Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices

The use of artificial intelligence to assess diabetic eye disease among the Greenlandic population

Background: Retina fundus images conducted in Greenland are telemedically assessed for diabetic retinopathy by ophthalmological nurses in Denmark. Applying an AI grading solution, in a Greenlandic setting, could potentially improve the efficiency and cost-effectiveness of DR screening.Method: We developed an AI model using retina fundus photos, performed on persons registered with diabetes in Greenland and Denmark, using Optos® ultra wide-field scanning laser ophthalmoscope, graded according to ICDR.Using the ResNet50 network we compared the model's ability to distinguish between different images of ICDR severity levels in a confusion matrix.Results: Comparing images with ICDR level 0 to images of ICDR level 4 resulted in an accuracy of 0.9655, AUC of 0.9905, sensitivity and specificity of 96.6%.Comparing ICDR levels 0,1,2 with ICDR levels 3,4, we achieved a performance with an accuracy of 0.8077, an AUC of 0.8728, a sensitivity of 84.6% and a specificity of 78.8%. For the other comparisons, we achieved a modest performance.Conclusion: We developed an AI model using Greenlandic data, to automatically detect DR on Optos retina fundus images. The sensitivity and specificity were too low for our model to be applied directly in a clinical setting, thus optimising the model should be prioritised.

Implementation of deep learning artificial intelligence in vision-threatening disease screenings for an underserved community during COVID-19

INTRODUCTION

Age-related macular degeneration, diabetic retinopathy, and glaucoma are vision-threatening diseases that are leading causes of vision loss. Many studies have validated deep learning artificial intelligence for image-based diagnosis of vision-threatening diseases. Our study prospectively investigated deep learning artificial intelligence applications in student-run non-mydriatic screenings for an underserved, primarily Hispanic community during COVID-19.

METHODS

Five supervised student-run community screenings were held in West New York, New Jersey. Participants underwent non-mydriatic 45-degree retinal imaging by medical students. Images were uploaded to a cloud-based deep learning artificial intelligence for vision-threatening disease referral. An on-site tele-ophthalmology grader and remote clinical ophthalmologist graded images, with adjudication by a senior ophthalmologist to establish the gold standard diagnosis, which was used to assess the performance of deep learning artificial intelligence.

RESULTS

A total of 385 eyes from 195 screening participants were included (mean age 52.43  ±  14.5 years, 40.0% female). A total of 48 participants were referred for at least one vision-threatening disease. Deep learning artificial intelligence marked 150/385 (38.9%) eyes as ungradable, compared to 10/385 (2.6%) ungradable as per the human gold standard (p < 0.001). Deep learning artificial intelligence had 63.2% sensitivity, 94.5% specificity, 32.0% positive predictive value, and 98.4% negative predictive value in vision-threatening disease referrals. Deep learning artificial intelligence successfully referred all 4 eyes with multiple vision-threatening diseases. Deep learning artificial intelligence graded images (35.6  ±  13.3 s) faster than the tele-ophthalmology grader (129  ±  41.0) and clinical ophthalmologist (68  ±  21.9, p < 0.001).

DISCUSSION

Deep learning artificial intelligence can increase the efficiency and accessibility of vision-threatening disease screenings, particularly in underserved communities. Deep learning artificial intelligence should be adaptable to different environments. Consideration should be given to how deep learning artificial intelligence can best be utilized in a real-world application, whether in computer-aided or autonomous diagnosis.

Multi-modal representation learning in retinal imaging using self-supervised learning for enhanced clinical predictions

Self-supervised learning has become the cornerstone of building generalizable and transferable artificial intelligence systems in medical imaging. In particular, contrastive representation learning techniques trained on large multi-modal datasets have demonstrated impressive capabilities of producing highly transferable representations for different downstream tasks. In ophthalmology, large multi-modal datasets are abundantly available and conveniently accessible as modern retinal imaging scanners acquire both 2D fundus images and 3D optical coherence tomography (OCT) scans to assess the eye. In this context, we introduce a novel multi-modal contrastive learning-based pipeline to facilitate learning joint representations for the two retinal imaging modalities. After self-supervised pre-training on 153,306 scan pairs, we show that such a pre-training framework can provide both a retrieval system and encoders that produce comprehensive OCT and fundus image representations that generalize well for various downstream tasks on three independent external datasets, explicitly focusing on clinically pertinent prediction tasks. In addition, we show that interchanging OCT with lower-cost fundus imaging can preserve the predictive power of the trained models.

Global review of publicly available image datasets for the anterior segment of the eye

This study comprehensively reviewed publicly available image datasets for the anterior segment, with a focus on cataract, refractive, and corneal surgeries. The goal was to assess characteristics of existing datasets and identify areas for improvement. PubMED and Google searches were performed using the search terms "refractive surgery," "anterior segment," "cornea," "corneal," "cataract" AND "database," with the related word of "imaging." Results of each of these searches were collated, identifying 26 publicly available anterior segment image datasets. Imaging modalities included optical coherence tomography, photography, and confocal microscopy. Most datasets were small, 80% originated in the U.S., China, or Europe. Over 50% of images were from normal eyes. Disease states represented included keratoconus, corneal ulcers, and Fuchs dystrophy. Most of the datasets were incompletely described. To promote accessibility going forward to 2030, the ESCRS Digital Health Special Interest Group will annually update a list of available image datasets for anterior segment at www.escrs.org .

Novel artificial intelligence for diabetic retinopathy and diabetic macular edema: what is new in 2024?

PURPOSE OF REVIEW

Given the increasing global burden of diabetic retinopathy and the rapid advancements in artificial intelligence, this review aims to summarize the current state of artificial intelligence technology in diabetic retinopathy detection and management, assessing its potential to improve care and visual outcomes in real-world settings.

RECENT FINDINGS

Most recent studies focused on the integration of artificial intelligence in the field of diabetic retinopathy screening, focusing on real-world efficacy and clinical implementation of such artificial intelligence models. Additionally, artificial intelligence holds the potential to predict diabetic retinopathy progression, enhance personalized treatment strategies, and identify systemic disease biomarkers from ocular images through 'oculomics', moving towards a more precise, efficient, and accessible care. The emergence of foundation model architectures and generative artificial intelligence, which more clearly reflect the clinical care process, may enable rapid advances in diabetic retinopathy care, research and medical education.

