were used to identify the most effective models, in which the Root Mean Squared Error (RMSE) was the primary one. The results highlight the significance of meteorological data as a crucial factor in ML models for predicting leishmaniasis outbreaks, while also emphasizing the importance of fine-tuning these models to achieve greater accuracy. Finally, data and the pseudo-code of the models are accessible through an open repository to encourage further studies in this area.
Purpose: This study aims to develop a heuristic for an outpatient appointment system considering patient classification. Design/methodology/approach: The proposed heuristic was applied in simulations with eighteen scenarios, combining different environmental factors. Total cost was adopted as a performance metric, composed of the patient's wait time and the service provider's idleness and overtime. The patients were divided into two classes according to their no-show probability, in an arrivals sequence with a binomial distribution. As a significance test of the results, Bonferroni-adjusted repeated measures analysis was applied. Findings: Having Dome rule as baseline, an increase in performance in terms of total cost (TC) TC ) was observed, which varied between 0.46 % and 5.94 % among the means of the simulated environments, validated using the proposed significance test. The greatest benefits were obtained in the scenarios with lower ratios between service provider costs and patient costs (CR), CR ), as well as lower coefficients of variation for service times (Cv). It was also found that the heuristic is more efficient when patients from the class with the highest no-show rate predominate in the session. Originality: The single study identified in the literature that contemplates recalculations adopts deterministic service times to make its model viable. The present research, in turn, makes more realistic assumptions for the simulated environments, considering the variables and probability distributions most commonly observed in practical contexts Practical implications: The proposed heuristic provided a significant increase in performance for some combinations of environmental factors analyzed, preserving flexibility in the choice of appointment slots and covering a wide range of healthcare services found in practice.
In many countries, the rapid aging of the population leads to an additional burden on already stretched long-term care systems. This often manifests itself in excessive waiting times for long-term care centers, and in abandonments (i.e., patients passing away while they are waiting). Interestingly, in practice, long waiting times are not caused by a lack of available total capacity in the system, but by systematic inefficiencies in the allocation of patients, each with their personal preferences and (in)flexibility, to geographically distributed care centers.Motivated by this, we propose a new and easy-to-implement method for the optimal allocation of patients-in-need to nursing homes, balancing the trade-off between the waiting time performance and the individual patients’ preferences and levels of flexibility. The optimal placement policy found by solving a Markov Decision Process demonstrates that for small instances, the mean optimality gap of the allocation model is equal to 1. 3%. We validate a simulation model for a real-life use case of allocating somatic patients to nursing homes in the Amsterdam area. The results show that if more patient replacements are approved, the allocation model can reduce the abandonment fraction under the current policy from 32.2% to 7.4% and waiting times at the same time. Moreover, with the allocation model individual preferences can be served better, which thus provides a powerful means to face the increasing need for patient-centered and sustainable long-term care solutions.
Layout design is considered a crucial aspect of healthcare architecture and its goal is to allow easy access to essential hospital services and effective patient care. Literature suggests that modifying or redesigning the inpatient unit layout is one of the ways to maximize patient visibility in an inpatient layout. However, prior work has been descriptive in nature and limited in their ability to derive optimal layouts. To fill this gap, we propose a non-linear optimization model that optimizes both equity and effectiveness in visibility by jointly determining the optimal location of two nurses and patient bed positions in multiple rooms. The bi-objective model is then converted into a single objective model utilizing the epsilon-constrained method, with equity in the objective function and effectiveness as a constraint. Patient visibility is estimated using a ray-casting algorithm that also considers nurses' line of sight, door positions, and obstruction levels. A progressive refinement algorithm embedded in the Particle Swarm Optimization framework is proposed to efficiently solve this model. Our results suggest that optimizing bed position in conjunction with nurse position can enhance equity by over 45.2% compared to just optimizing the nurse position. Similarly, angular layouts are superior to linear layout by up to 53% in patient equity. We also notice that increasing spatial distance between nurses in angular layouts can further increase equity. Our approach provides valuable insights and can serve as a benchmark tool for hospitals looking to improve the design of their inpatient units that promote patient safety and high-quality care.
