RESUMO
BACKGROUND: An artificial-intelligence (AI) model for predicting the prognosis or mortality of coronavirus disease 2019 (COVID-19) patients will allow efficient allocation of limited medical resources. We developed an early mortality prediction ensemble model for COVID-19 using AI models with initial chest X-ray and electronic health record (EHR) data. RESULTS: We used convolutional neural network (CNN) models (Inception-ResNet-V2 and EfficientNet) for chest X-ray analysis and multilayer perceptron (MLP), Extreme Gradient Boosting (XGBoost), and random forest (RF) models for EHR data analysis. The Gradient-weighted Class Activation Mapping and Shapley Additive Explanations (SHAP) methods were used to determine the effects of these features on COVID-19. We developed an ensemble model (Area under the receiver operating characteristic curve of 0.8698) using a soft voting method with weight differences for CNN, XGBoost, MLP, and RF models. To resolve the data imbalance, we conducted F1-score optimization by adjusting the cutoff values to optimize the model performance (F1 score of 0.77). CONCLUSIONS: Our study is meaningful in that we developed an early mortality prediction model using only the initial chest X-ray and EHR data of COVID-19 patients. Early prediction of the clinical courses of patients is helpful for not only treatment but also bed management. Our results confirmed the performance improvement of the ensemble model achieved by combining AI models. Through the SHAP method, laboratory tests that indicate the factors affecting COVID-19 mortality were discovered, highlighting the importance of these tests in managing COVID-19 patients.
Assuntos
COVID-19 , Aprendizado Profundo , Humanos , Registros Eletrônicos de Saúde , COVID-19/diagnóstico por imagem , Raios X , Inteligência ArtificialRESUMO
BACKGROUND: Paralysis of medical systems has emerged as a major problem not only in Korea but also globally because of the COVID-19 pandemic. Therefore, early identification and treatment of COVID-19 are crucial. This study aims to develop a machine-learning algorithm based on bio-signals that predicts the infection three days in advance before it progresses from mild to severe, which may necessitate high-flow oxygen therapy or mechanical ventilation. METHODS: The study included 2758 hospitalized patients with mild severity COVID-19 between July 2020 and October 2021. Bio-signals, clinical information, and laboratory findings were retrospectively collected from the electronic medical records of patients. Machine learning methods included random forest, random forest ranger, gradient boosting machine, and support vector machine (SVM). RESULTS: SVM showed the best performance in terms of accuracy, kappa, sensitivity, detection rate, balanced accuracy, and run-time; the area under the receiver operating characteristic curve was also quite high at 0.96. Body temperature and SpO2 three and four days before discharge or exacerbation were ranked high among SVM features. CONCLUSIONS: The proposed algorithm can predict the exacerbation of severity three days in advance in patients with mild COVID-19. This prediction can help effectively manage the reallocation of appropriate medical resources in clinical settings. Therefore, this algorithm can facilitate adequate oxygen therapy and mechanical ventilator preparation, thereby improving patient prognosis, increasing the efficiency of medical systems, and mitigating the damage caused by a global pandemic.
Assuntos
COVID-19 , Humanos , Estudos Retrospectivos , Pandemias , Morbidade , Aprendizado de Máquina , Algoritmos , OxigênioRESUMO
CONTEXT: Wall push-up plus (WPP) and wall slide (WS) are commonly prescribed in early rehabilitation to increase serratus anterior (SA) muscle activity. For individuals with scapular winging (SW), synergistic muscles such as upper trapezius (UT) and pectoralis major (PM) may compensate for weak SA during scapular movement. However, no studies have applied isometric horizontal abduction (IHA) during WS in individuals with SW nor have compared it with WPP with IHA. OBJECTIVES: This study compared the effects of IHA on shoulder muscle activity during WPP and WS exercises in individuals with SW. DESIGN: Cross-sectional study; 2-way repeated-measures analysis of variance was used to assess the statistical significance of observed differences in SA, UT, PM, lower trapezius (LT), and infraspinatus (IS) muscle activities. SETTING: Research laboratory. PATIENTS: We recruited 30 individuals with SW comprising 20 men and 10 women. INTERVENTIONS: The individuals performed WPP and WS exercises with and without IHA using Thera-Band. MAIN OUTCOME MEASURES: Surface electromyography was used to measure muscle activity of the SA, UT, PM, LT, and IS during the isometric phase of WPP and WS. Maximal voluntary isometric contractions were recorded to normalize electromyographic data. RESULTS: There was no significant interaction between IHA application and exercise type for any of the shoulder muscles. IHA application increased SA (P = .008), UT (P = .001), LT (P = .009), and IS (P = .000) activities and decreased PM (P = .001) activity compared with those without IHA. WS exercises elicited higher PM (P = .017) and LT (P = .011) activities than WPP. CONCLUSION: WPP and WS with IHA may be effective in increasing the muscle activities of shoulder stabilizers and preventing overactivation of PM activity. WPP may be recommended for individuals with overactivated PM, whereas WS may be used to increase LT activity.