SUMMARY

This review explores the emerging technology of artificial intelligence to assess the potential to improve patient outcomes and optimize personalized management in healthcare delivery and medical research. While artificial intelligence is expected to play an increasingly important role in diabetic retinopathy care, ongoing research and clinical trials are essential to address implementation issues and focus on long-term patient outcomes for successful real-world adoption of artificial intelligence in diabetic retinopathy.

AI in the clinical management of GA: A novel therapeutic universe requires novel tools

Regulatory approval of the first two therapeutic substances for the management of geographic atrophy (GA) secondary to age-related macular degeneration (AMD) is a major breakthrough following failure of numerous previous trials. However, in the absence of therapeutic standards, diagnostic tools are a key challenge as functional parameters in GA are hard to provide. The majority of anatomical biomarkers are subclinical, necessitating advanced and sensitive image analyses. In contrast to fundus autofluorescence (FAF), optical coherence tomography (OCT) provides high-resolution visualization of neurosensory layers, including photoreceptors, and other features that are beyond the scope of human expert assessment. Artificial intelligence (AI)-based methodology strongly enhances identification and quantification of clinically relevant GA-related sub-phenotypes. Introduction of OCT-based biomarker analysis provides novel insight into the pathomechanisms of disease progression and therapeutic, moving beyond the limitations of conventional descriptive assessment. Accordingly, the Food and Drug Administration (FDA) has provided a paradigm-shift in recognizing ellipsoid zone (EZ) attenuation as a primary outcome measure in GA clinical trials. In this review, the transition from previous to future GA classification and management is described. With the advent of AI tools, diagnostic and therapeutic concepts have changed substantially in monitoring and screening of GA disease. Novel technology combined with pathophysiological knowledge and understanding of the therapeutic response to GA treatments, is currently opening the path for an automated, efficient and individualized patient care with great potential to improve access to timely treatment and reduce health disparities.

Multimodal imaging in diabetic retinopathy and macular edema: An update about biomarkers

Diabetic macular edema (DME), defined as retinal thickening near, or involving the fovea caused by fluid accumulation in the retina, can lead to vision impairment and blindness in patients with diabetes. Current knowledge of retina anatomy and function and DME pathophysiology has taken great advantage of the availability of several techniques for visualizing the retina. Combining these techniques in a multimodal imaging approach to DME is recommended to improve diagnosis and to guide treatment decisions. We review the recent literature about the following retinal imaging technologies: optical coherence tomography (OCT), OCT angiography (OCTA), wide-field and ultrawide-field techniques applied to fundus photography, fluorescein angiography, and OCTA. The emphasis will be on characteristic DME features identified by these imaging technologies and their potential or established role as diagnostic, prognostic, or predictive biomarkers. The role of artificial intelligence in the assessment and interpretation of retina images is also discussed.

Synchronous Diagnosis of Diabetic Retinopathy by a Handheld Retinal Camera, Artificial Intelligence, and Simultaneous Specialist Confirmation

PURPOSE

Diabetic retinopathy (DR) is a leading cause of preventable blindness, particularly in underserved regions where access to ophthalmic care is limited. This study presents a proof of concept for utilizing a portable handheld retinal camera with an embedded artificial intelligence (AI) platform, complemented by a synchronous remote confirmation by retina specialists, for DR screening in an underserved rural area.

DESIGN

Retrospective cohort study.

SUBJECTS

A total of 1115 individuals with diabetes.

METHODS

A retrospective analysis of a screening initiative conducted in 4 municipalities in Northeastern Brazil, targeting the diabetic population. A portable handheld retinal camera captured macula-centered and disc-centered images, which were analyzed by the AI system. Immediate push notifications were sent out to retina specialists upon the detection of significant abnormalities, enabling synchronous verification and confirmation, with on-site patient feedback within minutes. Referral criteria were established, and all referred patients underwent a complete ophthalmic work-up and subsequent treatment.

MAIN OUTCOME MEASURES

Proof-of-concept implementation success.

RESULTS

Out of 2052 invited individuals, 1115 participated, with a mean age of 60.93 years and diabetes duration of 7.52 years; 66.03% were women. The screening covered 2222 eyes, revealing various retinal conditions. Referable eyes for DR were 11.84%, with an additional 13% for other conditions (diagnoses included various stages of DR, media opacity, nevus, drusen, enlarged cup-to-disc ratio, pigmentary changes, and other). Artificial intelligence performance for overall detection of referable cases (both DR and other conditions) was as follows: sensitivity 84.23% (95% confidence interval (CI), 82.63-85.84), specificity 80.79% (95% CI, 79.05-82.53). When we assessed whether AI matched any clinical diagnosis, be it referable or not, sensitivity was 85.67% (95% CI, 84.12-87.22), specificity was 98.86 (95% CI, 98.39-99.33), and area under the curve was 0.92 (95% CI, 0.91-0.94).

CONCLUSIONS

The integration of a portable device, AI analysis, and synchronous medical validation has the potential to play a crucial role in preventing blindness from DR, especially in socially unequal scenarios.

FINANCIAL DISCLOSURE(S)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

The AI revolution in glaucoma: Bridging challenges with opportunities

Recent advancements in artificial intelligence (AI) herald transformative potentials for reshaping glaucoma clinical management, improving screening efficacy, sharpening diagnosis precision, and refining the detection of disease progression. However, incorporating AI into healthcare usages faces significant hurdles in terms of developing algorithms and putting them into practice. When creating algorithms, issues arise due to the intensive effort required to label data, inconsistent diagnostic standards, and a lack of thorough testing, which often limits the algorithms' widespread applicability. Additionally, the "black box" nature of AI algorithms may cause doctors to be wary or skeptical. When it comes to using these tools, challenges include dealing with lower-quality images in real situations and the systems' limited ability to work well with diverse ethnic groups and different diagnostic equipment. Looking ahead, new developments aim to protect data privacy through federated learning paradigms, improving algorithm generalizability by diversifying input data modalities, and augmenting datasets with synthetic imagery. The integration of smartphones appears promising for using AI algorithms in both clinical and non-clinical settings. Furthermore, bringing in large language models (LLMs) to act as interactive tool in medicine may signify a significant change in how healthcare will be delivered in the future. By navigating through these challenges and leveraging on these as opportunities, the field of glaucoma AI will not only have improved algorithmic accuracy and optimized data integration but also a paradigmatic shift towards enhanced clinical acceptance and a transformative improvement in glaucoma care.

