Reptile Identification for Endemic and Invasive Alien Species Using Transfer Learning Approaches.

The Canary Islands are considered a hotspot of biodiversity and have high levels of endemicity, including endemic reptile species. Nowadays, some invasive alien species of reptiles are proliferating with no control in different parts of the territory, creating a dangerous situation for the ecosystems of this archipelago. Despite the fact that the regional authorities have initiated actions to try to control the proliferation of invasive species, the problem has not been solved as it depends on sporadic sightings, and it is impossible to determine when these species appear. Since no studies for automatically identifying certain species of reptiles endemic to the Canary Islands have been found in the current state-of-the-art, from the Signals and Communications Department of the Las Palmas de Gran Canaria University (ULPGC), we consider the possibility of developing a detection system based on automatic species recognition using deep learning (DL) techniques. So this research conducts an initial identification study of some species of interest by implementing different neural network models based on transfer learning approaches. This study concludes with a comparison in which the best performance is achieved by integrating the EfficientNetV2B3 base model, which has a mean Accuracy of 98.75%.

Sky Image Classification Based on Transfer Learning Approaches.

Cloudy conditions at a local scale pose a significant challenge for forecasting renewable energy generation through photovoltaic panels. Consequently, having real-time knowledge of sky conditions becomes highly valuable. This information could inform decision-making processes in system operations, such as determining whether conditions are favorable for activating a standalone system requiring a minimum level of radiation or whether sky conditions might lead to higher energy consumption than generation during adverse cloudy conditions. This research leveraged convolutional neural networks (CNNs) and transfer learning (TL) classification techniques, testing various architectures from the EfficientNet family and two ResNet models for classifying sky images. Cross-validation methods were applied across different experiments, where the most favorable outcome was achieved with the EfficientNetV2-B1 and EfficientNetV2-B2 models boasting a mean Accuracy of 98.09%. This study underscores the efficacy of the architectures employed for sky image classification, while also highlighting the models yielding the best results.

Robust Identification System for Spanish Sign Language Based on Three-Dimensional Frame Information.

Nowadays, according to the World Health Organization (WHO), of the world's population suffers from a hearing disorder that makes oral communication with other people challenging. At the same time, in an era of technological evolution and digitization, designing tools that could help these people to communicate daily is the base of much scientific research such as that discussed herein. This article describes one of the techniques designed to transcribe Spanish Sign Language (SSL). A Leap Motion volumetric sensor has been used in this research due to its capacity to recognize hand movements in 3 dimensions. In order to carry out this research project, an impaired hearing subject has collaborated in the recording of 176 dynamic words. Finally, for the development of the research, Dynamic Time Warping (DTW) has been used to compare the samples and predict the input with an accuracy of 95.17%.

Information and Communication Technologies in Patients With Immune-Mediated Inflammatory Diseases: Cross-sectional Survey.