In a home care setting, high-quality care is typically associated with continuity of care. In addition, the increasing pressure due to labor shortages calls for cost-efficient operations. This paper focuses on obtaining cost-efficient daily schedules over a longer time horizon, with balanced shift lengths, while ensuring continuity of care (using the continuity of care index). To address this challenge, we propose a novel method based on blueprint routes. This method generates daily schedules by constructing optimized shifts and routes with regard to travel time, (time window) waiting time, and shift costs based on hourly wages. To ensure continuity of care, the daily scheduling decisions are strategically guided using the concept named blueprint routes. The blueprint routes are pre-optimized (partial) routes that help to align the daily schedules to achieve continuity of care in the subsequent nurse-to-shift assignment. Model-based evolutionary algorithms are employed to overcome the NP-hardness of the routing problem and nurse-to-shift assignment. Real-life-based numerical experiments demonstrate that continuity of care does not have to compromise home care schedule costs significantly.
We develop a modeling framework for rostering, absence and demand uncertainty realization, and rerostering to perform detailed quantitative analyses of the robustness of nurse rosters. The framework reflects a real-life problem observed at the Department of Neonatal Intensive Care (DNIC) at St. Olavs Hospital in Trondheim, Norway, but is general and has a high transfer value with respect to using it to analyze roster robustness at other departments. We present multiple proactive strategies to enhance the stability of a roster and a reactive rerostering problem used to improve the flexibility. An extensive case study is performed using historical data from the department. The results show that there is a great potential to improve the stability and flexibility of the rosters using the best combination of strategies. We show that allowing nurses to trade extra weekend work for extra days off, assign surplus work hours evenly over all work shifts, and consider the absence profile of nurses when making the rosters are key strategies to create robust rosters.
The COVID-19 pandemic has tested the resilience of Supply Chains (SCs), which has faced many restrictions and affected their global response. Worldwide stockouts were witnessed due to SCs disruptions, which may endanger lives since some products are critical to responding to this global threat, such as ventilators and Personal Protective Equipment (PPE). Thus, this work aims to help deal with the pandemic impacts on critical SCs, addressing the distribution of materials that are used to cope with the pandemic and considering the resilience of its SCs, dealing with a gap of few studies combining simulation and optimization approaches to tackle this situation. We propose a dynamic framework based on a stochastic population model to address pandemic behavior and an optimization model to support decision-making in a PPE supply chain subject to a pandemic-driven disruption that can be updated anytime necessary. We develop a social objective function that aims to deliver PPE where they are most needed. The proposed approach is illustrated by an example involving real data from a Brazilian company that distributes PPE during the COVID-19 pandemic. We find that profit was inversely correlated with social gain, suggesting that optimizing profits is a poor strategy for addressing public health or social crisis. Still, our model furnishes results with an acceptable profit while prioritizing its effect on coping with the pandemic. As implications, our framework can be applied to support decision makers to improve SCs’ resilience and better allocate resources during disruptive circumstances in which the uncertainty is high, such as future pandemics.
With the growing demand for healthcare resources, pressure on efficient usage of available bed capacity is increasing. Peaks in bed demand corresponds to overcrowding in upstream units such as emergency department or operating rooms. With a balanced schedule in elective surgeries integrated into the master surgical schedule, peak traffic can be leveled across the week without changing resource capacity. Hence, overcrowding is reduced without turning away any patients or increasing bed capacity. This study formulates the integration of master surgical and elective surgery scheduling problems as an Integer Programming model to minimize the fluctuation in the required ward beds for elective inpatients admitted for surgery to the hospital, by changing the day of surgery. This demonstrates the opportunities for smoothing the expected patient demand for beds by adjusting the operating room schedule. This decision is made at the tactical level. The model has been examined using data on the elective patient demand for beds in the hospital during typical weeks driven from Hamilton Health Sciences in Ontario, Canada. The integer programming model has been solved using GAMS/CoinCBC MIP Solver. The model enhances bed management by not only smoothing but also reducing the peak demand. The optimal schedule reduces the peak patient demand for bed by about 3-19% for the test samples. The model can be extended to cover the demand for other resources such as ICU beds.