Assuntos
Músculo Esquelético , Ombro , Masculino , Humanos , Feminino , Ombro/fisiologia , Estudos Transversais , Músculo Esquelético/fisiologia , Escápula/fisiologia , EletromiografiaRESUMO
CONTEXT: Side bridge exercises strengthen the hip, trunk, and abdominal muscles and challenge the trunk muscles without the high lumbar compression associated with trunk extension or curls. Previous research using electromyography (EMG) reports that performance of the side bridge exercise highly activates the gluteus medius (Gmed). However, to the best of our knowledge, no previous research has investigated EMG amplitude in the hip and trunk muscles during side bridge exercise in subjects with Gmed weakness. OBJECTIVE: The purpose of this study was to examine the EMG activity of the hip and trunk muscles during 3 variations of the side bridge exercise (side bridge, side bridge with knee flexion, and side bridge with knee flexion and hip abduction of the top leg) in subjects with Gmed weakness. DESIGN: Repeated-measures experimental design. SETTING: Research laboratory. PATIENTS: Thirty subjects (15 females and 15 males) with Gmed weakness participated in this study. INTERVENTION: Each subject performed 3 variations of the side bridge exercise in random order. MAIN OUTCOME MEASURES: Surface EMG was used to measure the muscle activities of the rectus abdominis, external oblique, longissimus thoracis, multifidus, Gmed, gluteus maximus, and tensor fasciae latae (TFL), and Gmed/TFL muscle activity ratio during 3 variations of the side bridge exercise. RESULTS: There were significant differences in Gmed (F2,56 = 110.054, P < .001), gluteus maximus (F2,56 = 36.416, P < .001), and TFL (F2,56 = 108.342, P < .001) muscles among the 3 side bridge exercises. There were significant differences in the Gmed/TFL muscle ratio (F2,56 = 20.738, P < .001). CONCLUSION: Among 3 side bridge exercises, the side bridge with knee flexion may be effective for the individuals with Gmed weakness among 3 side bridge exercises to strengthen the gluteal muscles, considering the difficulty of the exercise and relative contribution of Gmed and TFL.
Assuntos
Músculo Esquelético , Coxa da Perna , Nádegas , Eletromiografia , Terapia por Exercício , Feminino , Articulação do Quadril , Humanos , MasculinoRESUMO
CONTEXT: The improvement of hip joint stability can significantly impact knee and rearfoot mechanics. Individuals with pes planus have a weak abductor hallucis (AbdH), and the tibialis anterior (TA) may activate to compensate for this. As yet, no studies have applied isometric hip abduction (IHA) for hip stability during short-foot exercise (SFE). OBJECTIVE: To compare the effects of IHA on the muscle activity of the AbdH, TA, peroneus longus (PL), and gluteus medius (Gmed), as well as the medial longitudinal arch (MLA) angle during sitting and standing SFE. DESIGN: Two-way repeated analyses of variance were used to determine the statistical significance of AbdH, TA, PL, and Gmed electromyography activity, as well as the change in MLA angle. SETTING: University research laboratory. PARTICIPANTS: Thirty-two participants with pes planus. INTERVENTION(S): The participants performed SFE with and without isometric hip abduction in sitting and standing positions. MAIN OUTCOME MEASURES: Surface electromyography was used to measure the activity of the AbdH, TA, PL, and Gmed muscles, and Image J was used to measure the MLA angle. RESULTS: Significant interactions between exercise type and position were observed in terms of the PL muscle activity and in the change in MLA angle only, while other muscles showed significant main effects. The IHA during SFE significantly increased the AbdH muscle activity, while the TA muscle activity was significantly lower. The muscle activity of Gmed and PL was significantly increased in the standing position compared with sitting, but there was no significant difference with or without IHA. The change in the MLA angle was significantly greater in SFE with IHA in a standing position than in the other SFE conditions. CONCLUSIONS: IHA may be an effective method for reducing compensatory TA activity and increasing AbdH muscle activity during SFE for individuals with pes planus.