BACKGROUND: Information and communication technologies (ICTs) are changing the traditional health care model and redefining personalized health. ICTs offer effective communication and real-time monitoring of patients and provide additional data to support clinical decision-making, improve the quality of care, and contribute to the empowerment of patients. However, evidence on the use of ICTs and digital preferences of immune-mediated inflammatory disease (IMID) patients is scarce. OBJECTIVE: The aim of this study is to describe the degree of use of ICTs in patients with IMIDs (including rheumatic diseases, inflammatory bowel diseases, and psoriasis), identify their needs, and analyze their interest in the use of apps as tools for better management of their disease. METHODS: A questionnaire was created by a multidisciplinary team including pharmacists, rheumatologists, gastroenterologists, dermatologists, and nurses with experience in ICTs applied to the field of IMID. The survey included 27 questions organized into 3 blocks: (1) sociodemographic characteristics, (2) ICT use for health-related information, and (3) patient expectations about mobile health. RESULTS: A total of 472 questionnaires were analyzed. Overall, 52.9% (250/472) of patients were diagnosed with a rheumatologic disease, 39.4% (186/472) with inflammatory bowel disease, and 12.3% (58/472) with psoriasis. The state of health was considered good by 45.6% (215/472) of patients. Patients were interested in staying informed about health issues in 86.9% (410/427) of cases and sought health-related information mainly from the internet (334/472, 70.8%) and health care professionals (318/472, 67.4%). Overall, 13.6% (64/472) did not trust the health information they found in internet. Of the patients, 42.8% (202/472) had a health app, and 42.2% (199/472) had found it on their own. Patients would like a health app to help mainly to manage appointments (281/472, 59.5%), obtain information about their diseases and treatments (274/472, 58.1%), and get in contact with health professionals (250/472, 53.0%). Overall, 90.0% (425/472) of patients reported they would use an app to manage their IMID if their health professional recommended it, and 58.0% (274/472) would pay or probably be willing to pay for it. CONCLUSIONS: IMID patients were very interested in finding health-related information via ICTs, especially using smartphones and apps recommended by health professionals. Appointment management, advice on disease and treatment management, and personalized communication with health professionals were the most desired app features identified. Health professionals should play an essential role in recommending and validating these tools to ensure they are of high quality.

Analysis of the Nosema Cells Identification for Microscopic Images.

The use of image processing tools, machine learning, and deep learning approaches has become very useful and robust in recent years. This paper introduces the detection of the Nosema disease, which is considered to be one of the most economically significant diseases today. This work shows a solution for recognizing and identifying Nosema cells between the other existing objects in the microscopic image. Two main strategies are examined. The first strategy uses image processing tools to extract the most valuable information and features from the dataset of microscopic images. Then, machine learning methods are applied, such as a neural network (ANN) and support vector machine (SVM) for detecting and classifying the Nosema disease cells. The second strategy explores deep learning and transfers learning. Several approaches were examined, including a convolutional neural network (CNN) classifier and several methods of transfer learning (AlexNet, VGG-16 and VGG-19), which were fine-tuned and applied to the object sub-images in order to identify the Nosema images from the other object images. The best accuracy was reached by the VGG-16 pre-trained neural network with 96.25%.

Perspective and Evolution of Gesture Recognition for Sign Language: A Review.

This review analyses the different gesture recognition systems through a timeline, showing the different types of technology, and specifying which are the most important features and their achieved recognition rates. At the end of the review, Leap Motion sensor possibilities are described in detail, in order to consider its application on the field of sign language. This device has many positive characteristics that make it a good option for sign language. One of the most important conclusions is the ability of the Leap Motion sensor to provide 3D information from the hands for due identification.

Efficient Multi-Object Detection and Smart Navigation Using Artificial Intelligence for Visually Impaired People.

Visually impaired people face numerous difficulties in their daily life, and technological interventions may assist them to meet these challenges. This paper proposes an artificial intelligence-based fully automatic assistive technology to recognize different objects, and auditory inputs are provided to the user in real time, which gives better understanding to the visually impaired person about their surroundings. A deep-learning model is trained with multiple images of objects that are highly relevant to the visually impaired person. Training images are augmented and manually annotated to bring more robustness to the trained model. In addition to computer vision-based techniques for object recognition, a distance-measuring sensor is integrated to make the device more comprehensive by recognizing obstacles while navigating from one place to another. The auditory information that is conveyed to the user after scene segmentation and obstacle identification is optimized to obtain more information in less time for faster processing of video frames. The average accuracy of this proposed method is 95.19% and 99.69% for object detection and recognition, respectively. The time complexity is low, allowing a user to perceive the surrounding scene in real time.

Genetic variation at the glycosaminoglycan metabolism pathway contributes to the risk of psoriatic arthritis but not psoriasis.

