Machine Learning-Based Approach for Identifying Research Gaps: COVID-19 as a Case Study

BACKGROUND

Research gaps refer to unanswered questions in the existing body of knowledge, either due to a lack of studies or inconclusive results. Research gaps are essential starting points and motivation in scientific research. Traditional methods for identifying research gaps, such as literature reviews and expert opinions, can be time consuming, labor intensive, and prone to bias. They may also fall short when dealing with rapidly evolving or time-sensitive subjects. Thus, innovative scalable approaches are needed to identify research gaps, systematically assess the literature, and prioritize areas for further study in the topic of interest.

OBJECTIVE

In this paper, we propose a machine learning-based approach for identifying research gaps through the analysis of scientific literature. We used the COVID-19 pandemic as a case study.

METHODS

We conducted an analysis to identify research gaps in COVID-19 literature using the COVID-19 Open Research (CORD-19) data set, which comprises 1,121,433 papers related to the COVID-19 pandemic. Our approach is based on the BERTopic topic modeling technique, which leverages transformers and class-based term frequency-inverse document frequency to create dense clusters allowing for easily interpretable topics. Our BERTopic-based approach involves 3 stages: embedding documents, clustering documents (dimension reduction and clustering), and representing topics (generating candidates and maximizing candidate relevance).

RESULTS

After applying the study selection criteria, we included 33,206 abstracts in the analysis of this study. The final list of research gaps identified 21 different areas, which were grouped into 6 principal topics. These topics were: "virus of COVID-19," "risk factors of COVID-19," "prevention of COVID-19," "treatment of COVID-19," "health care delivery during COVID-19," "and impact of COVID-19." The most prominent topic, observed in over half of the analyzed studies, was "the impact of COVID-19."

CONCLUSIONS

The proposed machine learning-based approach has the potential to identify research gaps in scientific literature. This study is not intended to replace individual literature research within a selected topic. Instead, it can serve as a guide to formulate precise literature search queries in specific areas associated with research questions that previous publications have earmarked for future exploration. Future research should leverage an up-to-date list of studies that are retrieved from the most common databases in the target area. When feasible, full texts or, at minimum, discussion sections should be analyzed rather than limiting their analysis to abstracts. Furthermore, future studies could evaluate more efficient modeling algorithms, especially those combining topic modeling with statistical uncertainty quantification, such as conformal prediction.

Novel Nebulized Milrinone Formulation for the Treatment of Acute Heart Failure Requiring Inotropic Therapy: A Phase 1 Study

BACKGROUND

Nonintravenous inotropic-delivery options are needed for patients with inotropic-dependent heart failure (HF) to reduce the costs, infections and thrombotic risks associated with chronic central venous catheters and home infusion services.

METHODS

We developed a novel, concentrated formulation of nebulized milrinone for inhalation and evaluated the feasibility, safety and pharmacokinetic profile in a prospective, single-arm, phase I clinical trial. We enrolled 10 patients with stage D HF requiring inotropic therapy during a hospital admission for acute HF. Milrinone 60 mg/4 mL was inhaled via nebulization 3 times daily for 48 hours. The coprimary outcomes were adverse events and pharmacokinetic profiles of inhaled milrinone. Acute changes in hemodynamic parameters were secondary outcomes.

RESULTS

A concentrated nebulized milrinone formulation was well tolerated, without hypotensive events, arrhythmias or inhalation-related adverse events requiring discontinuation. Nebulized milrinone produced serum concentrations in the goal therapeutic range with a median plasma milrinone trough concentration of 39 (17-66) ng/mL and a median peak concentration of 207 (134-293) ng/mL. There were no serious adverse events. From baseline to 24 hours, mean pulmonary artery saturation increased (60% ± 7%-65 ± 5%; P = 0.001), and mean cardiac index increased (2.0 ± 0.5 mL/min/1.73m2-2.5 ± 0.1 mL/min/1.73m2; P = 0.001) with nebulized milrinone.

CONCLUSIONS

In a proof-of-concept study, a concentrated, nebulized milrinone formulation for inhalation was safe and produced therapeutic serum milrinone concentrations. Nebulized milrinone was associated with improved hemodynamic parameters of cardiac output in a population with advanced HF. These promising results require further investigation in a longer-term trial in patients with inotrope-dependent advanced HF.

Heart transplantation from COVID-positive donors with 6-month follow-up: A case series

BACKGROUND

Data on long term outcomes in heart transplant recipients from Coronavirus disease 2019 (COVID-19) positive donors are limited.

METHODS AND RESULTS

We present a series of nine patients who underwent heart transplants from COVID-19 PCR-positive donors between November 2021 to August 2022 with mean follow-up of 12.12 ± 3 months. All the recipients received two doses of COVID-19 vaccine and had at least 6 months follow-up. Eight recipients had acceptable long-term outcomes; one patient died during index admission from primary graft dysfunction. Details regarding donor and recipient characteristics, management and outcomes are provided. Two patients developed deep vein thrombosis, and one patient underwent pacemaker implantation for sinus node dysfunction. Among the surviving eight patients, none developed COVID-19 infection during follow-up period. There was no significant difference in outcome parameters when compared to patients who received hearts from donors who tested negative for COVID-19 during the same time period at our center.

CONCLUSION

Keeping in mind the significant waitlist mortality in patients awaiting heart transplantation, COVID-19-positive donors should be considered for heart transplantation to help expand the donor pool and potentially reduce waitlist mortality.

Extracting factors associated with vaccination from Twitter data and mapping to behavioral models

Social media platform, particularly Twitter, is a rich data source that allows monitoring of public opinions and attitudes toward vaccines.Established behavioral models like the 5C psychological antecedents model and the Health Belief Model (HBM) provide a well-structured framework for analyzing shifts in vaccine-related behavior. This study examines if the extracted data from Twitter contains valuable insights regarding public attitudes toward vaccines and can be mapped to two behavioral models. This study focuses on the Arab population, and a search was carried out on Twitter using: ' تلقيحي OR تطعيم OR تطعيمات OR لقاح OR لقاحات' for two years from January 2020 to January 2022. Then, BERTopicmodeling was applied, and several topics were extracted. Finally, the topics were manually mapped to the factors of the 5C model and HBM. 1,068,466 unique users posted 3,368,258 vaccine-related tweets in Arabic. Topic modeling generated 25 topics, which were mapped to the 15 factors of the 5C model and HBM. Among the users, 32.87%were male, and 18.06% were female. A significant 55.77% of the users were from the MENA (Middle East and North Africa) region. Twitter users were more inclined to accept vaccines when they trusted vaccine safety and effectiveness, but vaccine hesitancy increased due to conspiracy theories and misinformation. The association of topics with these theoretical frameworks reveals the availability and diversity of Twitter data that can predict behavioral change toward vaccines. It allows the preparation of timely and effective interventions for vaccination programs compared to traditional methods.