Expanded Field OCT Angiography Biomarkers for Predicting Clinically Significant Outcomes in Non-Proliferative Diabetic Retinopathy

PURPOSE

To evaluate the utility of extended field swept-source Optical Coherence Tomography Angiography (SS-OCTA) imaging biomarkers in predicting the occurrence of clinically significant outcomes in eyes with Non-Proliferative Diabetic Retinopathy (NPDR).

DESIGN

Retrospective clinical case-control study.

METHODS

Single-center clinical study. 88 eyes with NPDR from 57 participants (median age: 64.0 years; mean duration of diabetes: 15.8 years) with at least two consecutive SS-OCTA scans over a follow-up period of at least six months were included. The presence of intraretinal microvascular abnormalities (IRMAs) at baseline and the stability of IRMAs during follow-up period on 12 × 12-mm angiograms were evaluated. Baseline nonperfusion ischemia index (ISI) and other SS-OCTA metrics were calculated on FIJI and ARI Network. Significant clinical outcomes were defined as occurrence of one or more of the following events at the last available clinical visit:1. significant DR progression (2-step DR progression or progression to proliferative DR (PDR)); 2) development of new center-involving diabetic macular edema (CI-DME); and 3) initiation of treatment with PRP or anti-VEGF injections during the follow-up period. Mixed-effects Cox regression models was used to explore these outcomes.

RESULTS

Following a clinical follow-up period lasting 25.1 ± 10.8 months, we observed significant clinical outcomes in 17 eyes (19.3%). Among these, 7 eyes (8.0%) experienced significant progression and 4 eyes (4.5%) developed CI-DME. Anti-VEGF injections were initiated in 15 eyes (17.0%), while PRP was initiated in 2 eyes (2.3%). Upon adjusting for age, the duration of DM, and prior Anti-VEGF treatments, our analysis revealed that non-stable IRMAs during the follow-up periods and a higher ischemia index at baseline were significantly associated with the occurrence of significant clinical outcomes with HRs of 3.88 (95% CI: 1.56-9.64; p=0.004) and 1.05 (95% CI: 1.02-1.09; p=0.004), respectively.

CONCLUSIONS

In conclusion, NPDR eyes with non-stable IRMAs over time and more ischemia at baseline are in higher risk of developing significant clinical outcomes. Our findings suggest that expanded field SS-OCTA may offer additional prognostic benefits for clinical DR staging and predicting high-risk patients.

Grading of diabetic retinopathy using a pre-segmenting deep learning classification model: Validation of an automated algorithm

PURPOSE

To validate the performance of autonomous diabetic retinopathy (DR) grading by comparing a human grader and a self-developed deep-learning (DL) algorithm with gold-standard evaluation.

METHODS

We included 500, 6-field retinal images graded by an expert ophthalmologist (gold standard) according to the International Clinical Diabetic Retinopathy Disease Severity Scale as represented with DR levels 0-4 (97, 100, 100, 103, 100, respectively). Weighted kappa was calculated to measure the DR classification agreement for (1) a certified human grader without, and (2) with assistance from a DL algorithm and (3) the DL operating autonomously. Using any DR (level 0 vs. 1-4) as a cutoff, we calculated sensitivity, specificity, as well as positive and negative predictive values (PPV and NPV). Finally, we assessed lesion discrepancies between Model 3 and the gold standard.

RESULTS

As compared to the gold standard, weighted kappa for Models 1-3 was 0.88, 0.89 and 0.72, sensitivities were 95%, 94% and 78% and specificities were 82%, 84% and 81%. Extrapolating to a real-world DR prevalence of 23.8%, the PPV were 63%, 64% and 57% and the NPV were 98%, 98% and 92%. Discrepancies between the gold standard and Model 3 were mainly incorrect detection of artefacts (n = 49), missed microaneurysms (n = 26) and inconsistencies between the segmentation and classification (n = 51).

CONCLUSION

While the autonomous DL algorithm for DR classification only performed on par with a human grader for some measures in a high-risk population, extrapolations to a real-world population demonstrated an excellent 92% NPV, which could make it clinically feasible to use autonomously to identify non-DR patients.

Trends and hotspots in the field of diabetic retinopathy imaging research from 2000-2023

Background

Diabetic retinopathy (DR) poses a major threat to diabetic patients' vision and is a critical public health issue. Imaging applications for DR have grown since the 21st century, aiding diagnosis, grading, and screening. This study uses bibliometric analysis to assess the field's advancements and key areas of interest.

Methods

This study performed a bibliometric analysis of DR imaging articles collected from the Web of Science Core Collection database between January 1st, 2000, and December 31st, 2023. The literature information was then analyzed through CiteSpace.

Results

The United States and China led in the number of publications, with 719 and 609, respectively. The University of London topped the institution list with 139 papers. Tien Yin Wong was the most prolific researcher. Invest. Ophthalmol. Vis. Sci. published the most articles (105). Notable burst keywords were "deep learning," "artificial intelligence," et al.

Conclusion

The United States is at the forefront of DR research, with the University of London as the top institution and Invest. Ophthalmol. Vis. Sci. as the most published journal. Tien Yin Wong is the most influential researcher. Hotspots like "deep learning," and "artificial intelligence," have seen a significant rise, indicating artificial intelligence's growing role in DR imaging.