OBJECTIVE: Psoriatic arthritis (PsA) is a chronic inflammatory arthritis affecting up to 30% of patients with psoriasis (Ps). To date, most of the known risk loci for PsA are shared with Ps, and identifying disease-specific variation has proven very challenging. The objective of the present study was to identify genetic variation specific for PsA. METHODS: We performed a genome-wide association study in a cohort of 835 patients with PsA and 1558 controls from Spain. Genetic association was tested at the single marker level and at the pathway level. Meta-analysis was performed with a case-control cohort of 2847 individuals from North America. To confirm the specificity of the genetic associations with PsA, we tested the associated variation using a purely cutaneous psoriasis cohort (PsC, n=614) and a rheumatoid arthritis cohort (RA, n=1191). Using network and drug-repurposing analyses, we further investigated the potential of the PsA-specific associations to guide the development of new drugs in PsA. RESULTS: We identified a new PsA risk single-nucleotide polymorphism at B3GNT2 locus (p=1.10e-08). At the pathway level, we found 14 genetic pathways significantly associated with PsA (pFDR<0.05). From these, the glycosaminoglycan (GAG) metabolism pathway was confirmed to be disease-specific after comparing the PsA cohort with the cohorts of patients with PsC and RA. Finally, we identified candidate drug targets in the GAG metabolism pathway as well as new PsA indications for approved drugs. CONCLUSION: These findings provide insights into the biological mechanisms that are specific for PsA and could contribute to develop more effective therapies.

Validity and reliability of the TargetScan ISSF Pistol & Rifle application for measuring shooting performance.

Several technologies, with varying economic costs, are used to measure the performance of Olympic sport shooters, including electronic targets, laser systems, and mobile phone applications. Nevertheless, the International Shooting Sport Federation (ISSF) does not currently accept any mobile applications as valid measurement systems. The objective of the study was to check the validity and reliability of the TargetScan ISSF Pistol & Rifle application, an automatic mobile application for measuring the performance of shooters via image analysis. A total of 1440 shots were fired by 24 shooters (720 shots each for the air rifle and air pistol events), and the target sheets were analyzed with the application installed on an iPhone X. The official competition results and the measurements from the application were then compared. The results showed an almost perfect reliability for air pistol shots (ICC = .999) and no significant differences between the official average scores and those measured by the application. The reliability was also high for air rifle shots (ICC = .998), but the average scores registered by the application were statistically lower than the official results. We concluded that TargetScan ISSF Pistol & Rifle is a low-cost, valid, and reliable application for measuring amateur and semi-professional shooting performance which may be useful to shooters, coaches, clubs, and federations worldwide.

Xenopus laevis as a Bioindicator of Endocrine Disruptors in the Region of Central Chile.

One of the direct causes of biodiversity loss is environmental pollution resulting from the use of chemicals. Different kinds of chemicals, such as persistent organic pollutants and some heavy metals, can be endocrine disruptors, which act at low doses over a long period of time and have a negative effect on the reproductive and thyroid system in vertebrates worldwide. Research on the effects of endocrine disruptors and the use of bioindicators in neotropical ecosystems where pressure on biodiversity is high is scarce. In Chile, although endocrine disruptors have been detected at different concentrations in the environments of some ecosystems, few studies have been performed on their biological effects in the field. In this work, Xenopus laevis (African clawed frog), an introduced species, is used as a bioindicator for the presence of endocrine disruptors in aquatic systems with different degrees of contamination in a Mediterranean zone in central Chile. For the first time for Chile, alterations are described that can be linked to exposure to endocrine disruptors, such as vitellogenin induction, decreased testosterone in male frogs, and histological changes in gonads. Dioxin-like and oestrogenic activity was detected in sediments at locations where it seem to be related to alterations found in the frogs. In addition, an analysis of land use/cover use revealed that urban soil was the best model to explain the variations in frog health indicators. This study points to the usefulness of an invasive species as a bioindicator for the presence of endocrine-disruptive chemicals.

Identifying markers of sustained remission in rheumatoid arthritis patients on long-term tapered biological disease-modifying antirheumatic drugs.