Longitudinal analysis of behavioral factors and techniques used to identify vaccine hesitancy among Twitter users: Scoping review

While vaccines have played a pivotal role in the fight against infectious diseases, individuals engage in online resources to find vaccine-related support and information. The benefits and consequences of these online peers are unclear and mainly cause a behavioral shift in user sentiment toward vaccination. This scoping review aims to identify the community and individual factors that longitudinally influence public behavior toward vaccination. The secondary aim is to gain insight into techniques and methodologies used to extract these factors from Twitter data. We followed PRISMA-ScR guidelines to search various online repositories. From this search process, a total of 28 most relevant articles out of 705 relevant studies. Three main themes emerged including individual and community factors influencing public attitude toward vaccination, and techniques employed to identify these factors. Anti-vax, Pro-vax, and neutral are the major communities, while misinformation, vaccine campaign, and user demographics are the common individual factors assessed during this reviewing process. Twitter user sentiment (positive, negative, and neutral) and emotions (fear, trust, sadness) were also discussed to identify the intentions to accept or refuse vaccines. SVM, LDA, BERT are the techniques used for topic modeling, while Louvain, NodeXL, and Infomap algorithms are used for community detection. This research is notable for being the first systematic review that emphasizes the dearth of longitudinal studies and the methodological and underlying practical constraints underpinning the lucrative implementation of an explainable and longitudinal behavior analysis system. Moreover, new possible research directions are suggested for the researchers to perform accurate human behavior analysis.

Artificial Intelligence-Based Methods for Integrating Local and Global Features for Brain Cancer Imaging: Scoping Review

BACKGROUND

Transformer-based models are gaining popularity in medical imaging and cancer imaging applications. Many recent studies have demonstrated the use of transformer-based models for brain cancer imaging applications such as diagnosis and tumor segmentation.

OBJECTIVE

This study aims to review how different vision transformers (ViTs) contributed to advancing brain cancer diagnosis and tumor segmentation using brain image data. This study examines the different architectures developed for enhancing the task of brain tumor segmentation. Furthermore, it explores how the ViT-based models augmented the performance of convolutional neural networks for brain cancer imaging.

METHODS

This review performed the study search and study selection following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. The search comprised 4 popular scientific databases: PubMed, Scopus, IEEE Xplore, and Google Scholar. The search terms were formulated to cover the interventions (ie, ViTs) and the target application (ie, brain cancer imaging). The title and abstract for study selection were performed by 2 reviewers independently and validated by a third reviewer. Data extraction was performed by 2 reviewers and validated by a third reviewer. Finally, the data were synthesized using a narrative approach.

RESULTS

Of the 736 retrieved studies, 22 (3%) were included in this review. These studies were published in 2021 and 2022. The most commonly addressed task in these studies was tumor segmentation using ViTs. No study reported early detection of brain cancer. Among the different ViT architectures, Shifted Window transformer-based architectures have recently become the most popular choice of the research community. Among the included architectures, UNet transformer and TransUNet had the highest number of parameters and thus needed a cluster of as many as 8 graphics processing units for model training. The brain tumor segmentation challenge data set was the most popular data set used in the included studies. ViT was used in different combinations with convolutional neural networks to capture both the global and local context of the input brain imaging data.

CONCLUSIONS

It can be argued that the computational complexity of transformer architectures is a bottleneck in advancing the field and enabling clinical transformations. This review provides the current state of knowledge on the topic, and the findings of this review will be helpful for researchers in the field of medical artificial intelligence and its applications in brain cancer.

Transcatheter aortic valve replacement for aortic insufficiency in a patient with aortic root Thrombus and left ventricular assist device: A risk worth taking?

A 61-year-old man with end-stage ischemic cardiomyopathy post HeartMate 3 (Abbott laboratories, Chicago, Illinois, USA) left ventricular assist device (LVAD) implant was hospitalized after he had recurrent ventricular tachycardia requiring implantable cardioverter-defibrillator shocks. His transthoracic echocardiogram and computed tomography angiography of the chest showed presence of trace aortic insufficiency (AI) and aortic root thrombus (ART) of non-coronary cusp without obstruction of right or left coronary artery ostium despite therapeutic international normalized ratio. He presented again 3 months later with worsening heart failure signs and symptoms. Transesophageal echocardiogram showed progression to severe AI and persistent ART. Despite hemodynamically guided LVAD speed optimization, inotropic support, and diuresis, the patient continued to deteriorate with worsening renal function. The patient was not a transplant candidate due to frailty. After multi-disciplinary discussion he underwent successful 29-Sapien S3 (Edwards Lifesciences, Irvine, CA, USA) transcatheter aortic valve replacement utilizing distal protection filters in bilateral internal carotid arteries for stroke prevention. This case provides novel insight to physicians treating LVAD patients regarding management of severe AI in the setting of ART.

Learning objective

We report a rare approach employed for management of aortic insufficiency (AI) in a patient who also had an aortic root thrombus and left ventricular assist device (LVAD) that traditionally requires cardiac transplantation. Our patient had a favorable outcome with a minimally invasive transcatheter aortic valve replacement. With this case, we hope to generate awareness amongst physicians treating patients about management alternatives and approach of a commonly encountered, life-threatening complication of AI in patients with LVAD.

Multimodal deep learning for liver cancer applications: a scoping review

Background

Hepatocellular carcinoma is a malignant neoplasm of the liver and a leading cause of cancer-related deaths worldwide. The multimodal data combines several modalities, such as medical images, clinical parameters, and electronic health record (EHR) reports, from diverse sources to accomplish the diagnosis of liver cancer. The introduction of deep learning models with multimodal data can enhance the diagnosis and improve physicians' decision-making for cancer patients.

Objective

This scoping review explores the use of multimodal deep learning techniques (i.e., combining medical images and EHR data) in diagnosing and prognosis of hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA).

Methodology

A comprehensive literature search was conducted in six databases along with forward and backward references list checking of the included studies. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension for scoping review guidelines were followed for the study selection process. The data was extracted and synthesized from the included studies through thematic analysis.

Results

Ten studies were included in this review. These studies utilized multimodal deep learning to predict and diagnose hepatocellular carcinoma (HCC), but no studies examined cholangiocarcinoma (CCA). Four imaging modalities (CT, MRI, WSI, and DSA) and 51 unique EHR records (clinical parameters and biomarkers) were used in these studies. The most frequently used medical imaging modalities were CT scans followed by MRI, whereas the most common EHR parameters used were age, gender, alpha-fetoprotein AFP, albumin, coagulation factors, and bilirubin. Ten unique deep-learning techniques were applied to both EHR modalities and imaging modalities for two main purposes, prediction and diagnosis.