Artificial Intelligence-Based Digital Biomarkers for Type 2 Diabetes: A Review

Type 2 diabetes mellitus (T2DM), a complex metabolic disorder that burdens the health care system, requires early detection and treatment. Recent strides in digital health technologies, coupled with artificial intelligence (AI), may have the potential to revolutionize T2DM screening, diagnosis of complications, and management through the development of digital biomarkers. This review provides an overview of the potential applications of AI-driven biomarkers in the context of screening, diagnosing complications, and managing patients with T2DM. The benefits of using multisensor devices to develop digital biomarkers are discussed. The summary of these findings and patterns between model architecture and sensor type are presented. In addition, we highlight the pivotal role of AI techniques in clinical intervention and implementation, encompassing clinical decision support systems, telemedicine interventions, and population health initiatives. Challenges such as data privacy, algorithm interpretability, and regulatory considerations are also highlighted, alongside future research directions to explore the use of AI-driven digital biomarkers in T2DM screening and management.

RiskPath: Explainable deep learning for multistep biomedical prediction in longitudinal data

Predicting individual and population risk for disease outcomes and identifying persons at elevated risk is a key prerequisite for targeting interventions to improve health. However, current risk stratification tools for the common, chronic diseases that develop over the lifecourse and represent the majority of disease morbidity, mortality and healthcare costs are aging and achieve only moderate predictive performance. In some common, highly morbid conditions such as mental illness no risk stratification tools are yet available. There is an urgent need to improve predictive performance for chronic diseases and understand how cumulative, multifactorial risks aggregate over time so that intervention programs can be targeted earlier and more effectively in the disease course. Chronic diseases are the end outcomes of multifactorial risks that increment over years and represent cumulative, temporally-sensitive risk pathways. However, tools in current clinical use were constructed in older data and utilize inputs from a single data collection step. Here, we present RiskPath, a multistep deep learning method for temporally-sensitive biomedical risk prediction tailored for the constraints and demands of biomedical practice that achieves very strong performance and full translational explainability. RiskPath delineates and quantifies cumulative multifactorial risk pathways and allows the user to explore performance-complexity tradeoffs and constrain models as required by clinical use cases. Our results highlight the potential for developing a new generation of risk stratification tools and risk pathway mapping in time-dependent diseases and health outcomes by leveraging powerful timeseries deep learning methods in the wealth of biomedical data now appearing in large, longitudinal open science datasets.

Foundation models in ophthalmology

Foundation models represent a paradigm shift in artificial intelligence (AI), evolving from narrow models designed for specific tasks to versatile, generalisable models adaptable to a myriad of diverse applications. Ophthalmology as a specialty has the potential to act as an exemplar for other medical specialties, offering a blueprint for integrating foundation models broadly into clinical practice. This review hopes to serve as a roadmap for eyecare professionals seeking to better understand foundation models, while equipping readers with the tools to explore the use of foundation models in their own research and practice. We begin by outlining the key concepts and technological advances which have enabled the development of these models, providing an overview of novel training approaches and modern AI architectures. Next, we summarise existing literature on the topic of foundation models in ophthalmology, encompassing progress in vision foundation models, large language models and large multimodal models. Finally, we outline major challenges relating to privacy, bias and clinical validation, and propose key steps forward to maximise the benefit of this powerful technology.

Diagnostic Performance of the Offline Medios Artificial Intelligence for Glaucoma Detection in a Rural Tele-Ophthalmology Setting

PURPOSE

This study assesses the diagnostic efficacy of offline Medios Artificial Intelligence (AI) glaucoma software in a primary eye care setting, using nonmydriatic fundus images from Remidio's Fundus-on-Phone (FOP NM-10). Artificial intelligence results were compared with tele-ophthalmologists' diagnoses and with a glaucoma specialist's assessment for those participants referred to a tertiary eye care hospital.

DESIGN

Prospective cross-sectional study PARTICIPANTS: Three hundred three participants from 6 satellite vision centers of a tertiary eye hospital.

METHODS

At the vision center, participants underwent comprehensive eye evaluations, including clinical history, visual acuity measurement, slit lamp examination, intraocular pressure measurement, and fundus photography using the FOP NM-10 camera. Medios AI-Glaucoma software analyzed 42-degree disc-centric fundus images, categorizing them as normal, glaucoma, or suspect. Tele-ophthalmologists who were glaucoma fellows with a minimum of 3 years of ophthalmology and 1 year of glaucoma fellowship training, masked to artificial intelligence (AI) results, remotely diagnosed subjects based on the history and disc appearance. All participants labeled as disc suspects or glaucoma by AI or tele-ophthalmologists underwent further comprehensive glaucoma evaluation at the base hospital, including clinical examination, Humphrey visual field analysis, and OCT. Artificial intelligence and tele-ophthalmologist diagnoses were then compared with a glaucoma specialist's diagnosis.

MAIN OUTCOME MEASURES

Sensitivity and specificity of Medios AI.

RESULTS

Out of 303 participants, 299 with at least one eye of sufficient image quality were included in the study. The remaining 4 participants did not have sufficient image quality in both eyes. Medios AI identified 39 participants (13%) with referable glaucoma. The AI exhibited a sensitivity of 0.91 (95% confidence interval [CI]: 0.71-0.99) and specificity of 0.93 (95% CI: 0.89-0.96) in detecting referable glaucoma (definite perimetric glaucoma) when compared to tele-ophthalmologist. The agreement between AI and the glaucoma specialist was 80.3%, surpassing the 55.3% agreement between the tele-ophthalmologist and the glaucoma specialist amongst those participants who were referred to the base hospital. Both AI and the tele-ophthalmologist relied on fundus photos for diagnoses, whereas the glaucoma specialist's assessments at the base hospital were aided by additional tools such as Humphrey visual field analysis and OCT. Furthermore, AI had fewer false positive referrals (2 out of 10) compared to the tele-ophthalmologist (9 out of 10).

CONCLUSIONS

Medios offline AI exhibited promising sensitivity and specificity in detecting referable glaucoma from remote vision centers in southern India when compared with teleophthalmologists. It also demonstrated better agreement with glaucoma specialist's diagnosis for referable glaucoma participants.