To identify features associated with long-term persistent remission in rheumatoid arthritis (RA) patients on tapered biological disease-modifying antirheumatic drugs (bDMARD) (tap-bDMARD) therapy. We carried out a 40-month (m) extension follow-up study of 77 RA patients from a previous 12 m tap-bDMARD study. Disease activity was assessed at baseline and every 3 months. Doppler US investigation of 42 joints for the presence and grade (0-3) of B-mode synovial hypertrophy (SH) and synovial power Doppler signal (i.e., Doppler synovitis) was performed before starting the tap-bDMARD strategy by a rheumatologist blinded to clinical and laboratory data. At the 40 m mark, 44 (57.1%) patients failed the tap-bDMARD strategy, while 33 (42.9%) succeeded. Patients who presented a failed tap-bDMARD had significantly longer disease duration, a longer time from symptom onset to synthetic (s) DMARD start, longer duration of sDMARD treatment, a greater number of sDMARDs, and a higher baseline DAS28 and SDAI than patients with successful tap-bDMARD at 40 months. In logistic regression analysis, the presence of baseline Doppler synovitis, a DAS28 ≥ 2.2, and the presence of rheumatoid factor were identified as predictors of tap-bDMARD failure at 40 m. In those patients who succeed tap-bDMARD at 12 m, a smoking habit was significantly more frequently found in tap-bDMARD failures at 40 m. Our results showed that DAS28 and the presence of Doppler synovitis, RF and a smoking habit predicted long-term tap-bDMARD failure.

The feet in systemic lupus erythematosus; are we underestimating their involvement and functional impact?

OBJECTIVES: To evaluate biomechanical and ultrasound (US) abnormalities in SLE patients as compared with controls and to assess the relationship between these abnormalities and SLE activity. METHODS: Fifty-four consecutive female patients with SLE with and without foot pain and 60 female controls (30 with foot pain and 30 without foot pain) were recruited. SLE activity was assessed by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI). SLE patients and controls blindly underwent a comprehensive podiatric, biomechanical and US evaluation of the feet. US assessment included detection of B-mode synovitis, tenosynovitis, enthesopathy, bone changes and synovial, tenosynovial and entheseal power Doppler (PD) signal. RESULTS: Thirty-one (57.4%) SLE patients had bilateral foot pain and 5 (9.3%) had unilateral foot pain. Metatarsalgia was the most common location for pain but without significant difference between groups (p=0.284). Toe joint deformities were significantly more common in SLE feet as compared with control feet (p<0.0005). SLE feet showed significantly more biomechanical abnormalities than control feet (p<0.05). B-mode synovitis in the tibiotalar joint was strongly associated with having SLE (p<0.0005) and the presence of synovial PD signal in the MTP joints was found only in painful feet of SLE patients. SLEDAI was significantly higher in patients with foot pain than in those with painless feet (p=0.008). However, SLEDAI did not discriminate between patients with and without biomechanical or US abnormalities. CONCLUSIONS: SLE patients showed more biomechanical and US abnormalities in the feet than controls, which were not captured by standardised assessment of the disease activity.

Predictive value of Doppler ultrasound-detected synovitis in relation to failed tapering of biologic therapy in patients with rheumatoid arthritis.

