key: cord-0966079-gtk1kzp9 authors: Elleuch, Mohamed Ali; Hassena, Amal Ben; Abdelhedi, Mohamed; Pinto, Francisco Silva title: Real-time prediction of COVID-19 patients health situations using Artificial Neural Networks and Fuzzy Interval Mathematical modelling date: 2021-06-24 journal: Appl Soft Comput DOI: 10.1016/j.asoc.2021.107643 sha: 10977e045cf6d6fc8df43576d99a1305ff820a52 doc_id: 966079 cord_uid: gtk1kzp9 At the end of 2019, the SARS-CoV-2 virus caused an outbreak of COVID-19 disease. The spread of this once-in-a-century pathogen increases demand for appropriate medical care, which strains the capacity and resources of hospitals in a critical way. Given the limited time available to prepare for the required demand, health care administrators fear they will not be ready to face patient’s influx. To aid health managers with the Prioritization and Scheduling COVID-19 Patients problem, a tool based on Artificial Intelligence (AI) through the Artificial Neural Networks (ANN) method, and Operations Research (OR) through a Fuzzy Interval Mathematical model was developed. The results indicated that combining both models provides an effective assessment under scarce initial information to select a suitable list of patients for a set of hospitals. The proposed approach allows to achieve a key goal: minimizing death rates under each hospital constraints of available resources. Furthermore, there is a serious concern regarding the resurgence of the COVID-19 virus which could cause a more severe pandemic. Thus, the main outcome of this study is the application of the above-mentioned approaches, especially when combining them, as efficient tools serving health establishments to manage critical resources. In Mars 2020, COVID-19 was declared as a pandemic disease [1] . Although very limited 34 clinical information is available about the causing virus SARS-CoV-2 [2], the mortality rate 35 of the disease was estimated at 5.7% [3] . In fact, COVID-19 presents higher risks to elderly 36 people and those with underlying diseases such as cardiac dysfunctions and obesity [1, 4] . As 37 a result of the rapid increase of COVID-19 patients influx and hospitals limited capacity, the 38 general principles became closer to those used in wars by venerating quick healings. Since World War I, when the medical services of the French army adopted a protocol to manage 40 injured soldiers, that the triage procedure has been continuously improving its efficiency in 41 sorting, classifying and distributing sick and injured patients to medical staff [5] . In 2020, 42 Italian healthcare workers facing increased numbers of COVID-19 patients have discussed a 43 potential age limit to access to medical care [6] . The medical personnel were actually unable 44 to respond to all requests due to resources constraints, e.g. lack of ventilators needed for the 45 most critical cases. The typical symptoms of the disease include fever, cough and breathing difficulty. A In fact, the Prioritization and Scheduling Patients (PSP) problem is usually considered as a 53 complex and combinatorial problem [1] . The PSP is conducted to provide health care/services 54 for each patient in due course. Improper patient prioritization can lead to incorrect strategic Luscombe and Kozan [35] made an exception to this rule as the authors assumed that each 138 emergency patient incoming after triage receipts a specific treatment before taking an 139 additional treatment decision. Overall, many problems resulting from the COVID-19 pandemic require each one a different The problem can be figured as follows: For each COVID-19 patient, a priority level (for each parameter ± , ± and ± ) is assigned 207 by the hospital based on his/her level of situation. In a third step, this paper discusses the use of a combined approach to solve a multi-objective In this paper, the PSCOVP problem was studied to provide decision makers with a reliable 214 tool to select patients (similarly to the principles of wartime triage), to achieve the main goals 215 of lowest death rates and maximum survival rates, under the restrictions of hospitals' 216 available capacities (Fig 2) . were used via MATLAB software. "tansig" was the applied activation function and the 240 network training function was gradient descent w/momentum & adaptive lr back propagation. The back propagation algorithm involved three layers: an input layer formed by 6 neurons, a 242 hidden layer covering 10 neurons and two output layers (Fig.3) . Each one of these layers 243 contains one node or more [49] . The hyperbolic tangent function is defined as: In ANN modelling, the activation function is important to manipulate complicated situations. Its main role is to convert input signals to outputs. The output signal is used as an 'input' to Equation (7) of the FIM model presents the bed capacity constraint. It ensures that the bed 301 COVID-19 hospital capacity is respected. Constraint (10) The objective allows integrating the estimated need for ventilation using the ANN method. The objective allows integrating the patients estimated health status using the ANN method. The objective functions (11 and 12) seek to minimize the total weighted overachievement for 348 goals and . weights on the proposed model. Results are shown in Table SD 2 and Table 4 . Unsuccessful decision 0 (00%) 0 (00%) 0 (00%) 0 (00%) 0 (00%) 0 (00%) 0 (00%) Word-Health-Organization Stability