Assuntos
Nádegas/fisiologia , Terapia por Exercício/métodos , Pé Chato/fisiopatologia , Pé Chato/terapia , Força Muscular/fisiologia , Músculo Esquelético/fisiologia , Adulto , Eletromiografia , Feminino , Humanos , Contração Isométrica , Masculino , Adulto JovemRESUMO
Background: The recent use of artificial intelligence (AI) in medical research is noteworthy. However, most research has focused on medical imaging. Although the importance of laboratory tests in the clinical field is acknowledged by clinicians, they are undervalued in medical AI research. Our study aims to develop an early prediction AI model for pneumonia mortality, primarily using laboratory test results. Materials and methods: We developed a mortality prediction model using initial laboratory results and basic clinical information of patients with pneumonia. Several machine learning (ML) models and a deep learning method-multilayer perceptron (MLP)-were selected for model development. The area under the receiver operating characteristic curve (AUROC) and F1-score were optimized to improve model performance. In addition, an ensemble model was developed by blending several models to improve the prediction performance. We used 80,940 data instances for model development. Results: Among the ML models, XGBoost exhibited the best performance (AUROC = 0.8989, accuracy = 0.88, F1-score = 0.80). MLP achieved an AUROC of 0.8498, accuracy of 0.86, and F1-score of 0.75. The performance of the ensemble model was the best among the developed models, with an AUROC of 0.9006, accuracy of 0.90, and F1-score of 0.81. Several laboratory tests were conducted to identify risk factors that affect pneumonia mortality using the "Feature importance" technique and SHapley Additive exPlanations. We identified several laboratory results, including systolic blood pressure, serum glucose level, age, aspartate aminotransferase-to-alanine aminotransferase ratio, and monocyte-to-lymphocyte ratio, as significant predictors of mortality in patients with pneumonia. Conclusions: Our study demonstrates that the ensemble model, incorporating XGBoost, CatBoost, and LGBM techniques, outperforms individual ML and deep learning models in predicting pneumonia mortality. Our findings emphasize the importance of integrating AI techniques to leverage laboratory test data effectively, offering a promising direction for advancing AI applications in medical research and clinical decision-making.
RESUMO
To assess the diagnostic utility of bone turnover markers (BTMs) and demographic variables for identifying individuals with osteoporosis. A cross-sectional study involving 280 participants was conducted. Serum BTM values were obtained from 88 patients with osteoporosis and 192 controls without osteoporosis. Six machine learning models, including extreme gradient boosting (XGBoost), light gradient boosting machine (LGBM), CatBoost, random forest, support vector machine, and k-nearest neighbors, were employed to evaluate osteoporosis diagnosis. The performance measures included the area under the receiver operating characteristic curve (AUROC), F1-score, and accuracy. After AUROC optimization, LGBM exhibited the highest AUROC of 0.706. Post F1-score optimization, LGBM's F1-score was improved from 0.50 to 0.65. Combining the top three optimized models (LGBM, XGBoost, and CatBoost) resulted in an AUROC of 0.706, an F1-score of 0.65, and an accuracy of 0.73. BTMs, along with age and sex, were found to contribute significantly to osteoporosis diagnosis. This study demonstrates the potential of machine learning models utilizing BTMs and demographic variables for diagnosing preexisting osteoporosis. The findings highlight the clinical relevance of accessible clinical data in osteoporosis assessment, providing a promising tool for early diagnosis and management.
Assuntos
Biomarcadores , Aprendizado de Máquina , Osteoporose , Humanos , Osteoporose/diagnóstico , Feminino , Masculino , Estudos Transversais , Biomarcadores/sangue , Pessoa de Meia-Idade , Idoso , Remodelação Óssea/fisiologia , Curva ROCRESUMO
The aim of this study was to investigate the additional effects of cerebellar rTMS on the motor recovery of facilitatory rTMS over affected primary motor cortex (M1) in subacute stroke patients. Twenty-eight subacute stroke patients were recruited in this single-blind, randomized, controlled trial. The Cr-Cbll group received Cr-Cbll rTMS stimulation consisting of high-frequency rTMS over affected M1 (10 min), motor training (10 min), and high-frequency rTMS over contralesional Cbll (10 min). The Cr-sham group received sham rTMS instead of high-frequency rTMS over the cerebellum. Ten daily sessions were performed for 2 weeks. A Fugl-Meyer Assessment (FMA) was measured before (T0), immediately after (T1), and 2 months after the intervention (T2). A total of 20 participants (10 in the Cr-Cbll group and 10 in the Cr-sham group) completed the intervention. There was no significant difference in clinical characteristics between the two groups at T0. FMA was significantly improved after the intervention in both Cr-Cbll and Cr-sham groups (p < 0.05). However, there was no significant interaction in FMA between time and group. In conclusion, these results could not demonstrate that rTMS over the contralesional cerebellum has additional effects to facilitatory rTMS over the affected M1 for improving motor function in subacute stroke patients.
