A5046405075- Retinal Imaging and Analysis
- Retinal and Optic Conditions
- Retinal Diseases and Treatments
- Artificial Intelligence in Healthcare and Education
- Artificial Intelligence in Healthcare
- Glaucoma and retinal disorders
- Intraocular Surgery and Lenses
- Ocular Diseases and Behçet’s Syndrome
- COVID-19 diagnosis using AI
- Ocular Infections and Treatments
- Facial Trauma and Fracture Management
- Traumatic Ocular and Foreign Body Injuries
- Systemic Lupus Erythematosus Research
- Ocular Disorders and Treatments
- Ethics in Clinical Research
- Acute Ischemic Stroke Management
- Ophthalmology and Visual Health Research
- COVID-19 and healthcare impacts
- Blood Pressure and Hypertension Studies
- Health Systems, Economic Evaluations, Quality of Life
- Injury Epidemiology and Prevention
- Maternal and Neonatal Healthcare
- Ophthalmology and Eye Disorders
- Telemedicine and Telehealth Implementation
- Mobile Health and mHealth Applications
Universidade Federal de São Paulo
2018-2025
Massachusetts Institute of Technology
2022-2025
Advanced Neural Dynamics (United States)
2024
University of Chile
2024
In-Q-Tel
2021
Instituto Dante Pazzanese de Cardiologia
2021
Ophthalmology Associates (United States)
2021
Associação Paulista de Medicina
2019-2020
Fundação de Apoio à Universidade Federal de São Paulo
2020
Universidade de São Paulo
2019
Abstract Objective Large-scale multi-modal deep learning models and datasets have revolutionized various domains such as healthcare, underscoring the critical role of computational power. However, in resource-constrained regions like Low Middle-Income Countries (LMICs), GPU data access is limited, leaving many dependent solely on CPUs. To address this, we advocate leveraging vector embeddings for flexible efficient methodologies, aiming to democratize multimodal across diverse contexts....
The aim of this study was to evaluate the performance ChatGPT-4.0 in answering 2022 Brazilian National Examination for Medical Degree Revalidation (Revalida) and as a tool provide feedback on quality examination.A total two independent physicians entered all examination questions into ChatGPT-4.0. After comparing outputs with test solutions, they classified large language model answers adequate, inadequate, or indeterminate. In cases disagreement, adjudicated achieved consensus decision...
<title>Abstract</title> In the big data era, integrating diverse modalities poses significant challenges, particularly in complex fields like healthcare. This paper introduces a new process model for multimodal Data Fusion Mining, embeddings and Cross-Industry Standard Process Mining with existing Information Group model. Our aims to decrease computational costs, complexity, bias while improving efficiency reliability. We also propose "disentangled dense fusion," novel embedding fusion...
Abstract Introduction The Brazilian Multilabel Ophthalmological Dataset (BRSET) addresses the scarcity of publicly available ophthalmological datasets in Latin America. BRSET comprises 16,266 color fundus retinal photos from 8,524 patients, aiming to enhance data representativeness, serving as a research and teaching tool. It contains sociodemographic information, enabling investigations into differential model performance across demographic groups. Methods Data three São Paulo outpatient...
Abstract Background Postoperative endophthalmitis (PSE) is a severe ocular complication that can lead to irreversible vision loss or even globe atrophy. The Endophthalmitis Vitrectomy Study (EVS) historically guided PSE management but increasingly questioned due advances in pars plana vitrectomy (PPV) techniques and its narrow focus on cataract surgery. This study aimed compare PPV followed by intravitreal antibiotic injection at the end of surgery (PPV + IVAIES) with alone (IVAI) managing...
Artificial Intelligence (AI) represents a significant milestone in health care's digital transformation. However, traditional care education and training often lack competencies. To promote safe effective AI implementation, professionals must acquire basic knowledge of machine learning neural networks, critical evaluation data sets, integration within clinical workflows, bias control, human-machine interaction settings. Additionally, they should understand the legal ethical aspects impact...
Abstract Aims This study aims to compare the performance of a handheld fundus camera (Eyer) and standard tabletop cameras (Visucam 500, Visucam 540, Canon CR-2) for diabetic retinopathy macular edema screening. Methods was multicenter, cross-sectional that included images from 327 individuals with diabetes. The participants underwent pharmacological mydriasis photography in two fields (macula optic disk centered) both strategies. All were acquired by trained healthcare professionals,...
PurposeTo evaluate the performance of artificial intelligence (AI) systems embedded in a mobile, handheld retinal camera, with single image protocol, detecting both diabetic retinopathy (DR) and more-than-mild (mtmDR).DesignMulticenter cross-sectional diagnostic study, conducted at three diabetes care eye facilities.ParticipantsA total 327 individuals mellitus (Type 1 or Type 2) underwent imaging protocol enabling expert reading automated analysis.MethodsParticipants fundus photographs using...
Abstract To assess the feasibility of code-free deep learning (CFDL) platforms in prediction binary outcomes from fundus images ophthalmology, evaluating two distinct online-based (Google Vertex and Amazon Rekognition), datasets. Two publicly available datasets, Messidor-2 BRSET, were utilized for model development. The consists photographs diabetic patients BRSET is a multi-label dataset. CFDL used to create models, with no preprocessing images, by single ophthalmologist without coding...
Abstract Purpose Diabetic retinopathy (DR) screening in low- and middle-income countries (LMICs) faces challenges due to limited access specialized care. Portable retinal cameras provide a practical alternative, but image quality, influenced by mydriasis, affects artificial intelligence (AI) model performance. This study examines the role of mydriasis improving quality AI-based DR detection resource-limited settings. Methods We compared proportion gradable images between mydriatic...
Abstract Diabetic retinopathy (DR) is a serious diabetes complication that can lead to vision loss, making timely identification crucial. Existing data-driven algorithms for DR staging from digital fundus images (DFIs) often struggle with generalization due distribution shifts between training and target domains. To address this, DRStageNet, deep learning model, was developed using six public independent datasets 91,984 DFIs diverse demographics. Five pretrained self-supervised transformers...
Objectives Health research that significantly impacts global clinical practice and policy is often published in high-impact factor (IF) medical journals. These outlets play a pivotal role the worldwide dissemination of novel knowledge. However, researchers identifying as women those affiliated with institutions low- middle-income countries (LMICs) have been largely under-represented high-IF journals across multiple fields medicine. To evaluate disparities gender geographical representation...
This paper introduces the Team Card (TC) as a protocol to address harmful biases in development of clinical artificial intelligence (AI) systems by emphasizing often-overlooked role researchers' positionality. While bias medical AI, particularly Clinical Decision Support (CDS) tools, is frequently attributed issues data quality, this limited framing neglects how worldviews-shaped their training, backgrounds, and experiences-can influence AI design deployment. These unexamined subjectivities...