OBJECTIVE: To investigate the predictive value of synovitis detected by Doppler US in relation to failed tapering of biologic therapy (BT) in RA patients in sustained clinical remission. METHODS: A total of 77 RA patients (52 women, 25 men) in sustained clinical remission, treated with a stable dosage of BT were prospectively recruited. BT was tapered according to an agreed strategy implemented in clinical practice (i.e. increasing the interval between doses for s.c. BT and reducing the dose for i.v. BT). BT tapering failure was assessed at 6 and 12 months. Doppler US investigation of 42 joints for the presence and grade (0-3) of B-mode synovial hypertrophy and synovial power Doppler signal (i.e. Doppler synovitis) was performed at baseline by a rheumatologist blinded to clinical and laboratory data. Hand and foot radiographs were obtained at baseline and at 12-month follow-up. RESULTS: Of the 77 patients, 46 (59.7%) were on s.c. BT and 31 (40.3%) on i.v. BT. At 12 months, 35 patients (45.5%) presented BT tapering failure, 23 of them (29.9% of all patients) in the first 6 months of BT tapering. In logistic regression analysis, the baseline DAS28 and the global score of Doppler synovitis were identified as independent predictors of BT tapering failure at 12 and 6 months. The presence of Doppler synovitis was the strongest predictor for BT tapering failure. No patient showed radiographic progression. CONCLUSION: Our results suggest that the presence of Doppler-detected synovitis may predict BT tapering failure in RA patients in sustained clinical remission.

Seasonal strength performance and its relationship with training load on elite runners.

The aim of this study was to analyze the time-course of force production of elite middle and long-distance runners throughout an entire season and at the end of the off-season, as well as its relationships with training load and hormonal responses. Training load was recorded daily throughout an entire season by measuring and evaluating the session distance (km), training zone and session-RPE in a group of 15 elite middle and long-distance runners (12 men, 3 women; age = 26.3 ± 5.1yrs, BMI = 19.7 ± 1.1). Also, basal salivary-free cortisol levels were measured weekly, and 50-metre sprints, mean propulsive velocity (MPV), mean propulsive power (MPP), repetition maximum (RM) and peak rate of force development (RFD) of half-squats were measured 4 times during the season, and once more after the off-season break. There were no significant variations in force production during the season or after the off-season break, except for the RFD (-30.2%, p = 0.005) values, which changed significantly from the beginning to the end of the season. Significant correlations were found between session-RPE and MPV (r = -0.650, p = 0.004), MPP (r = -0.602, p = 0.009), RM (r = -0.650, p = 0.004), and the 50-metre sprint (r = 0.560, p = 0.015). Meanwhile, salivary-free cortisol correlated significantly with the 50-metre sprint (r = 0.737, p < 0.001) and the RM ( r = -0.514, p = 0.025). Finally, the training zone correlated with the 50-metre sprint (r = -0.463, p = 0.041). Session-RPE, training zone and salivary-free cortisol levels are related to force production in elite middle and long-distance runners. Monitoring these variables could be a useful tool in controlling the training programs of elite athletes. Key pointsSession-RPE, training zone and salivary free cortisol levels correlate significantly with strength-related variables in middle and long-distance elite runners.A month of active rest during the off-season break is enough to prevent decreases in force production of such athletes.Monitoring training loads through session-RPE is a suitable and simple method for controlling the training process in elite middle and long-distance runners.

Does ultrasound-scored synovitis depend on the pharmacokinetics of subcutaneous anti-TNF agents in patients with rheumatoid arthritis?

OBJECTIVE: The aim of this study was to investigate the influence of the pharmacokinetics of s.c. anti-TNF agents on the grade of US-detected synovitis in RA patients. METHODS: Fifty RA patients were prospectively recruited from the Biologic Therapy Unit of our hospital. Inclusion criteria were being in treatment with s.c. anti-TNF agents and having had neither changes in therapy nor local corticosteroid injections in the previous 3 months. Patients underwent clinical, laboratory [28-joint DAS (DAS28) and Simplified Disease Activity Index (SDAI)] and US assessment at two time points, i.e. at peak plasma drug concentration and at trough plasma drug concentration. US assessments were performed blindly to the anti-TNF agent, the administration time and the clinical and laboratory data. Twenty-eight joints were investigated for the presence and grade (0-3) of B-mode synovitis and synovial power Doppler signal. Global indices for B-mode synovitis (BSI) and Doppler synovitis (DSI) were calculated for 12 joints and for wrist-hand-ankle-foot joints. B-mode US remission was defined as a BSI <1 and Doppler US remission as a DSI <1. RESULTS: There were no significant differences between the clinical, laboratory and B-mode and Doppler US parameters at peak time and trough time (P = 0.132-0.986). There were no significant differences between the proportion of patients with active disease and those in remission according to DAS28, SDAI, B-mode US and Doppler US at peak time and trough time assessments (P = 0.070-1). CONCLUSION: Our results suggested that s.c. anti-TNF pharmacokinetics do not significantly influence US-scored synovitis in RA patients.