RESUMO
Objective: Acute leukemia (AL) is a life-threatening malignant disease that occurs in the bone marrow and blood, and is classified as either acute myeloid leukemia (AML) or acute lymphoblastic leukemia (ALL). Diagnosing AL warrants testing methods, such as flow cytometry, which require trained professionals, time, and money. We aimed to develop a model that can classify peripheral blood images of 12 cell types, including pathological cells associated with AL, using artificial intelligence. Methods: We acquired 42,386 single-cell images of peripheral blood slides from 282 patients (82 with AML, 40 with ALL, and 160 with immature granulocytes). Results: The performance of EfficientNet-V2 (B2) using the original image size exhibited the greatest accuracy (accuracy, 0.8779; precision, 0.7221; recall, 0.7225; and F1 score, 0.7210). The next-best accuracy was achieved by EfficientNet-V1 (B1), with a 256 × 256 pixels image. F1 score was the greatest for EfficientNet-V1 (B1) with the original image size. EfficientNet-V1 (B1) and EfficientNet-V2 (B2) were used to develop an ensemble model, and the accuracy (0.8858) and F1 score (0.7361) were improved. The classification performance of the developed ensemble model for the 12 cell types was good, with an area under the receiver operating characteristic curve above 0.9, and F1 scores for myeloblasts and lymphoblasts of 0.8873 and 0.8006, respectively. Conclusions: The performance of the developed ensemble model for the 12 cell classifications was satisfactory, particularly for myeloblasts and lymphoblasts. We believe that the application of our model will benefit healthcare settings where the rapid and accurate diagnosis of AL is difficult.
RESUMO
Objective: Modern healthcare systems face challenges related to the stable and sufficient blood supply of blood due to shortages. This study aimed to predict the monthly blood transfusion requirements in medical institutions using an artificial intelligence model based on national open big data related to transfusion. Methods: Data regarding blood types and components in Korea from January 2010 to December 2021 were obtained from the Health Insurance Review and Assessment Service and Statistics Korea. The data were collected from a single medical institution. Using the obtained information, predictive models were developed, including eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LGBM), and category boosting (CatBoost). An ensemble model was created using these three models. Results: The prediction performance of XGBoost, LGBM, and CatBoost demonstrated a mean absolute error ranging from 14.6657 for AB+ red blood cells (RBCs) to 84.0433 for A+ platelet concentrate (PC) and a root mean squared error ranging from 18.5374 for AB+ RBCs to 118.6245 for B+ PC. The error range was further improved by creating ensemble models, wherein the department requesting blood was the most influential parameter affecting transfusion prediction performance for different blood products and types. Except for the department, the features that affected the prediction performance varied for each product and blood type, including the number of RBC antibody screens, crossmatch, nationwide blood donations, and surgeries. Conclusion: Based on blood-related open big data, the developed blood-demand prediction algorithm can efficiently provide medical facilities with an appropriate volume of blood ahead of time.
RESUMO
Recent advancements in endoscopic procedures have resulted in a growing diagnosis of early colorectal cancer (CRC) cases, where classical en bloc lymph node (LN) dissection is not performed and treatment is terminated with the removal of the main cancer lesion by endoscopy without pathologic LN staging. Although many studies report noninferior outcomes of endoscopic resection in comparison to surgical resection, a cautious approach to completing treatment with endoscopic resection alone is recommended because LN metastases may be present even in early-stage CRC. In most countries, including the United States, Europe, and South Korea, the guidelines for additional surgery after endoscopic resection are very similar. If LN metastasis is suspected, even in T1 stage or lower lesions, further surgery is an essential treatment modality, but confirmation of the presence of LN metastasis is perhaps the most difficult part of this process. Another paradoxical recent trend is the expansion of more extensive and complete surgical lymphadenectomy for CRC. The success rate of surgery has improved dramatically over the past decade with the introduction of surgical devices and minimally invasive surgery, and the associated risks have been significantly reduced. While the burden of surgery on patients is understandable, the indications for surgery in early colon cancer need to be carefully reviewed to improve cure rates. In this process, we believe that an integrated decision-making process with surgeons, radiologists, and pathologists, in addition to the opinions of endoscopists, will be an important process to improve the cure rate.