FINANCIAL DISCLOSURE(S)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Artificial Intelligence-Based Screening System for Diabetic Retinopathy in Primary Care

BACKGROUND

This study aimed to test an artificial intelligence-based reading system (AIRS) capable of reading retinographies of type 2 diabetic (T2DM) patients and a predictive algorithm (DRPA) that predicts the risk of each patient with T2DM of developing diabetic retinopathy (DR).

METHODS

We tested the ability of the AIRS to read and classify 15,297 retinal photographs from our database of diabetics and 1200 retinal images taken with Messidor-2 into the different DR categories. We tested the DRPA in a sample of 40,129 T2DM patients. The results obtained by the AIRS and the DRPA were then compared with those provided by four retina specialists regarding sensitivity (S), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), accuracy (ACC), and area under the curve (AUC).

RESULTS

The results of testing the AIRS for identifying referral DR (RDR) in our database were ACC = 98.6, S = 96.7, SP = 99.8, PPV = 99.0, NPV = 98.0, and AUC = 0.958, and in Messidor-2 were ACC = 96.78%, S = 94.64%, SP = 99.14%, PPV = 90.54%, NPV = 99.53%, and AUC = 0.918. The results of our DRPA when predicting the presence of any type of DR were ACC = 0.97, S = 0.89, SP = 0.98, PPV = 0.79, NPV = 0.98, and AUC = 0.92.

CONCLUSIONS

The AIRS performed well when reading and classifying the retinographies of T2DM patients with RDR. The DRPA performed well in predicting the absence of DR based on some clinical variables.

Ethical Considerations in the Design and Conduct of Clinical Trials of Artificial Intelligence

Importance

Safe integration of artificial intelligence (AI) into clinical settings often requires randomized clinical trials (RCT) to compare AI efficacy with conventional care. Diabetic retinopathy (DR) screening is at the forefront of clinical AI applications, marked by the first US Food and Drug Administration (FDA) De Novo authorization for an autonomous AI for such use.

Objective

To determine the generalizability of the 7 ethical research principles for clinical trials endorsed by the National Institute of Health (NIH), and identify ethical concerns unique to clinical trials of AI.

Design, Setting, and Participants

This qualitative study included semistructured interviews conducted with 11 investigators engaged in the design and implementation of clinical trials of AI for DR screening from November 11, 2022, to February 20, 2023. The study was a collaboration with the ACCESS (AI for Children's Diabetic Eye Exams) trial, the first clinical trial of autonomous AI in pediatrics. Participant recruitment initially utilized purposeful sampling, and later expanded with snowball sampling. Study methodology for analysis combined a deductive approach to explore investigators' perspectives of the 7 ethical principles for clinical research endorsed by the NIH and an inductive approach to uncover the broader ethical considerations implementing clinical trials of AI within care delivery.

Results

A total of 11 participants (mean [SD] age, 47.5 [12.0] years; 7 male [64%], 4 female [36%]; 3 Asian [27%], 8 White [73%]) were included, with diverse expertise in ethics, ophthalmology, translational medicine, biostatistics, and AI development. Key themes revealed several ethical challenges unique to clinical trials of AI. These themes included difficulties in measuring social value, establishing scientific validity, ensuring fair participant selection, evaluating risk-benefit ratios across various patient subgroups, and addressing the complexities inherent in the data use terms of informed consent.

Conclusions and Relevance

This qualitative study identified practical ethical challenges that investigators need to consider and negotiate when conducting AI clinical trials, exemplified by the DR screening use-case. These considerations call for further guidance on where to focus empirical and normative ethical efforts to best support conduct clinical trials of AI and minimize unintended harm to trial participants.

Implementation of Artificial Intelligence-Based Diabetic Retinopathy Screening in a Tertiary Care Hospital in Quebec: Prospective Validation Study

BACKGROUND

Diabetic retinopathy (DR) affects about 25% of people with diabetes in Canada. Early detection of DR is essential for preventing vision loss.

OBJECTIVE

We evaluated the real-world performance of an artificial intelligence (AI) system that analyzes fundus images for DR screening in a Quebec tertiary care center.

METHODS

We prospectively recruited adult patients with diabetes at the Centre hospitalier de l'Université de Montréal (CHUM) in Montreal, Quebec, Canada. Patients underwent dual-pathway screening: first by the Computer Assisted Retinal Analysis (CARA) AI system (index test), then by standard ophthalmological examination (reference standard). We measured the AI system's sensitivity and specificity for detecting referable disease at the patient level, along with its performance for detecting any retinopathy and diabetic macular edema (DME) at the eye level, and potential cost savings.

RESULTS

This study included 115 patients. CARA demonstrated a sensitivity of 87.5% (95% CI 71.9-95.0) and specificity of 66.2% (95% CI 54.3-76.3) for detecting referable disease at the patient level. For any retinopathy detection at the eye level, CARA showed 88.2% sensitivity (95% CI 76.6-94.5) and 71.4% specificity (95% CI 63.7-78.1). For DME detection, CARA had 100% sensitivity (95% CI 64.6-100) and 81.9% specificity (95% CI 75.6-86.8). Potential yearly savings from implementing CARA at the CHUM were estimated at CAD $245,635 (US $177,643.23, as of July 26, 2024) considering 5000 patients with diabetes.

CONCLUSIONS

Our study indicates that integrating a semiautomated AI system for DR screening demonstrates high sensitivity for detecting referable disease in a real-world setting. This system has the potential to improve screening efficiency and reduce costs at the CHUM, but more work is needed to validate it.

Exploring general practice staff perspectives on a teaching concept based on instruction videos for diabetic retinopathy screening - an interview study

OBJECTIVE

The aim of this study is to explore general practice staff perspectives regarding a teaching concept based on instructional videos for conducting DR screenings. Furthermore, this study aims to investigate the competencies acquired by the staff through this teaching concept.

DESIGN AND SETTING

Qualitative cross-sectional study conducted in general practice clinics in the North Denmark Region.

METHOD

A teaching concept was developed based on instruction videos to teach general practice staff to conduct diabetic retinopathy screenings with automated grading through artificial intelligence. Semi-structured interviews were performed with 16 staff members to investigate their perspectives on the concept and acquired competencies.