This work addresses the Multi-Objective Mammography Unit Location–allocation Problem (MOMULAP), aiming to meet three objectives: maximize mammography screening coverage, minimize the total distance traveled weighted by the number of users, and maximize equity in access to mammography screening. We introduce a mixed-integer nonlinear programming (MINLP) formulation to represent the MOMULAP and algorithms based on the Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Strength Pareto Evolutionary Algorithm (SPEA2) for treating it. The algorithms were tested with data from seven Brazilian states. In these states, the number of cities ranges from 139 to 853, equipment from 23 to 347 units, and estimated annual demand for screenings from 96,592 to 1,739,085. The solutions provided by this work allow health managers to choose the appropriate location and allocation of the mammography units, considering different objectives.
Designing high quality nurse rostering plans is essential for health care facilities in order to guarantee efficiency, safety and quality-of-care balanced with staff well-being. We introduce a new real-world formulation for the nurse rostering problem, arising in many Italian healthcare institutions, which has been developed in collaboration with a primary software company in the field. It considers nurses with different skills, special shifts depending on the skills, time work-load limits, and different types of days-off. In addition, preferences and incompatibilities between nurses are taken into account. We propose a MIP model and a local search method, driven by a Simulated Annealing metaheuristic, based on a combination of two neighborhoods. The solution method was tested on 34 real-world instances coming from various healthcare institutions in North Italy. The dataset is available at https://bitbucket.org/satt/nrp-instances, along with our best solutions. (c) 2023 Elsevier Ltd. All rights reserved.
During Covid-19, medical service networks (MSNs) faced new challenges, such as an impressive increase in hospital visits, a shortage of hospital beds and staff, and insufficient information to estimate the number of mild and critical cases. In addition, governments were encountered to implement appropriate quarantine policies. Dealing with these problems became more complex and challenging when a new wave of disease occurred. This study develops a mixed-integer linear programming model for reorganizing an MSN to manage future pandemic waves. The model aims at reallocation medical staff to prevent a shortage of hospital beds. A fuzzy approach is employed to estimate the uncertain number of patients in each period. As a result, direct hospital visits are decreased by 60% on average, and shortages of beds are avoided by adding the fewest beds possible in each period. The model can also optimize several performance ratios, e.g., the ratio of hospitalized patients to the specialized personnel assigned to each hospital, which is decreased by approximately 40% in our case.
Growing healthcare needs leverage the potential savings of using resources efficiently. To that end, ProMoBed is a comprehensive model that supports strategic planning of bed capacity in inpatient hospitals. The model consists of an extrapolation and simulation component, the former supplying input for the latter. The extrapolation model forecasts admission rates and the average Length of Stay for pathology groups, and corrects for demographic changes. Subsequently, the simulation model emulates the demand for bed capacity, and makes service-level based bed capacity suggestions. Additionally, the model uses the Shapley value principle to disaggregate the effects on demand for inpatient days due to different causes. Results from the extrapolation model are applied to regions in Belgium, showing expected divergence in inpatient day demand evolution.