Prevalence of leisure-time sedentary behaviour and sociodemographic correlates: a cross-sectional study in Spanish adults.

BACKGROUND: Being physically inactive has been linked to a higher mortality and poorer quality of life. This cross-sectional study examines the prevalence of leisure-time sedentary behaviour in a population of Spanish adults and its correlates with several sociodemographic variables. METHODS: Data were collected from 1,330 subjects living in Madrid (age: 18-65 years, 51.6% women) by telephone interview. The sampling error was ±2.7% for a 95.5% confidence level. Leisure-time sedentary behaviour was assessed using the Global Physical Activity Questionnaire (version 2). Further factors examined were: country of birth, sex, age, civil state, education level, employment and economic status and physical activity of parents. RESULTS: 76.3% of the subjects interviewed reported a mostly sedentary leisure-time lifestyle. The remaining subjects (23.7%) reported a moderate to high level of physical activity, meeting minimum recommendations. Logistic regression adjusted for all variables identified the following population subsets as showing a greater risk of sedentary behaviour: women (odds ratio (OR) = 2.14; 95% confidence interval (CI): 1.64, 2.79), participants aged 41-50 years (OR = 1.64; 95%CI:1.05, 2.51), those with a middle economic status (OR = 1.48; 95% CI: 1.04, 2.10) or not providing information about their income (OR = 1.97; 95% CI: 1.05, 3.67), and those whose father (OR = 1.53; 95% CI: 1.13, 2.07) and/or mother (OR = 1.41; 95% CI: 1.01, 1.97) were never physically active during leisure-time. CONCLUSIONS: The high prevalence of self-reported sedentary behaviour recorded suggests the need for public health policies targeted at increasing leisure-time physical activity levels. Our data identified several population subsets as priority candidates for possible interventions pursuing this goal.

Objectively measured physical activity has a negative but weak association with academic performance in children and adolescents.

AIM: There is an emerging body of evidence on the potential effects of regular physical activity on academic performance. The aim of this study was to add to the debate, by examining the association between objectively measured physical activity and academic performance in a relatively large sample of children and adolescents. METHODS: The Spanish UP & DOWN study is a 3-year longitudinal study designed to assess the impact, overtime, of physical activity and sedentary behaviours on health indicators. This present analysis was conducted with 1778 children and adolescents aged 6-18 years. Physical activity was objectively measured by accelerometry. Academic performance was assessed using school grades. RESULTS: Physical activity was inversely associated with all academic performance indicators after adjustment for potential confounders, including neonatal variables, fatness and fitness (all p < 0.05). This association became nonsignificant among quartiles of physical activity. There were only slight differences in academic performance between the lowest and the second quartile of physical activity, compared to the highest quartile, with very small effect size (d < 0.20). CONCLUSION: Objectively measured physical activity may influence academic performance during both childhood and adolescence, but this association was negative and very weak. Longitudinal and intervention studies are necessary to further our understanding.

Characteristics of extracurricular physical activity and cognitive performance in adolescents. The AVENA study.