RESULTS

This study found no substantial resistance to the teaching concept from staff; however, participants' satisfaction with the methods employed in the instruction session, the progression of learning curves, screening competencies, and their acceptance of a known knowledge gap during screenings varied slightly among the participants.

CONCLUSION

This study showed that the teaching concept can be used to teach general practice staff to conduct diabetic retinopathy screenings. Staffs' perspectives on the teaching concept and acquired competencies varied, and this study suggest few adjustments to the concept to accommodate staff's preferences and establish more consistent competencies.

Ocular biomarkers: useful incidental findings by deep learning algorithms in fundus photographs

BACKGROUND/OBJECTIVES

Artificial intelligence can assist with ocular image analysis for screening and diagnosis, but it is not yet capable of autonomous full-spectrum screening. Hypothetically, false-positive results may have unrealized screening potential arising from signals persisting despite training and/or ambiguous signals such as from biomarker overlap or high comorbidity. The study aimed to explore the potential to detect clinically useful incidental ocular biomarkers by screening fundus photographs of hypertensive adults using diabetic deep learning algorithms.

SUBJECTS/METHODS

Patients referred for treatment-resistant hypertension were imaged at a hospital unit in Perth, Australia, between 2016 and 2022. The same 45° colour fundus photograph selected for each of the 433 participants imaged was processed by three deep learning algorithms. Two expert retinal specialists graded all false-positive results for diabetic retinopathy in non-diabetic participants.

RESULTS

Of the 29 non-diabetic participants misclassified as positive for diabetic retinopathy, 28 (97%) had clinically useful retinal biomarkers. The models designed to screen for fewer diseases captured more incidental disease. All three algorithms showed a positive correlation between severity of hypertensive retinopathy and misclassified diabetic retinopathy.

CONCLUSIONS

The results suggest that diabetic deep learning models may be responsive to hypertensive and other clinically useful retinal biomarkers within an at-risk, hypertensive cohort. Observing that models trained for fewer diseases captured more incidental pathology increases confidence in signalling hypotheses aligned with using self-supervised learning to develop autonomous comprehensive screening. Meanwhile, non-referable and false-positive outputs of other deep learning screening models could be explored for immediate clinical use in other populations.

Role of Artificial Intelligence in Retinal Diseases

Artificial intelligence (AI) has already found its way into ophthalmology, with the first approved algorithms that can be used in clinical routine. Retinal diseases in particular are proving to be an important area of application for AI, as they are the main cause of blindness and the number of patients suffering from retinal diseases is constantly increasing. At the same time, regular imaging using high-resolution modalities in a standardised and reproducible manner generates immense amounts of data that can hardly be processed by human experts. In addition, ophthalmology is constantly experiencing new developments and breakthroughs that require a re-evaluation of patient management in routine clinical practice. AI is able to analyse these volumes of data efficiently and objectively and also provide new insights into disease progression and therapeutic mechanisms by identifying relevant biomarkers. AI can make a significant contribution to screening, classification and prognosis of various retinal diseases and can ultimately be a clinical decision support system, that significantly reduces the burden on both everyday clinical practice and the healthcare system, by making more efficient use of costly and time-consuming resources.

Current research and future strategies for the management of vision-threatening diabetic retinopathy

Diabetic retinopathy (DR) is a major ocular complication of diabetes and the leading cause of blindness and visual impairment, particularly among adults of working-age adults. Although the medical and economic burden of DR is significant and its global prevalence is expected to increase, particularly in low- and middle-income countries, a large portion of vision loss caused by DR remains preventable through early detection and timely intervention. This perspective reviewed the latest developments in research and innovation in three areas, first novel biomarkers (including advanced imaging modalities, serum biomarkers, and artificial intelligence technology) to predict the incidence and progression of DR, second, screening and early detection of referable DR and vision-threatening DR (VTDR), and finally, novel therapeutic strategies for VTDR, including diabetic macular oedema (DME), with the goal of reducing diabetic blindness.

Bibliometric analysis of artificial intelligence in healthcare research: Trends and future directions

Objective

The presence of artificial intelligence (AI) in healthcare is a powerful and game-changing force that is completely transforming the industry as a whole. Using sophisticated algorithms and data analytics, AI has unparalleled prospects for improving patient care, streamlining operational efficiency, and fostering innovation across the healthcare ecosystem. This study conducts a comprehensive bibliometric analysis of research on AI in healthcare, utilising the SCOPUS database as the primary data source.

Methods

Preliminary findings from 2013 identified 153 publications on AI and healthcare. Between 2019 and 2023, the number of publications increased exponentially, indicating significant growth and development in the field. The analysis employs various bibliometric indicators to assess research production performance, science mapping techniques, and thematic mapping analysis.

Results

The study reveals insights into research hotspots, thematic focus, and emerging trends in AI and healthcare research. Based on an extensive examination of the Scopus database provides a brief overview and suggests potential avenues for further investigation.

Conclusion

This article provides valuable contributions to understanding the current landscape of AI in healthcare, offering insights for future research directions and informing strategic decision making in the field.

Ethical trade-offs in AI for mental health

It is expected that machine learning algorithms will enable better diagnosis, prognosis, and treatment in psychiatry. A central argument for deploying algorithmic methods in clinical decision-making in psychiatry is that they may enable not only faster and more accurate clinical judgments but also that they may provide a more objective foundation for clinical decisions. This article argues that the outputs of algorithms are never objective in the sense of being unaffected by human values and possibly biased choices. And it suggests that the best way to approach this is to ensure awareness of and transparency about the ethical trade-offs that must be made when developing an algorithm for mental health.

Using multimodal ultrasound including full-time-series contrast-enhanced ultrasound cines for identifying the nature of thyroid nodules

Background

Based on the conventional ultrasound images of thyroid nodules, contrast-enhanced ultrasound (CEUS) videos were analyzed to investigate whether CEUS improves the classification accuracy of benign and malignant thyroid nodules using machine learning (ML) radiomics and compared with radiologists.