of SARS-CoV-2 in different environmental conditions. The 513 Lancet Microbe Real estimates of 515 mortality following COVID-19 infection. The Lancet Infectious Diseases The epidemiological characteristics of an outbreak of 2019 novel 517 coronavirus diseases (COVID-19)-China Triage and management of the injured in World War I: The diuturnity of 519 Craniomaxillofacial trauma & reconstruction, 1(01) The Extraordinary Decisions Facing Italian Doctors Ventilator-associated pneumonia in patients with 524 SARS-CoV-2-associated acute respiratory distress syndrome requiring ECMO: a retrospective 525 cohort study Impact of non-527 pharmaceutical interventions (NPIs) to reduce COVID-19 mortality and healthcare demand, 528 Imperial College COVID-19 Response Team Real-530 time measurement of the uncertain epidemiological appearances of COVID-19 infections Data Science 533 Techniques for COVID-19 in Intensive Care Units COVID-19 Detection in Chest X-ray Images using a Deep 537 Learning Approach Multi-objective permutation flow shop scheduling 540 problem: Literature review, classification and current trends FMS short term planning problems: a review. Manufacturing Research and 542 Technology Health expenditure Operational research in the management of the operating theatre: a 545 survey Optimizing contracted resource capacity with two advance 547 cancelation modes Optimal advance scheduling Scheduling operating rooms: achievements, challenges and pitfalls An integer programming approach to elective surgery 553 scheduling Addressing conflicting stakeholders' priorities in 555 surgical scheduling by goal programming Solving the operating room scheduling problem with 558 prioritized lists of patients Different stakeholders' perspectives for a surgical case 560 assignment problem: Deterministic and robust approaches Scheduling elective surgery under uncertainty and downstream capacity 563 constraints Appointment scheduling of oncology outpatients Dynamic scheduling with due dates and time windows: an 567 application to chemotherapy patient appointment booking Dynamic optimization of chemotherapy outpatient scheduling with uncertainty. Health care 571 management science Prioritizations of individual surgeons' patients waiting for 573 elective procedures: A systematic review and future directions. Perioperative Care and Operating 574 Room Management Adaptive appointment scheduling for patient-centered 576 medical homes Minimizing Bed 578 Occupancy Variance by Scheduling Patients under Uncertainty Evaluation of 581 appointment scheduling rules: A multi-performance measurement approach An integrated approach to demand and capacity planning in outpatient 583 clinics Literature review on multi-appointment scheduling 585 problems in hospitals A Conceptual Framework for Improving Critical Care 587 Patient Flow and Bed Use Simulation of va hospital length of stay 589 for analyzing additional inpatient bed capacity Dynamic resource allocation to improve emergency 592 department efficiency in real time How simulation modelling can help reduce the impact of COVID-19 Sample-Efficient Deep 598 Learning for COVID-19 Diagnosis Based on CT Scans pengaruh perubahan waktu gilir kerja sebagai dampak covid-19 601 terhadap kinerja karyawan (studi kasus : pt. nusa halmahera mineral, maluku utara). jurnal 602 manajemen pendidikan dan ilmu sosial A Framework for Rationing Ventilators and Critical Care Beds 604 During the COVID-19 Pandemic Hospital Capacity and 606 Operations in the Coronavirus Disease 2019 (COVID-19) Pandemic-Planning for the Nth 607 Patient A mathematical model for the novel coronavirus epidemic in 609 Prognostic medication: prediction by 611 a macroscopic equation model for actual medical histories of illness with various recovery 612 speeds Prediction of Recovery From Severe 614 Hemorrhagic Shock Using Logistic Regression. IEEE journal of translational engineering in 615 health and medicine Building predictive models for MERS-617 CoV infections using data mining techniques Potential Effects of Coronaviruses 620 on the Cardiovascular System: A Review Pivot table data crunching. Indianapolis: Que Ultrasonic velocity as a tool for 625 geotechnical parameters prediction within carbonate rocks aggregates An example of artificial neural network (ANN) application 628 for indirect estimation of rock parameters DEANN: A healthcare analytic 631 methodology of data envelopment analysis and artificial neural networks for the prediction of 632 organ recipient functional status Comparative Assessment of the Hybrid 634 Genetic Algorithm-Artificial Neural Network and Genetic Programming Methods for the 635 Prediction of Longitudinal Velocity Field around a Single Straight Groyne. Applied Soft 636 Computing A Meta-Goal Programming Approach to Cardinal 638 Preferences Aggregation in Multicriteria Problems Modified two-phase fuzzy goal programming integrated with 640 IF-TOPSIS for green supplier selection The authors received no funding for this research and have no further conflicts of interest. Any errors and omissions are the responsibility of the authors. The usual disclaimer applies. final situation (death or survival). To overcome this problem, this work presents a model able 484 to solve multi-objective problems with priority objectives. First, the ANN procedure was used 485 to predict the death or recovery of the patients suffering from COVID-19 and their need for