Conclusion

The use of multimodal data and deep learning techniques can help in the diagnosis and prediction of HCC. However, there is a limited number of works and available datasets for liver cancer, thus limiting the overall advancements of AI for liver cancer applications. Hence, more research should be undertaken to explore further the potential of multimodal deep learning in liver cancer applications.

A scoping review of artificial intelligence-based methods for diabetes risk prediction

The increasing prevalence of type 2 diabetes mellitus (T2DM) and its associated health complications highlight the need to develop predictive models for early diagnosis and intervention. While many artificial intelligence (AI) models for T2DM risk prediction have emerged, a comprehensive review of their advancements and challenges is currently lacking. This scoping review maps out the existing literature on AI-based models for T2DM prediction, adhering to the PRISMA extension for Scoping Reviews guidelines. A systematic search of longitudinal studies was conducted across four databases, including PubMed, Scopus, IEEE-Xplore, and Google Scholar. Forty studies that met our inclusion criteria were reviewed. Classical machine learning (ML) models dominated these studies, with electronic health records (EHR) being the predominant data modality, followed by multi-omics, while medical imaging was the least utilized. Most studies employed unimodal AI models, with only ten adopting multimodal approaches. Both unimodal and multimodal models showed promising results, with the latter being superior. Almost all studies performed internal validation, but only five conducted external validation. Most studies utilized the area under the curve (AUC) for discrimination measures. Notably, only five studies provided insights into the calibration of their models. Half of the studies used interpretability methods to identify key risk predictors revealed by their models. Although a minority highlighted novel risk predictors, the majority reported commonly known ones. Our review provides valuable insights into the current state and limitations of AI-based models for T2DM prediction and highlights the challenges associated with their development and clinical integration.

Wnt10b protects cardiomyocytes against doxorubicin-induced cell death via MAPK modulation

BACKGROUND

Doxorubicin, an anthracycline chemotherapeutic known to incur heart damage, decreases heart function in up to 11% of patients. Recent investigations have implicated the Wnt signaling cascade as a key modulator of cardiac tissue repair after myocardial infarction. Wnt upregulation in murine models resulted in stimulation of angiogenesis and suppression of fibrosis after ischemic insult. However, the molecular mechanisms of Wnt in mitigating doxorubicin-induced cardiac insult require further investigation. Identifying cardioprotective mechanisms of Wnt is imperative to reducing debilitating cardiovascular adverse events in oncologic patients undergoing treatment.

METHODS

Exposing human cardiomyocyte AC16 cells to varying concentrations of Wnt10b and DOX, we observed key metrics of cell viability. To assess the viability and apoptotic rates, we utilized MTT and TUNEL assays. We quantified cell and mitochondrial membrane stability via LDH release and JC-1 staining. To investigate how Wnt10b mitigates doxorubicin-induced apoptosis, we introduced pharmacologic inhibitors of key enzymes involved in apoptosis: FR180204 and SB203580, ERK1/2 and p38 inhibitors. Further, we quantified apoptotic executor enzymes, caspase 3/7, via immunofluorescence.

RESULTS

AC16 cells exposed solely to doxorubicin were shrunken with distorted morphology. Cardioprotective effects of Wnt10b were demonstrated via a reduction in apoptosis, from 70.1% to 50.1%. LDH release was also reduced between doxorubicin and combination groups from 2.27-fold to 1.56-fold relative to the healthy AC16 control group. Mitochondrial membrane stability was increased from 0.67-fold in the doxorubicin group to 5.73 in co-treated groups relative to control. Apoptotic protein expression was stifled by Wnt10b, with caspase3/7 expression reduced from 2.4- to 1.3-fold, and both a 20% decrease in p38 and 40% increase in ERK1/2 activity.

CONCLUSION

Our data with the AC16 cell model demonstrates that Wnt10b provides defense mechanisms against doxorubicin-induced cardiotoxicity and apoptosis. Further, we explain a mechanism of this beneficial effect involving the mitochondria through simultaneous suppression of pro-apoptotic p38 and anti-apoptotic ERK1/2 activities.

Orthostatic variation of pulmonary artery pressure in ambulatory heart failure patients

AIM

To study effect of change in position (supine and standing) on pulmonary artery pressure (PAP) in ambulatory heart failure (HF) patients.

METHODS

Seventeen patients with CardioMEMS® sensor and stable heart failure were consented and included in this single center study. Supine and standing measurements were obtained with at least 5 min interval between the two positions. These measurements included PAP readings utilizing the manufacturer handheld interrogator obtaining 10 s data in addition to the systemic blood pressure and heart rate recordings.

RESULTS

Mean supine and standing readings and their difference (Δ) were as follows respectively: Systolic PAP were 33.4 (± 11.19), 23.6 (± 10) and Δ was 9.9 mmHg (p = 0.0001), diastolic PAP were 14.2 (± 5.6), 7.9 (± 5.7) and Δ was 6.3 mmHg (p = 0.0001) and mean PAP were 21.8 (± 7.8), 14 (± 7.2) and Δ was 7.4 mmHg (p = 0.0001) while the systemic blood pressure did not vary significantly.

CONCLUSION

There is orthostatic variation of PAP in ambulatory HF patients demonstrating a mean decline with standing in diastolic PAP by 6.3 mmHg, systolic PAP by 9.9 mmHg and mean PAP by 7.4 mmHg in absence of significant orthostatic variation in systemic blood pressure or heart rate. These findings have significant clinical implications and inform that PAP in each patient should always be measured in the same position. Since initial readings at the time of implant were taken in supine position, it may be best to use supine position or to obtain a baseline standing PAP reading if standing PAP is planned on being used.

Improving diagnosis and prognosis of lung cancer using vision transformers: a scoping review

BACKGROUND

Vision transformer-based methods are advancing the field of medical artificial intelligence and cancer imaging, including lung cancer applications. Recently, many researchers have developed vision transformer-based AI methods for lung cancer diagnosis and prognosis.

OBJECTIVE

This scoping review aims to identify the recent developments on vision transformer-based AI methods for lung cancer imaging applications. It provides key insights into how vision transformers complemented the performance of AI and deep learning methods for lung cancer. Furthermore, the review also identifies the datasets that contributed to advancing the field.

METHODS

In this review, we searched Pubmed, Scopus, IEEEXplore, and Google Scholar online databases. The search terms included intervention terms (vision transformers) and the task (i.e., lung cancer, adenocarcinoma, etc.). Two reviewers independently screened the title and abstract to select relevant studies and performed the data extraction. A third reviewer was consulted to validate the inclusion and exclusion. Finally, the narrative approach was used to synthesize the data.