RESUMO
Colorectal cancer is the 3rd leading cause of cancer-related deaths in Korea, ranking 4th and 3rd among men and women, respectively. It is also the most common cause of cancer-related deaths in women older than 64 years. This study assessed the National Cancer Screening Program for colorectal cancer and examined its efficacy in enhancing public health. The fecal occult blood test (FOBT), a traditional noninvasive colorectal cancer screening test that can be performed on an outpatient basis was replaced with the fecal immunochemical test (FIT) because of the latter's better predictive value. Since 2004, the Government of South Korea has recommended an annual FIT for people aged 50 years and older as the first step in colorectal cancer screening. Individuals who test positive on the FIT are scheduled for follow-up screening procedures, such as colonoscopy or double-contrast barium enema, whereas those who have a negative FOBT are not recommended for colonoscopy. Colonoscopy, as a screening tool in Korea, has definite merits because it is highly accessible to patients and is performed by qualified specialists. Although the domestic colorectal cancer screening rate is relatively stable, there is scope for improvement. Owing to the low cost of colonoscopy and the wealth of skilled endoscopy specialists, the number of intention-to-screen procedures for colonoscopy has increased. As Korea is rapidly becoming an ultra-elderly society, it is time to reconsider the revision of the classical screening program and recommend region-specific, cost-effective guidelines.
RESUMO
BACKGROUND: Adolescent idiopathic scoliosis (AIS), which is the most common type of scoliosis, is a progressive disease that occurs in children aged 10-16 years. Abnormal curvature in AIS provokes spinal asymmetry of the upper body alignment and might deteriorate postural balancing and control ability. OBJECTIVE: To evaluate the effect of exercise interventions on balance and postural stability in patients with adolescent idiopathic scoliosis. METHODS: Embase, Scopus, Pubmed (Medline) and Web of Science databases were searched using the terms idiopathic scoliosis, physiotherapy, and balance. The articles selected were published in English in peer-reviewed journals from 2012 to July 2022. RESULTS: Ten studies met the inclusion criteria. The PEDro scale values ranged from 2 to 6 (mean, 3.6), indicating a low level of scientific rigor. In the sample studies, spinal stabilization exercises were most often trialed (n= 3), followed by Schroth's exercise (n= 2), stretching and self-elongation exercise (n= 2), the exercise protocol of Blount and Moe, physiotherapeutic scoliosis-specific exercise, and proprioceptive neuromuscular facilitation exercise (all n= 1). CONCLUSIONS: Physical therapists will be able to apply hippotherapy, Schroth exercise, physiotherapy scoliosis-specific exercise, trunk stabilization, proprioceptive neuromuscular facilitation exercise, spinal stabilization exercise, core stabilization exercise, and body awareness therapy to manage balance impairments in patients with adolescent idiopathic scoliosis, and further studies are needed to provide stronger evidence.
Assuntos
Cifose , Exercícios de Alongamento Muscular , Escoliose , Criança , Humanos , Adolescente , Escoliose/terapia , Coluna Vertebral , Terapia por Exercício/métodos , Equilíbrio Postural , Modalidades de FisioterapiaRESUMO
BACKGROUND: Non-Coronavirus disease 2019 (COVID-19) pneumonia and COVID-19 have similar clinical features but last for different periods, and consequently, require different treatment protocols. Therefore, they must be differentially diagnosed. This study uses artificial intelligence (AI) to classify the two forms of pneumonia using mainly laboratory test data. METHODS: Various AI models are applied, including boosting models known for deftly solving classification problems. In addition, important features that affect the classification prediction performance are identified using the feature importance technique and SHapley Additive exPlanations method. Despite the data imbalance, the developed model exhibits robust performance. RESULTS: eXtreme gradient boosting, category boosting, and light gradient boosted machine yield an area under the receiver operating characteristic of 0.99 or more, accuracy of 0.96-0.97, and F1-score of 0.96-0.97. In addition, D-dimer, eosinophil, glucose, aspartate aminotransferase, and basophil, which are rather nonspecific laboratory test results, are demonstrated to be important features in differentiating the two disease groups. CONCLUSIONS: The boosting model, which excels in producing classification models using categorical data, excels in developing classification models using linear numerical data, such as laboratory tests. Finally, the proposed model can be applied in various fields to solve classification problems.