The literature in Intensive Care Units (ICUs) data analysis focuses on predictions of length-of-stay (LOS) and mortality based on patient acuity scores such as Acute Physiology and Chronic Health Evaluation (APACHE), Sequential Organ Failure Assessment (SOFA), to name a few. Unlike ICUs in other areas around the world, ICUs in Ontario, Canada, collect two primary intensive care scoring scales, a therapeutic acuity score called the “Multiple Organs Dysfunctional Score” (MODS) and a nursing workload score called the “Nine Equivalents Nursing Manpower Use Score” (NEMS). The dataset analyzed in this study contains patients’ NEMS and MODS scores measured upon patient admission into the ICU and other characteristics commonly found in the literature. Data were collected between January 1st, 2015 and May 31st, 2021, at two teaching hospital ICUs in Ontario, Canada. In this work, we developed logistic regression, random forests (RF) and neural networks (NN) models for mortality (discharged or deceased) and LOS (short or long stay) predictions. Considering the effect of mortality outcome on LOS, we also combined mortality and LOS to create a new categorical health outcome called LMClass (short stay & discharged, short stay & deceased, or long stay without specifying mortality outcomes), and then applied multinomial regression, RF and NN for its prediction. Among the models evaluated, logistic regression for mortality prediction results in the highest area under the curve (AUC) of 0.795 and also for LMClass prediction the highest accuracy of 0.630. In contrast, in LOS prediction, RF outperforms the other methods with the highest AUC of 0.689. This study also demonstrates that MODS and NEMS, as well as their components measured upon patient arrival, significantly contribute to health outcome prediction in ICUs.
The nurse scheduling problem (NSP) has become significant in recent years due to its direct impact on patient healthcare. This problem involves assigning nurses’ shifts while fulfilling a set of hard constraints associated with labor regulations and soft constraints related to personal preferences, workload balance, among others. Most studies have focused on providing solutions for deterministic scenarios without considering unexpected disruptions, such as an unscheduled nurse absence. This study integrates two of the most common approaches to address absenteeism: preventive and reactive. First, we propose a multiobjective linear model for staff scheduling that preventively assigns backup nurses for each day. The NSP is known to be an NP-hard problem. Therefore, we used a genetic algorithm to obtain solutions in a reasonable amount of time. To mitigate the effect of unscheduled nurse absences, we propose two reactive rescheduling policies, one that seeks to maintain the baseline schedule and another that prioritizes the exclusive use of backup nurses. We used Montecarlo simulation under different problem settings to compare the proposed policies with a policy that does not use the preventive approach. The probability that a nurse will accept an additional shift to cover an absence was also considered. Simulation results suggest that both of our preventive–reactive policies outperform the non-preventive policy, especially in the presence of a small probability that a nurse will accept an additional shift. Finally, we used the proposed policies to create the monthly nursing schedule in a reference hospital in Bogotá-Colombia.
Planning the use of chemotherapy drugs in outpatient treatment is a complex problem due to the variability of cancer, resulting in different chemotherapy protocols. This process involves factors such as the cyclical nature of treatment protocols and clinical resources. Within this context, an optimization model is needed to plan the use of chemotherapy drugs in the preparation of doses for patient treatment, considering the operational particularities of treatment centers. This study proposed a linear programming model based on the multiple knapsack problem, in order to optimize the combination of different vials of chemotherapy drugs, minimizing the total cost of treatment. The model, based on the daily schedule of patients, provides the best combination of drug vials and supports the preparation process of each dose prescribed by the doctor for each patient, respecting the treatment protocol and resource limitations. The model was implemented in the CPLEX 12.9.0 application, and the computational tests were performed with real data. The results demonstrated that the costs when applying the model were 5% lower when compared to the current manner in which the oncology pharmacy combines the drugs vials.
Beds are a critical resource for hospitals, requiring effective management to ensure the quality of care for patients. Beds operate in a closed-loop circuit and must be thoroughly cleaned between patients' arrivals to prevent infections. Hospitals must implement efficient logistics systems to collect, transport, store, and clean unclean beds from discharged patients. These systems must be robust and efficient to meet the varying bed supply needs, given the available resources such as beds, staff and machines. This study aims to develop a decision support tool to optimise bed cleaning logistics and ensure the availability of sterile beds for incoming patients at all times. The study is based on the bed flow and cleaning organisation of a Danish public hospital. A discrete event simulation model (DES) of the back-end bed flow has been developed. The paper also presents a tension level indicator to reflect the behaviour of cleaning staff when facing variations in demand and bed stock. Using the organisational set-up (staff schedules, policies, and bed fleet size), the DES model: (1) evaluates the ability to provide sterile beds in a reasonable time, (2) measures the stress on cleaning staff, and (3) visualises resource usage. This study illustrates how to incorporate the staff's perceived workload and resulting behaviour into a DES model to capture the behavioural aspect of staff's decision-making.