RESULTS

Of the 314 retrieved studies, this review included 34 studies published from 2020 to 2022. The most commonly addressed task in these studies was the classification of lung cancer types, such as lung squamous cell carcinoma versus lung adenocarcinoma, and identifying benign versus malignant pulmonary nodules. Other applications included survival prediction of lung cancer patients and segmentation of lungs. The studies lacked clear strategies for clinical transformation. SWIN transformer was a popular choice of the researchers; however, many other architectures were also reported where vision transformer was combined with convolutional neural networks or UNet model. Researchers have used the publicly available lung cancer datasets of the lung imaging database consortium and the cancer genome atlas. One study used a cluster of 48 GPUs, while other studies used one, two, or four GPUs.

CONCLUSION

It can be concluded that vision transformer-based models are increasingly in popularity for developing AI methods for lung cancer applications. However, their computational complexity and clinical relevance are important factors to be considered for future research work. This review provides valuable insights for researchers in the field of AI and healthcare to advance the state-of-the-art in lung cancer diagnosis and prognosis. We provide an interactive dashboard on lung-cancer.onrender.com/ .

Cytomegalovirus infection in heart transplant patient presenting as appendicitis

Cytomegalovirus (CMV) may manifest in various ways. While immunocompetent hosts may be asymptomatic or present with a mononucleosis-like illness, immunocompromised patients can have organ-specific disease capable of significant morbidity and mortality. CMV appendicitis is a particularly rare presentation. A 22-year-old female with a history of orthotopic heart transplantation presented to our hospital with a three-day history of worsening abdominal pain. A computed tomography scan of her abdomen was consistent with acute uncomplicated appendicitis, and she underwent laparoscopic appendectomy. Pathology revealed acute appendicitis with numerous large cells with intranuclear "owl's eye" inclusions characteristic of CMV. Her CMV viral load was elevated at 327,018 IU/ml. She was started on ganciclovir which resulted in improvement of her CMV level to 30,118 IU/ml within three weeks. CMV is a frequent cause of opportunistic infection in solid organ transplant patients and commonly involves the gastrointestinal tract. Acute appendicitis is a rarely reported complication to consider in the differential diagnosis of abdominal pain in immunocompromised patients.

Learning objective

Heart transplant recipients are at increased risk for opportunistic infections. Cytomegalovirus (CMV) is a frequent culprit and can present with a broad range of disease. A particularly rare presentation is that of acute appendicitis. We describe a case of a young woman with CMV appendicitis following orthotopic heart transplant.

Artificial Intelligence-Based Methods for Precision Cardiovascular Medicine

Precision medicine has the potential to revolutionize the way cardiovascular diseases are diagnosed, predicted, and treated by tailoring treatment strategies to the individual characteristics of each patient. Artificial intelligence (AI) has recently emerged as a promising tool for improving the accuracy and efficiency of precision cardiovascular medicine. In this scoping review, we aimed to identify and summarize the current state of the literature on the use of AI in precision cardiovascular medicine. A comprehensive search of electronic databases, including Scopes, Google Scholar, and PubMed, was conducted to identify relevant studies. After applying inclusion and exclusion criteria, a total of 28 studies were included in the review. We found that AI is being increasingly applied in various areas of cardiovascular medicine, including the diagnosis, prognosis of cardiovascular diseases, risk prediction and stratification, and treatment planning. As a result, most of these studies focused on prediction (50%), followed by diagnosis (21%), phenotyping (14%), and risk stratification (14%). A variety of machine learning models were utilized in these studies, with logistic regression being the most used (36%), followed by random forest (32%), support vector machine (25%), and deep learning models such as neural networks (18%). Other models, such as hierarchical clustering (11%), Cox regression (11%), and natural language processing (4%), were also utilized. The data sources used in these studies included electronic health records (79%), imaging data (43%), and omics data (4%). We found that AI is being increasingly applied in various areas of cardiovascular medicine, including the diagnosis, prognosis of cardiovascular diseases, risk prediction and stratification, and treatment planning. The results of the review showed that AI has the potential to improve the performance of cardiovascular disease diagnosis and prognosis, as well as to identify individuals at high risk of developing cardiovascular diseases. However, further research is needed to fully evaluate the clinical utility and effectiveness of AI-based approaches in precision cardiovascular medicine. Overall, our review provided a comprehensive overview of the current state of knowledge in the field of AI-based methods for precision cardiovascular medicine and offered new insights for researchers interested in this research area.

Trends and Outcomes of Cardiogenic Shock in Patients With End-Stage Renal Disease: Insights From USRDS Database

BACKGROUND

There is a paucity of data regarding epidemiology, temporal trends, and outcomes of patients with cardiogenic shock (CS) and end-stage renal disease (chronic kidney disease stage V on hemodialysis).

METHODS

This is a retrospective cohort study using the United States Renal Data System database from January 1, 2006 to December 31, 2019. We analyzed trends of CS, percutaneous mechanical support (intraaortic balloon pump, percutaneous ventricular assist device [Impella and Tandemheart], and extracorporeal membrane oxygenation) utilization, index mortality, 30-day mortality, and 1-year all-cause mortality in end-stage renal disease patients.

RESULTS

A total of 43 825 end-stage renal disease patients were hospitalized with CS (median age, 67.8 years [IQR, 59.4-75.8] and 59.1% men). From 2006 to 2019, the incidence of CS increased from 275 to 578 per 100 000 patients (Ptrend<0.001). The index mortality rate declined from 54.1% in 2006 to 40.8% in 2019 (Ptrend=0.44), and the 1-year all-cause mortality decreased from 63% in 2006 to 61.8% in 2018 (Ptrend=0.73), but neither trend was statistically significant. There was a significantly decreased utilization of intra-aortic balloon pumps from 17 832 to 7992 (Ptrend<0.001), increased utilization of percutaneous ventricular assist device from 137 to 5201 (Ptrend<0.001) and increase in extracorporeal membrane oxygenation use from 69 to 904 per 100 000 patients (Ptrend<0.001). After adjusting for covariates, there was no significant difference in index mortality between CS patients requiring percutaneous mechanical support versus those not requiring percutaneous mechanical support (odds ratio, 0.97 [CI, 0.91-1.02]; P=0.22). On multivariable regression analysis, older age, peripheral vascular disease, diabetes, and time on dialysis were independent predictors of higher index mortality.

CONCLUSIONS

The incidence of CS in end-stage renal disease patients has doubled without significant change in the trend of index mortality despite the use of percutaneous mechanical support.