Assuntos
COVID-19 , Pneumonia , Humanos , COVID-19/diagnóstico , Diagnóstico Diferencial , Inteligência Artificial , Pneumonia/diagnóstico , Aprendizado de Máquina , Teste para COVID-19RESUMO
BACKGROUND: Despite reports that the closed intensive care unit (ICU) system improves clinical outcomes, it has not been widely applied for various reasons. This study aimed to propose a better ICU system for critically ill patients by comparing the experience of open surgical ICU (OSICU) and closed surgical ICU (CSICU) systems in the same institution. METHODS AND FINDINGS: Our institution converted the ICU system from "open" to "closed" in February 2020, and enrolled patients were classified into the OSICU and CSICU groups at that time from March 2019 to February 2022. A total of 751 patients were categorized into the OSICU (n = 191) and CSICU (n = 560) groups. The mean age of the patients was 67 years in the OSICU group and 72 years in the CSICU group (p < 0.05). The acute physiology and chronic health evaluation II score was 21.8 ± 7.65 in the CSICU group, which was higher than the score 17.4 ± 7.97 in the OSICU group (p < 0.05). The sequential organ failure assessment scores were 2.0 ± 2.29 in the OSICU group and 4.1 ± 3.06 in the CSICU group (p < 0.05). After correction for bias by logistic regression analysis for all-cause mortality, the odds ratio in the CSICU group was 0.089 (95% confidence interval [CI]: 0.014-0.568, p < 0.05). CONCLUSIONS: Despite considering the various factors of increased patient severity, a CSICU system is more beneficial for critically ill patients. Therefore, we propose that the CSICU system be applied worldwide.
Assuntos
Estado Terminal , Unidades de Terapia Intensiva , Humanos , Idoso , Fatores de Risco , Escores de Disfunção Orgânica , Estudos RetrospectivosRESUMO
BACKGROUND: Thyroid storm is a life-threatening emergency. Reportedly, the prevalence of thyroid storm is 1%-2% among patients admitted to the hospital for thyrotoxicosis. Burch and Wartofsky (1993) introduced a scoring system using precise clinical criteria to identify thyroid storms. Only 17 cases of thyroid storm with a score > 70 points have been reported. Although thyroid storms are uncommon, their clinical findings resemble those of sepsis. CASE SUMMARY: A 48-year-old man was referred to the emergency room from a local clinic owing to suspicion of gastric ulcer perforation; medications for hypertension, diabetes mellitus, and hyperthyroidism had been suspended 1 year prior to this visit. We performed an emergency distal gastrectomy with Billroth II anastomosis for gastric cardia cancer perforation, and the patient was referred to the surgical intensive care unit (ICU). On the 2nd d in the ICU, his body temperature (BT) increased to 41.3 °C at 19:00, with the thyroid storm score (90 points) peaking at 18:00 (BT; 41.2ºC, pulse rate; 138/min, irritable status). The patient was administered propylthiouracil, intravenous glucocorticoids, acetaminophen, and Lugol's solution daily. Subsequently, we performed bladder irrigation with cold saline using a Foley catheter and applied a hypothermic blanket to decrease the patient's BT. His vital signs were stable on the 8th day in the ICU. CONCLUSION: Thyroid storms are uncommon, with few reports in the literature; however, their clinical findings resemble those of sepsis and require further investigation. Since an untreated thyroid storm results in a high mortality rate, it should be investigated when managing sepsis.
RESUMO
This study was designed to develop machine-learning models to predict COVID-19 mortality and identify its key features based on clinical characteristics and laboratory tests. For this, deep-learning (DL) and machine-learning (ML) models were developed using receiver operating characteristic (ROC) area under the curve (AUC) and F1 score optimization of 87 parameters. Of the two, the DL model exhibited better performance (AUC 0.8721, accuracy 0.84, and F1 score 0.76). However, we also blended DL with ML, and the ensemble model performed the best (AUC 0.8811, accuracy 0.85, and F1 score 0.77). The DL model is generally unable to extract feature importance; however, we succeeded by using the Shapley Additive exPlanations method for each model. This study demonstrated both the applicability of DL and ML models for classifying COVID-19 mortality using hospital-structured data and that the ensemble model had the best predictive ability.