The aim of this report was to examine the associations between some characteristics (i.e., structure and number) of extracurricular physical activity (EPA) and cognitive performance in adolescents. A total of 1662 adolescents (880 girls; 13.0-18.5 years) from 5 Spanish cities (Granada, Madrid, Murcia, Santander and Zaragoza) were included in this study. Structure (organised and non-organised) and number of EPAs, and participation at vigorous intensity during EPA were self-reported. Cognitive performance (verbal, numeric and reasoning abilities, and an overall score) was measured with the "SRA-Test of Educational Ability". Results showed that vigorous EPA was positively associated with all cognitive variables. Adolescents who practiced an organised EPA had higher scores in 3 of the 4 cognitive variables than those who practiced a non-organised EPA (all P < 0.05). Likewise, the group who participated in more than one EPA had higher cognitive performance in all variables than the group who participated in only one EPA (all P < 0.05). Regardless of potential confounder variables, including vigorous EPA, both structure and number of EPAs were each other independently associated with cognitive performance. Therefore, structure and number of EPAs may positively influence cognitive performance in adolescents. Participating in multiple, organised EPA may have benefits for cognitive performance.

The concurrent validity and reliability of a low-cost, high-speed camera-based method for measuring the flight time of vertical jumps.

Flight time is the most accurate and frequently used variable when assessing the height of vertical jumps. The purpose of this study was to analyze the validity and reliability of an alternative method (i.e., the HSC-Kinovea method) for measuring the flight time and height of vertical jumping using a low-cost high-speed Casio Exilim FH-25 camera (HSC). To this end, 25 subjects performed a total of 125 vertical jumps on an infrared (IR) platform while simultaneously being recorded with a HSC at 240 fps. Subsequently, 2 observers with no experience in video analysis analyzed the 125 videos independently using the open-license Kinovea 0.8.15 software. The flight times obtained were then converted into vertical jump heights, and the intraclass correlation coefficient (ICC), Bland-Altman plot, and Pearson correlation coefficient were calculated for those variables. The results showed a perfect correlation agreement (ICC = 1, p < 0.0001) between both observers' measurements of flight time and jump height and a highly reliable agreement (ICC = 0.997, p < 0.0001) between the observers' measurements of flight time and jump height using the HSC-Kinovea method and those obtained using the IR system, thus explaining 99.5% (p < 0.0001) of the differences (shared variance) obtained using the IR platform. As a result, besides requiring no previous experience in the use of this technology, the HSC-Kinovea method can be considered to provide similarly valid and reliable measurements of flight time and vertical jump height as more expensive equipment (i.e., IR). As such, coaches from many sports could use the HSC-Kinovea method to measure the flight time and height of their athlete's vertical jumps.

Novel cost-effective method for forecasting COVID-19 and hospital occupancy using deep learning.

The emergence of the COVID-19 pandemic in 2019 and its rapid global spread put healthcare systems around the world to the test. This crisis created an unprecedented level of stress in hospitals, exacerbating the already complex task of healthcare management. As a result, it led to a tragic increase in mortality rates and highlighted the urgent need for advanced predictive tools to support decision-making. To address these critical challenges, this research aims to develop and implement a predictive system capable of predicting pandemic evolution with accuracy (in terms of Mean Absolute error (MAE), Root Mean Square Error (RMSE), R2, and Mean Absolute Percentage Error (MAPE)) and low computational and economic cost. It uses a set of interconnected Long Short Term-memory (LSTM) with double bidirectional LSTM (BiLSTM) layers together with a novel preprocessing based on future time windows. This model accurately predicts COVID-19 cases and hospital occupancy over long periods of time using only 40% of the set to train. This results in a long-term prediction where each day we can query the cases for the next three days with very little data. The data utilized in this analysis were obtained from the "Hospital Insular" in Gran Canaria, Spain. These data describe the spread of the coronavirus disease (COVID-19) from its initial emergence in 2020 until March 29, 2022. The results show an improvement in MAE (< 161), RMSE (< 405), and MAPE (> 0.20) compared to other studies with similar conditions. This would be a powerful tool for the healthcare system, providing valuable information to decision-makers, allowing them to anticipate and strategize for possible scenarios, ultimately improving public health outcomes and optimizing the allocation of healthcare and economic resources.