Identifying the role of vision transformer for skin cancer-A scoping review

Introduction

Detecting and accurately diagnosing early melanocytic lesions is challenging due to extensive intra- and inter-observer variabilities. Dermoscopy images are widely used to identify and study skin cancer, but the blurred boundaries between lesions and besieging tissues can lead to incorrect identification. Artificial Intelligence (AI) models, including vision transformers, have been proposed as a solution, but variations in symptoms and underlying effects hinder their performance.

Objective

This scoping review synthesizes and analyzes the literature that uses vision transformers for skin lesion detection.

Methods

The review follows the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Revise) guidelines. The review searched online repositories such as IEEE Xplore, Scopus, Google Scholar, and PubMed to retrieve relevant articles. After screening and pre-processing, 28 studies that fulfilled the inclusion criteria were included.

Results and discussions

The review found that the use of vision transformers for skin cancer detection has rapidly increased from 2020 to 2022 and has shown outstanding performance for skin cancer detection using dermoscopy images. Along with highlighting intrinsic visual ambiguities, irregular skin lesion shapes, and many other unwanted challenges, the review also discusses the key problems that obfuscate the trustworthiness of vision transformers in skin cancer diagnosis. This review provides new insights for practitioners and researchers to understand the current state of knowledge in this specialized research domain and outlines the best segmentation techniques to identify accurate lesion boundaries and perform melanoma diagnosis. These findings will ultimately assist practitioners and researchers in making more authentic decisions promptly.

Predicting Long-Term Type 2 Diabetes with Artificial Intelligence (AI): A Scoping Review

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder that affects a significant portion of the global population. Artificial intelligence (AI) has emerged as a promising tool for predicting T2DM risk. To provide an overview of the AI techniques used for long-term prediction of T2DM and evaluate their performance, we conducted a scoping review using PRISMA-ScR. Of the 40 papers included in this review, 23 studies used Machine Learning (ML) as the most common AI technique, with Deep Learning (DL) models used exclusively in four studies. Of the 13 studies that used both ML and DL, 8 studies employed ensemble learning models, and SVM and RF were the most used individual classifiers. Our findings highlight the importance of accuracy and recall as validation metrics, with accuracy being used in 31 studies, followed by recall in 29 studies. These discoveries emphasize the critical role of high predictive accuracy and sensitivity in detecting positive T2DM cases.

Understanding Correlations of Loneliness in India and USA

Loneliness is a global public health issues contributing to a variety of mental and physical health issues. It also increases the risk of life-threatening conditions as well as contributes to burden on the economy in terms of the number of days lost to productivity. Loneliness is a highly varied concept though, which is a result of multiple factors. To understand loneliness this paper carries out a comparative analysis of USA and India through Twitter data on the keywords associated with loneliness. The comparative analysis on loneliness is in the vein of comparative public health literature and to contribute to develop a global public health map on loneliness. The results showed that the dynamics of loneliness through the topics correlated vary across geographical locations. Social media data can be used to capture the dynamics of loneliness which can vary from one place to another depending on the socioeconomic and cultural norms and sociopolitical policies.

Applications of Artificial Intelligence (AI) in Medical Education: A Scoping Review

Artificial Intelligence (AI) is increasingly used to support medical students' learning journeys, providing personalized experiences and improved outcomes. We conducted a scoping review to explore the current application and classifications of AI in medical education. Following the PRISMA-P guidelines, we searched four databases, ultimately including 22 studies. Our analysis identified four AI methods used in various medical education domains, with the majority of applications found in training labs. The use of AI in medical education has the potential to improve patient outcomes by equipping healthcare professionals with better skills and knowledge. Post-implementation refers to the outcomes of AI-based training, which showed improved practical skills among medical students. This scoping review highlights the need for further research to explore the effectiveness of AI applications in different aspects of medical education.

Revolutionizing Healthcare with Foundation AI Models

ChatGPT is a foundation Artificial Intelligence (AI) model that has opened up new opportunities in digital healthcare. Particularly, it can serve as a co-pilot tool for doctors in the interpretation, summarization, and completion of reports. Furthermore, it can build upon the ability to access the large literature and knowledge on the internet. So, chatGPT could generate acceptable responses for the medical examination. Hence. It offers the possibility of enhancing healthcare accessibility, expandability, and effectiveness. Nonetheless, chatGPT is vulnerable to inaccuracies, false information, and bias. This paper briefly describes the potential of Foundation AI models to transform future healthcare by presenting ChatGPT as an example tool.

Predicting Overall Survival in METABRIC Cohort Using Machine Learning

Triple-negative breast cancer (TNBC) is an aggressive form of breast cancer that presents very high relapse and mortality. However, due to differences in the genetic architecture associated with TNBC, patients have different outcomes and respond differently to available treatments. In this study, we predicted the overall survival of TNBC patients in the METABRIC cohort employing supervised machine learning to identify important clinical and genetic features that are associated with better survival. We achieved a slightly higher Concordance index than the state of art and identified biological pathways related to the top genes considered important by our model.

Using AI for Detection, Prediction and Classification of Retinal Detachment

The current state of machine learning (ML) and deep learning (DL) algorithms used to detect, classify and predict the onset of retinal detachment (RD) were examined in this scoping review. This severe eye condition can cause vision loss if left untreated. By analyzing the medical imaging modalities such as fundus photography, AI could help to detect peripheral detachment at an earlier stage. We have searched five databases: PubMed, Google Scholar, ScienceDirect, Scopus, and IEEE. Two reviewers independently carried out the selection of the studies and their data extractions. 32 studies fulfilled our eligibility criteria from the 666 references collected. In particular, based on the performance metrics employed in these studies, this scoping review provides a general overview of emerging trends and practices concerning using ML and DL algorithms for detecting, classifying, and predicting RD.

An Overview of Metabolomics Studies Based on Qatari Population

The aim of metabolomics research is to identify the metabolites that play a role in various biological traits and diseases. This scoping review provides an overview of the current state of metabolomics studies that focus on the Qatari population. Our findings indicate that few studies have been conducted on this population, with a focus on diabetes, dyslipidemia, and cardiovascular disease. Blood samples were the primary source of metabolite identification, and several potential biomarkers for these diseases were proposed. To the best of our knowledge, this is the first scoping review that presents an overview of metabolomics studies in Qatar.

The Pros and Cons of Using ChatGPT in Medical Education: A Scoping Review

This scoping review explores the advantages and disadvantages of using ChatGPT in medical education. We searched PubMed, Google Scholar, Medline, Scopus, and Science Direct to identify relevant studies. Two reviewers independently conducted study selection and data extraction, followed by a narrative synthesis. Out of 197 references, 25 studies met the eligibility criteria. The primary applications of ChatGPT in medical education include automated scoring, teaching assistance, personalized learning, research assistance, quick access to information, generating case scenarios and exam questions, content creation for learning facilitation, and language translation. We also discuss the challenges and limitations of using ChatGPT in medical education, such as its inability to reason beyond existing knowledge, generation of incorrect information, bias, potential undermining of students' critical thinking skills, and ethical concerns. These concerns include using ChatGPT for exam and assignment cheating by students and researchers, as well as issues related to patients' privacy.