RESUMO
BACKGROUND: The criteria for brain death determination have not been unified globally, and there is no global consensus on the apnea test, which is essential for determining brain death. Since the apnea test is associated with many complications, we aimed to determine an optimal duration of the apnea test. METHODS: We analyzed the results of the apnea test performed for brain death determination between August 2013 and February 2021 at a single institution in South Korea. Elevations in the partial pressure of carbon dioxide and mean arterial blood pressure fluctuations over time in the apnea test were recorded. RESULTS: In the 1st and 2nd tests, the mean partial pressure of carbon dioxide increased by more than 20 mmHg at 3 min after the apnea test compared to before the test (P < 0.05). At 4 min in the 1st test and 5 min in the 2nd test, the partial pressure of carbon dioxide exceeded 60 mmHg (P < 0.05). The fluctuation in the mean arterial blood pressure observed for 5 min during the apnea test was not significant. There was no significant fluctuation in the mean arterial blood pressure over time in the apnea test between patients with normal chest radiography findings and those with abnormal chest radiography findings (P = 0.888). CONCLUSIONS: Our study proposes that a short-term apnea test protocol is valid for the preservation of organs for donation.
Assuntos
Apneia , Morte Encefálica , Apneia/diagnóstico , Pressão Arterial , Morte Encefálica/diagnóstico , Dióxido de Carbono , Humanos , Pressão ParcialRESUMO
BACKGROUND: Sepsis and septic shock remain the leading causes of death in critically ill patients worldwide. Various biomarkers are available to determine the prognosis and therapeutic effects of sepsis. In this study, we investigated the effectiveness of presepsin as a sepsis biomarker. METHODS: Patients admitted to the intensive care unit with major or minor diagnosis of sepsis were categorized into survival and non-survival groups. The white blood cell count and serum C-reactive protein, procalcitonin, and presepsin levels were measured in all patients. RESULTS: The study included 40 patients (survival group, 32; non-survival group, 8; mortality rate, 20%). The maximum serum presepsin levels measured during intensive care unit admission were significantly higher in the non-survival group (median [interquartile range]: 4,205.5 pg/ml [1,155.8-10,094.0] vs. 741.5 pg/ml [520.0-1,317.5], P<0.05). No statistically significant intergroup differences were observed in the maximum, minimum, and mean values of the white blood cell count, as well as serum C-reactive protein, and procalcitonin levels. Based on the receiver operating characteristic curve, the area under the curve for presepsin as a predictor of sepsis mortality was 0.764. At a cut-off value of 1,898.5 pg/ml, the sensitivity and specificity of presepsin for prediction of sepsis-induced mortality were 75.0% and 87.5%, respectively. CONCLUSIONS: Early diagnosis of sepsis and prediction of sepsis-induced mortality are important for prompt initiation of treatment. Presepsin may serve as an effective biomarker for prediction of sepsis-induced mortality and for evaluation of treatment effectiveness.
RESUMO
BACKGROUND: The effectiveness of side-sling plank (SSP) exercises on trunk and hip muscle activation in subjects with gluteus medius (Gmed) weakness is unclear. OBJECTIVE: To quantify muscle activation of the rectus abdominis (RA), external oblique (EO), erector spinae (ES), lumbar multifidus (LM), Gmed, gluteus maximus (Gmax), and tensor fasciae latae (TFL) during SSP with three different hip rotations compared to side-lying hip abduction (SHA) exercise in subjects with Gmed weakness. METHODS: Twenty-two subjects with Gmed weakness were recruited. SHA and three types of SSP exercises were performed: SSP with neutral hip (SSP-N), hip lateral rotation (SSP-L), and hip medial rotation (SSP-M). Surface electromyography was used to measure the activation of the trunk and hip muscles. RESULTS: The trunk and hip muscles activations were generally significantly higher level during three SSP than SHA. SSP-M showed significantly lower EO activation while significantly higher ES and LM activation than SSP-L. Gmed activation was significantly higher during SSP-M than during SSP-L. TFL activation was significantly lower during SSP-M than during SSP-N and SSP-L. CONCLUSIONS: SSP could be prescribed for patients who have reduced Gmed strength after injuries. Especially, SSP-M could be applied for patients who have Gmed weakness with dominant TFL.