Materials and methods

The B-mode ultrasound (B-US), real-time elastography (RTE), color doppler flow images (CDFI) and CEUS cines of patients from two centers were retrospectively gathered. Then, the region of interest (ROI) was delineated to extract the radiomics features. Seven ML algorithms combined with four kinds of radiomics data (B-US, B-US + CDFI + RTE, CEUS, and B-US + CDFI + RTE + CEUS) were applied to establish 28 models. The diagnostic performance of ML models was compared with interpretations from expert and nonexpert readers.

Results

A total of 181 thyroid nodules from 181 patients of 64 men (mean age, 42 years +/- 12) and 117 women (mean age, 46 years +/- 12) were included. Adaptive boosting (AdaBoost) achieved the highest area under the receiver operating characteristic curve (AUC) of 0.89 in the test set among 28 models when combined with B-US + CDFI + RTE + CEUS data and an AUC of 0.72 and 0.66 when combined with B-US and B-US + CDFI + RTE data. The AUC achieved by senior and junior radiologists was 0.78 versus (vs.) 0.69 (p > 0.05), 0.79 vs. 0.64 (p < 0.05), and 0.88 vs. 0.69 (p < 0.05) combined with B-US, B-US+CDFI+RTE and B-US+CDFI+RTE+CEUS, respectively.

Conclusion

With the addition of CEUS, the diagnostic performance was enhanced for all seven classifiers and senior radiologists based on conventional ultrasound images, while no enhancement was observed for junior radiologists. The diagnostic performance of ML models was similar to senior radiologists, but superior to those junior radiologists.

Advancing healthcare with artificial intelligence: diagnostic accuracy of machine learning algorithm in diagnosis of diabetic retinopathy in the Brazilian population

BACKGROUND

In healthcare systems in general, access to diabetic retinopathy (DR) screening is limited. Artificial intelligence has the potential to increase care delivery. Therefore, we trained and evaluated the diagnostic accuracy of a machine learning algorithm for automated detection of DR.

METHODS

We included color fundus photographs from individuals from 4 databases (primary and specialized care settings), excluding uninterpretable images. The datasets consist of images from Brazilian patients, which differs from previous work. This modification allows for a more tailored application of the model to Brazilian patients, ensuring that the nuances and characteristics of this specific population are adequately captured. The sample was fractionated in training (70%) and testing (30%) samples. A convolutional neural network was trained for image classification. The reference test was the combined decision from three ophthalmologists. The sensitivity, specificity, and area under the ROC curve of the algorithm for detecting referable DR (moderate non-proliferative DR; severe non-proliferative DR; proliferative DR and/or clinically significant macular edema) were estimated.

RESULTS

A total of 15,816 images (4590 patients) were included. The overall prevalence of any degree of DR was 26.5%. Compared with human evaluators (manual method of diagnosing DR performed by an ophthalmologist), the deep learning algorithm achieved an area under the ROC curve of 0.98 (95% CI 0.97-0.98), with a specificity of 94.6% (95% CI 93.8-95.3) and a sensitivity of 93.5% (95% CI 92.2-94.9) at the point of greatest efficiency to detect referable DR.

CONCLUSIONS

A large database showed that this deep learning algorithm was accurate in detecting referable DR. This finding aids to universal healthcare systems like Brazil, optimizing screening processes and can serve as a tool for improving DR screening, making it more agile and expanding care access.

DeepMonitoring: a deep learning-based monitoring system for assessing the quality of cornea images captured by smartphones

Smartphone-based artificial intelligence (AI) diagnostic systems could assist high-risk patients to self-screen for corneal diseases (e.g., keratitis) instead of detecting them in traditional face-to-face medical practices, enabling the patients to proactively identify their own corneal diseases at an early stage. However, AI diagnostic systems have significantly diminished performance in low-quality images which are unavoidable in real-world environments (especially common in patient-recorded images) due to various factors, hindering the implementation of these systems in clinical practice. Here, we construct a deep learning-based image quality monitoring system (DeepMonitoring) not only to discern low-quality cornea images created by smartphones but also to identify the underlying factors contributing to the generation of such low-quality images, which can guide operators to acquire high-quality images in a timely manner. This system performs well across validation, internal, and external testing sets, with AUCs ranging from 0.984 to 0.999. DeepMonitoring holds the potential to filter out low-quality cornea images produced by smartphones, facilitating the application of smartphone-based AI diagnostic systems in real-world clinical settings, especially in the context of self-screening for corneal diseases.

Expert-Level Detection of Referable Glaucoma from Fundus Photographs in a Safety Net Population: The AI and Teleophthalmology in Los Angeles Initiative

Purpose

To develop and test a deep learning (DL) algorithm for detecting referable glaucoma in the Los Angeles County (LAC) Department of Health Services (DHS) teleretinal screening program.

Methods

Fundus photographs and patient-level labels of referable glaucoma (defined as cup-to-disc ratio [CDR] ≥ 0.6) provided by 21 trained optometrist graders were obtained from the LAC DHS teleretinal screening program. A DL algorithm based on the VGG-19 architecture was trained using patient-level labels generalized to images from both eyes. Area under the receiver operating curve (AUC), sensitivity, and specificity were calculated to assess algorithm performance using an independent test set that was also graded by 13 clinicians with one to 15 years of experience. Algorithm performance was tested using reference labels provided by either LAC DHS optometrists or an expert panel of 3 glaucoma specialists.

Results

12,098 images from 5,616 patients (2,086 referable glaucoma, 3,530 non-glaucoma) were used to train the DL algorithm. In this dataset, mean age was 56.8 ± 10.5 years with 54.8% females and 68.2% Latinos, 8.9% Blacks, 2.7% Caucasians, and 6.0% Asians. 1,000 images from 500 patients (250 referable glaucoma, 250 non-glaucoma) with similar demographics (p ≥ 0.57) were used to test the DL algorithm. Algorithm performance matched or exceeded that of all independent clinician graders in detecting patient-level referable glaucoma based on LAC DHS optometrist (AUC = 0.92) or expert panel (AUC = 0.93) reference labels. Clinician grader sensitivity (range: 0.33-0.99) and specificity (range: 0.68-0.98) ranged widely and did not correlate with years of experience (p ≥ 0.49). Algorithm performance (AUC = 0.93) also matched or exceeded the sensitivity (range: 0.78-1.00) and specificity (range: 0.32-0.87) of 6 LAC DHS optometrists in the subsets of the test dataset they graded based on expert panel reference labels.