Deep Learning in Colorectal Cancer Classification: A Scoping Review

Colorectal cancer (CRC) is one of the most common cancers worldwide, and its diagnosis and classification remain challenging for pathologists and imaging specialists. The use of artificial intelligence (AI) technology, specifically deep learning, has emerged as a potential solution to improve the accuracy and speed of classification while maintaining the quality of care. In this scoping review, we aimed to explore the utilization of deep learning for the classification of different types of colorectal cancer. We searched five databases and selected 45 studies that met our inclusion criteria. Our results show that deep learning models have been used to classify colorectal cancer using various types of data, with histopathology and endoscopy images being the most common. The majority of studies used CNN as their classification model. Our findings provide an overview of the current state of research on deep learning in the classification of colorectal cancer.

Impact of Biopsy Proven Liver Fibrosis on Patients Undergoing Evaluation and Treatment for Advanced Heart Failure Surgical Therapies

There is a paucity of data regarding the impact of liver fibrosis on patients with stage D heart failure (HF). We conducted a retrospective study (January 1, 2017 to December 12, 2020) in patients with stage D HF who underwent liver biopsy as part of their advanced HF therapy evaluation. Baseline characteristics and 1-year outcomes were compared between no- or mild-to-moderate-fibrosis (grade 0 to 2) and advanced-fibrosis (grade 3 to 4) groups. Of 519 patients with stage D HF, 136 who underwent liver biopsy (113 [83%] no or mild-to-moderate fibrosis and 23 [17%] advanced fibrosis) were included. A total of 71 patients (52%) received advanced HF therapies (23 heart transplantation, 48 left ventricular assist devices). One-year mortality was higher among patients with advanced fibrosis (52% vs 18%, p <0.001). Further subgroup analysis suggested a trend toward increased 1-year mortality among patients with advanced fibrosis who underwent advanced therapies (37% vs 13%, p = 0.09). There was a trend of lower likelihood of receiving advanced HF therapies in the advanced-fibrosis group, only 1 heart transplantation and 7 left ventricular assist devices, but it did not reach statistical significance (35% vs 56%, p = 0.06). After adjustment for confounders, degree of liver fibrosis was an independent predictor of mortality (odds ratio 6.2; 95% 1.27 to 30.29, p = 0.02). We conclude that advanced liver fibrosis is common among patients with stage D HF who undergo evaluation for advanced HF surgical therapies and significantly increases 1-year mortality. Further larger studies are needed to support our findings.

Early and aggressive presentation of wild-type transthyretin amyloid cardiomyopathy: A case report

BACKGROUND

Wild-type transthyretin amyloidosis (ATTRwt) is the most common form of transthyretin amyloid cardiomyopathy, occurring mostly over age of 60 years (mean age of 80 years). Mean survival without treatment is 3.6 years, making early detection imperative. We report an unusual case of a 58-year-old patient with ATTRwt cardiomyopathy requiring heart transplantation.

CASE SUMMARY

A 58-year-old male presented with progressive fatigue, shortness of breath, weight gain, leg swelling, orthopnoea, and paroxysmal nocturnal dyspnoea for several months. Approximately ten months before this clinical presentation, the patient had first received a diagnosis of heart failure with reduced ejection fraction (EF) of 15% to 20%. The patient was started on appropriate guideline-directed medical therapy with only mild improvement in his EF. Upon further investigation, echocardiogram, technetium pyrophosphate scan (Tc PYP), and cardiac magnetic resonance imaging (cMRI) suggested a diagnosis of amyloidosis, and ATTRwt was subsequently confirmed with native heart tissue biopsy, congo red staining, liquid chromatography-tandem mass spectrometry, and genetic testing. The patient was successfully treated with heart transplantation and is doing well post-transplant.

CONCLUSION

Wild-type ATTR amyloidosis should be kept on differentials in all patients (even less than 60 years old) with non-ischemic cardiomyopathy, especially in the setting of increased ventricular wall thickness and other classic echocardiogram, cMRI, and Tc PYP findings. Early diagnosis and management can be consequential in improving patient outcomes.

Palliative Inotropes in Advanced Heart Failure: Comparing Outcomes Between Milrinone and Dobutamine

BACKGROUND

We sought to describe and compare outcomes among advanced patients with heart failure (not candidates for orthotopic heart transplant/left ventricular assist device) on long-term milrinone or dobutamine, which are not well-studied in the contemporary era.

METHODS AND RESULTS

We included adults with refractory stage D heart failure who were not candidates for orthotopic heart transplant or left ventricular assist device and discharged on palliative dobutamine or milrinone. The primary outcome was 1-year survival. A 6-month predictor of survival analysis was conducted. A total of 248 patients (133 on milrinone, 115 on dobutamine) were included. There were no differences in baseline comorbidities between milrinone and dobutamine cohorts, except for the prevalence of chronic kidney disease, which was higher in the dobutamine group. On discharge, the proportion of patients on beta-blockers and mineralocorticoid antagonists was higher in milrinone group. Overall, the 1-year mortality rate was 70%. The dobutamine cohort had a significantly higher 1-year mortality rate (84% vs 58%, P <0.001). The type of inotrope did not predict survival at 6 months when adjusted for discharge medications and comorbidities. Beta-blockers and angiotensin-converting enzyme/angiotensin receptor blocker/angiotensin receptor neprilysin inhibitor continued at discharge predicted survival at 6 months.

CONCLUSIONS

The 1-year mortality from palliative inotropes remains high. Compared with dobutamine, use of milrinone was associated with improved survival owing to better optimization of guideline-directed medical therapy, primarily beta-blocker therapy.

Public attitudes on social media toward vaccination before and during the COVID-19 pandemic

With the success of COVID-19 vaccines in clinical trials, vaccination programs are being administered for the population with the hopes of herd immunity. However, the success of any vaccination program depends on the percentage of people willing to get vaccination which is influenced by social, economic, demographic, and vaccine-specific factors. Thus, it is important to understand public attitudes and perceptions toward vaccination. This study aims to measure public attitude toward vaccines and vaccinations before and during the COVID-19 pandemic, using public data from Twitter. A total of 880,586 tweets for 57,529 unique users were included in the study. Most of the tweets were posted in five languages: French, English, Swedish, Dutch, and Italian. These tweets were divided into two time periods: before COVID-19 (T1) and during COVID-19 (T2). This study observed the shift in the sentiments of the public attitude toward vaccines before and during COVID-19 pandemic. Both positive and negative shifts in sentiments were observed for the users of various languages but shifts toward positive sentiments were more prominent during the COVID-19 pandemic.