As pressure on healthcare systems continues to increase, it is becoming more and more important for hospitals to properly manage the high workload levels of their staff. Ensuring a balanced workload allocation between various groups of employees in a hospital has been shown to contribute considerably towards creating sustainable working conditions. However, allocating work to different organizational units in a fair manner is not straightforward when it involves complex decision-making processes. In this paper we set out to balance the workload of heterogeneous hospital wards by optimizing the patient admission scheduling problem. Given the multi-period nature of patient admission scheduling, we introduce a new equity objective that captures both spatial (between hospital wards) and temporal (between days in the planning period) workload balancing. The resulting bi-objective problem is solved using an exact criterion space search algorithm. Our computational study employs problem instances that have been generated based on real-world data. The results demonstrate how spatially and temporally balanced workload allocations can be generated by minimizing the proposed equity objective. Moreover, we analyze sets of nondominated solutions to gain various insights into the trade-off between schedule cost and workload balance.
The objective of this paper is to minimize the total cost of vaccine storage and distribution operations at centralized distribution centers (DCs) and at clinics so that clinics are provided with vaccines in a timely fashion while under resource and environment-protection constraints. A non-linear mathematical programming model is developed to improve the efficiency of large-scale influenza vaccination programs. The suggested model is tested and justified through computational experiments with real-life data from a Clalit HMO influenza vaccination case study. The investments in green (environment-protecting) activities recommended by the optimal plan are smaller than the expected monetary benefits associated with their effects. A possible application of this research is for optimizing vaccination plans for different subpopulations and various HMOs. Our vaccine supply chain model includes the costs of disposal, recycling, and utilizing clean technologies (i.e., low-pollution gas heating/cooling, electric transportation cars, energy saving policies). It integrates the operational cost/benefit parameters of vaccination programs with the costs/benefits of green activities.
The planning of radiotherapy treatments is based on the use of mathematical optimization models and algorithms. A treatment plan will correspond to an admissible solution to a problem that is defined by the medical prescription. The medical prescription defines the constraints that have to be satisfied, and that are patient dependent. To find a treatment plan complying with all the defined constraints, an objective function is built, having a set of parameters that can be tuned. In the clinical practice, the objective function parameters are tuned through a trial-and-error procedure. One common way of defining this objective function is to consider as parameters weights that are related with the importance of the corresponding structures of interest (volumes to treat and organs to spare), as well as upper and lower bounds that are related with the constraints defined by the medical prescription. Some treatment options, like the set of irradiation directions (beam angles), are usually defined beforehand by the planner, considering previous experiences with similar cases. This trial-and-error process is lengthy, since it can take up to several hours to calculate an admissible treatment plan for a single patient. There have been some research efforts to automate the treatment planning procedure, releasing the planner for other important tasks in the radiotherapy treatment workflow and guaranteeing the consistent calculation of high-quality plans. In this work derivative-free optimization algorithms are integrated with fuzzy inference systems that automatically tune the objective function parameters so that an admissible solution is found. This fully automated approach was tested and assessed considering six head-and-neck cancer cases already treated at the Portuguese Institute of Oncology at Coimbra (IPOC). For the clinical cases tested, comparisons between different treatment plans clearly favor the proposed approach. For a similar tumor coverage, it was possible to improve the sparing of the spinal cord, brainstem and parotids. Automating radiotherapy treatment planning can contribute to improved treatment plans in a consistent way.