In-hospital Outcomes of Patients with Septic Shock and Underlying Chronic Atrial Fibrillation: A Propensity Matched Analysis from A National Dataset

INTRODUCTION

Atrial fibrillation (AF) is one of the most common arrhythmias among hospitalized patients. Among patients admitted with septic shock (SS), the new occurrence of atrial fibrillation has been associated with an increase in intensive care unit (ICU) length of stay and in-hospital mortality. This is partially related to further reduction in cardiac output and thus worsening organ perfusion due to atrial fibrillation. However, there is a paucity of research on the outcomes of patients who have underlying chronic AF (UCAF) and then develop SS. This study aimed to identify the clinical characteristics and outcomes of patients with UCAF admitted with SS compared to patients with SS without UCAF.

METHODS

This study was a retrospective analysis of the 2016 and 2017 Nationwide Readmission Database. ICD-10 codes were used to identify patients with SS, and these patients were stratified into those with and without UCAF. Propensity matching analyses were performed to compare clinical outcomes and in-hospital mortality between the two groups.

RESULTS

A total of 353,422 patients with hospitalization for SS were identified, 5.8% (n = 20,772) of whom had UCAF. After 2:1 propensity matching, 20,719 patients were identified as having SS with UCAF, and 41,438 patients were identified as having SS without UCAF. Patients with SS and UCAF had a higher incidence of ischemic stroke [2.5% versus 2.2%, p = 0.012], length of stay [11.5 days versus 10.9 days, p < 0.001], mean total charges [$154,094 versus $144,037, p < 0.001] compared to those with SS without UCAF. In-hospital mortality was high in both groups, but was slightly higher among those with SS and UCAF than those with SS and no UCAF [34.4% versus 34.1%, p = 0.049].

CONCLUSIONS

This study identified UCAF as an adverse prognosticator for clinical outcomes. Patients with SS and UCAF need to be identified as a higher risk category of SS who will require more intensive management.

Rare case of chronic Q fever myocarditis in end stage heart failure patient: A case report

BACKGROUND

Q fever myocarditis is a rare disease manifestation of Q fever infection caused by Coxiella burnetii. It is associated with significant morbidity and mortality if left untreated. Prior studies have reported myocarditis in patients with acute Q fever. We present the first case of chronic myocarditis in an end-stage heart failure patient with chronic Q fever infection.

CASE SUMMARY

A 69-year-old male was admitted with dyspnea on exertion, hypotension and bilateral lower extremity edema for a few months. He has a past medical history of ischemic cardiomyopathy with left ventricular ejection fraction of 25%, implantable cardioverter defibrillator in place, bioprosthetic aortic valve and mitral valve replacement. He continued to have shortness of breath despite diuresis along with low grade fevers. Initial infectious work up came back negative. On further questioning, the patient was found to have close contact with farm animals and the recurrent fevers prompted the work-up for Q fever. Q fever serologies and cardiac positron emission tomography confirmed the diagnosis of chronic Q fever myocarditis. He was then successfully treated with doxycycline and hydroxychloroquine for 18 mo.

CONCLUSION

Chronic Q fever myocarditis, if left untreated, carries a poor prognosis. It should be kept in differentials, especially in patients with recurrent fevers and contact with farm animals.

Diurnal variation of pulmonary artery pressure in ambulatory heart failure patients

BACKGROUND

There is limited information on diurnal variation in pulmonary artery pressures (PAP) in ambulatory heart failure (HF) patients. We aimed to study the variation in morning and night-time PAP in HF patients with an implanted CardioMEMS^® sensor.

METHODS

In this prospective, single centre study we enrolled patients who had a cardioMEMS sensor and consented to participate (End stage renal disease and recent hospitalisation for acute HF were exclusions). Subjects were asked to transmit PAP and non-invasive blood pressure information in morning and at night-time for 7 consecutive days. Categorical and continuous variables were reported as percentages and mean ± SD respectively. Repeated measure ANOVA was used to compare the diurnal changes in PAP among different subgroups. Pierson correlation coefficient was performed to assess correlation between diurnal variation of PAP and left ventricular ejection fraction.

RESULTS

Thirty subjects were included in analysis. There was a significant nocturnal rise in PASP and mPAP compared to morning readings (+2.59 mmHg, p = 0.003 and +1.24 mmHg with p = 0.02 respectively) while night-time PADP did not change significantly (+0.48 mmHg, p = 0.18) without significant change in systemic blood pressure or pulse rate.

CONCLUSION

The described diurnal changes in PAP should be considered when managing ambulatory HF patients based on these readings. PADP can be used reliably without concern for the time of day the readings were recorded.HighlightsThere is a diurnal variation in PAP in ambulatory heart failure patientsPulmonary artery systolic and mean pulmonary artery pressures are higher at night-time than in morning.Pulmonary artery diastolic pressures do not vary significantly with time of day.These findings should inform clinical decisions in management of these patients about the time of the day readings are taken.

Clinical Trials on Alternative Medicines for COVID-19

In this study, we addressed the alternative medications that have been targeted in the clinical trials (CTs) to be evidenced as an adjuvant treatment against COVID-19. Based on the outcomes from CTs, we found that dietary supplements such as Lactoferrin, and Probiotics (as SivoMixx) can play a role enhancing the immunity thus can be used as prophylactics against COVID-19 infection. Vitamin D was proven as an effective adjuvant treatment against COVID-19, while Vitamin C role is uncertain and needs more investigation. Herbals such as Guduchi Ghan Vati can be used as prophylactic, while Resveratrol can be used to reduce the hospitalization risk of COVID-19 patients. On the contrary, there were no clinical improvements demonstrated when using Cannabidiol. This study is a part of a two-phase research study. In the first phase, we gathered evidence-based information on alternative therapeutics for COVID-19 that are under CT. In the second phase, we plan to build a mobile health application that will provide evidence based alternative therapy information to health consumers.

Artificial Intelligence-Based Mobile Application for Sensing Children Emotion Through Drawings

Children go through varied emotions such as happiness, sadness, and fear. At times, it may be difficult for children to express their emotions. Detecting and understanding the unexpressed emotions of children is very important to address their needs and prevent mental health issues. In this paper, we develop an artificial intelligence (AI) based Emotion Sensing Recognition App (ESRA) to help parents and teachers understand the emotions of children by analyzing their drawings. We collected 102 drawings from a local school in Doha and 521 drawings from Google and Instagram. Four different experiments were conducted using a combination of the two datasets. The deep learning model was trained using the Fastai library in Python. The model classifies the drawings into positive or negative emotions. The model accuracy ranged from 55% to 79% in the four experiments. This study showed that ESRA has the potential in identifying the emotions of children. However, the underlying algorithm needs to be trained and evaluated using more drawings to improve its current accuracy and to be able to identify more specific emotions.

Effectiveness of Serious Games for Visuospatial Abilities in Elderly Population with Cognitive Impairment: A Systematic Review and Meta-Analysis

We explore the effectiveness of serious games for visuospatial abilities among older adults with cognitive impairment by conducting a systematic review. Out of 548 identified publications, seven randomized controlled trials (RCTs) were included in this review. According to a meta-analysis of four RCTs, there is no statistically significant difference (p=0.28) in visuospatial abilities between serious game and control groups. Further, the included RCTs noted no statistically significant difference in the visuospatial ability when comparing serious games to conventional exercise (one study) and other serious games (two studies). One RCT demonstrated a statistically significant effect of serious games on the visuospatial ability when compared with conventional cognitive training. This review could not prove the effectiveness of serious games in enhancing visuospatial abilities for older adults with cognitive impairment. Thus, serious games should not be offered or used for enhancing visuospatial abilities amongst the elderly population with cognitive impairment. More robust RCTs are needed to make firm conclusions on the efficacy of serious games.

Effectiveness of Serious Games for Language Processing Amongst Elderly Population with Cognitive Impairment: A Systematic Review and Meta-Analysis

This article intended to carry out a systematic review on the effectiveness of serious games for language processing among older adults with cognitive impairment. Out of 548 retrieved records, six randomized controlled trials (RCTs) eventually met the eligibility criteria. A meta-analysis of four studies showed that serious games are more effective than no/passive interventions in enhancing language processing among older adults with cognitive impairment (p=0.008). Further, a statistically significant effect of serious games on language processing when compared with conventional cognitive activities and conventional exercises was reported in two RCTs. Other RCTs found that exergames are as effective as computerized cognitive training games in improving language processing. Serious games should be offered or used as complementary (i.e., not a substitute) to the current interventions. For there to be definitive conclusions about the efficacy of serious games on language processing more trials are needed.

Artificial Intelligence-Based Models for Predicting Vaccines Critical Tweets: An Experimental Study

We studied the suitability of Artificial Intelligence (AI)-based models to predict vaccine-critical tweets on the social media platform Twitter. We manually labeled a sample of 800 tweets as either "vaccine-critical" (i.e, anti-vaccine tweets, mentioned concerns related to vaccine safety and efficacy, and are against vaccine mandates or vaccine passports) or "other" (i.e., tweets that are neutral, report news, or are ambiguous) and used them to train and test AI-based models for automatically predicting vaccine-critical tweets. We fine-tuned two pre-trained deep learning-based language models, BERT and BERTweet, and implemented four classical AI-based models, Random Forest, Logistics Regression, Linear Support Vector Machines, and Multinomial Naïve Bayes. We evaluated these AI-based models using f1 score, accuracy, precision, and recall in three-fold cross-validation. We found that BERTweet outperformed all other models using these measures.

Federated Learning and Internet of Medical Things - Opportunities and Challenges

The recent advancements in artificial intelligence (AI) and the Internet of Medical Things (IoMT) have opened new horizons for healthcare technology. AI models, however, rely on large data that must be shared with the centralized entity developing the model. Data sharing leads to privacy preservation and legal issues. Federated Learning (FL) enables the training of AI models on distributed data. Hence, a large amount of IoMT data can be put into use without the need for sharing the data. This paper presents the opportunities offered by FL for privacy preservation in IoMT data. With FL, the complicated dynamics and agreements for data-sharing can be avoided. Furthermore, it describes the use cases of FL in facilitating collaborative efforts to develop AI for COVID-19 diagnosis. Since handling data from multiple sites poses its challenges, the paper also highlights the critical challenges associated with FL developments for IoMT data. Addressing these challenges will lead to gaining maximum benefit from data-driven AI technologies in IoMT.

Combating COVID-19 Using Generative Adversarial Networks and Artificial Intelligence for Medical Images: Scoping Review

Background

Research on the diagnosis of COVID-19 using lung images is limited by the scarcity of imaging data. Generative adversarial networks (GANs) are popular for synthesis and data augmentation. GANs have been explored for data augmentation to enhance the performance of artificial intelligence (AI) methods for the diagnosis of COVID-19 within lung computed tomography (CT) and X-ray images. However, the role of GANs in overcoming data scarcity for COVID-19 is not well understood.

Objective

This review presents a comprehensive study on the role of GANs in addressing the challenges related to COVID-19 data scarcity and diagnosis. It is the first review that summarizes different GAN methods and lung imaging data sets for COVID-19. It attempts to answer the questions related to applications of GANs, popular GAN architectures, frequently used image modalities, and the availability of source code.

Methods

A search was conducted on 5 databases, namely PubMed, IEEEXplore, Association for Computing Machinery (ACM) Digital Library, Scopus, and Google Scholar. The search was conducted from October 11-13, 2021. The search was conducted using intervention keywords, such as “generative adversarial networks” and “GANs,” and application keywords, such as “COVID-19” and “coronavirus.” The review was performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) guidelines for systematic and scoping reviews. Only those studies were included that reported GAN-based methods for analyzing chest X-ray images, chest CT images, and chest ultrasound images. Any studies that used deep learning methods but did not use GANs were excluded. No restrictions were imposed on the country of publication, study design, or outcomes. Only those studies that were in English and were published from 2020 to 2022 were included. No studies before 2020 were included.

Results

This review included 57 full-text studies that reported the use of GANs for different applications in COVID-19 lung imaging data. Most of the studies (n=42, 74%) used GANs for data augmentation to enhance the performance of AI techniques for COVID-19 diagnosis. Other popular applications of GANs were segmentation of lungs and superresolution of lung images. The cycleGAN and the conditional GAN were the most commonly used architectures, used in 9 studies each. In addition, 29 (51%) studies used chest X-ray images, while 21 (37%) studies used CT images for the training of GANs. For the majority of the studies (n=47, 82%), the experiments were conducted and results were reported using publicly available data. A secondary evaluation of the results by radiologists/clinicians was reported by only 2 (4%) studies.

Conclusions

Studies have shown that GANs have great potential to address the data scarcity challenge for lung images in COVID-19. Data synthesized with GANs have been helpful to improve the training of the convolutional neural network (CNN) models trained for the diagnosis of COVID-19. In addition, GANs have also contributed to enhancing the CNNs’ performance through the superresolution of the images and segmentation. This review also identified key limitations of the potential transformation of GAN-based methods in clinical applications.