Conclusions

A DL algorithm for detecting referable glaucoma developed using patient-level data provided by trained LAC DHS optometrists approximates or exceeds performance by ophthalmologists and optometrists, who exhibit variable sensitivity and specificity unrelated to experience level. Implementation of this algorithm in screening workflows could help reallocate eye care resources and provide more reproducible and timely glaucoma care.

Feature Identification Using Interpretability Machine Learning Predicting Risk Factors for Disease Severity of In-Patients with COVID-19 in South Florida

Objective: The objective of the study was to establish an AI-driven decision support system by identifying the most important features in the severity of disease for Intensive Care Unit (ICU) with Mechanical Ventilation (MV) requirement, ICU, and InterMediate Care Unit (IMCU) admission for hospitalized patients with COVID-19 in South Florida. The features implicated in the risk factors identified by the model interpretability can be used to forecast treatment plans faster before critical conditions exacerbate. Methods: We analyzed eHR data from 5371 patients diagnosed with COVID-19 from South Florida Memorial Healthcare Systems admitted between March 2020 and January 2021 to predict the need for ICU with MV, ICU, and IMCU admission. A Random Forest classifier was trained on patients' data augmented by SMOTE, collected at hospital admission. We then compared the importance of features utilizing different model interpretability analyses, such as SHAP, MDI, and Permutation Importance. Results: The models for ICU with MV, ICU, and IMCU admission identified the following factors overlapping as the most important predictors among the three outcomes: age, race, sex, BMI, diarrhea, diabetes, hypertension, early stages of kidney disease, and pneumonia. It was observed that individuals over 65 years ('older adults'), males, current smokers, and BMI classified as 'overweight' and 'obese' were at greater risk of severity of illness. The severity was intensified by the co-occurrence of two interacting features (e.g., diarrhea and diabetes). Conclusions: The top features identified by the models' interpretability were from the 'sociodemographic characteristics', 'pre-hospital comorbidities', and 'medications' categories. However, 'pre-hospital comorbidities' played a vital role in different critical conditions. In addition to individual feature importance, the feature interactions also provide crucial information for predicting the most likely outcome of patients' conditions when urgent treatment plans are needed during the surge of patients during the pandemic.

Cost-effectiveness of incorporating self-imaging optical coherence tomography into fundus photography-based diabetic retinopathy screening

Diabetic macular edema (DME) has emerged as the foremost cause of vision loss in the population with diabetes. Early detection of DME is paramount, yet the prevailing screening, relying on two-dimensional and labor-intensive fundus photography (FP), results in frequent unwarranted referrals and overlooked diagnoses. Self-imaging optical coherence tomography (SI-OCT), offering fully automated, three-dimensional macular imaging, holds the potential to enhance DR screening. We conducted an observational study within a cohort of 1822 participants with diabetes, who received comprehensive assessments, including visual acuity testing, FP, and SI-OCT examinations. We compared the performance of three screening strategies: the conventional FP-based strategy, a combination strategy of FP and SI-OCT, and a simulated combination strategy of FP and manual SD-OCT. Additionally, we undertook a cost-effectiveness analysis utilizing Markov models to evaluate the costs and benefits of the three strategies for referable DR. We found that the FP + SI-OCT strategy demonstrated superior sensitivity (87.69% vs 61.53%) and specificity (98.29% vs 92.47%) in detecting DME when compared to the FP-based strategy. Importantly, the FP + SI-OCT strategy outperformed the FP-based strategy, with an incremental cost-effectiveness ratio (ICER) of $8016 per quality-adjusted life year (QALY), while the FP + SD-OCT strategy was less cost-effective, with an ICER of $45,754/QALY. Our results were robust to extensive sensitivity analyses, with the FP + SI-OCT strategy standing as the dominant choice in 69.36% of simulations conducted at the current willingness-to-pay threshold. In summary, incorporating SI-OCT into FP-based screening offers substantial enhancements in sensitivity, specificity for detecting DME, and most notably, cost-effectiveness for DR screening.

Accuracy of Autonomous Artificial Intelligence-Based Diabetic Retinopathy Screening in Real-Life Clinical Practice

Background: In diabetic retinopathy, early detection and intervention are crucial in preventing vision loss and improving patient outcomes. In the era of artificial intelligence (AI) and machine learning, new promising diagnostic tools have emerged. The IDX-DR machine (Digital Diagnostics, Coralville, IA, USA) represents a diagnostic tool that combines advanced imaging techniques, AI algorithms, and deep learning methodologies to identify and classify diabetic retinopathy. Methods: All patients that participated in our AI-based DR screening were considered for this study. For this study, all retinal images were additionally reviewed retrospectively by two experienced retinal specialists. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for the IDX-DR machine compared to the graders' responses. Results: We included a total of 2282 images from 1141 patients who were screened between January 2021 and January 2023 at the Jules Gonin Eye Hospital in Lausanne, Switzerland. Sensitivity was calculated to be 100% for 'no DR', 'mild DR', and 'moderate DR'. Specificity for no DR', 'mild DR', 'moderate DR', and 'severe DR' was calculated to be, respectively, 78.4%, 81.2%, 93.4%, and 97.6%. PPV was calculated to be, respectively, 36.7%, 24.6%, 1.4%, and 0%. NPV was calculated to be 100% for each category. Accuracy was calculated to be higher than 80% for 'no DR', 'mild DR', and 'moderate DR'. Conclusions: In this study, based in Jules Gonin Eye Hospital in Lausanne, we compared the autonomous diagnostic AI system of the IDX-DR machine detecting diabetic retinopathy to human gradings established by two experienced retinal specialists. Our results showed that the ID-x DR machine constantly overestimates the DR stages, thus permitting the clinicians to fully trust negative results delivered by the screening software. Nevertheless, all fundus images classified as 'mild DR' or greater should always be controlled by a specialist in order to assert whether the predicted stage is truly present.

Impact of Gold-Standard Label Errors on Evaluating Performance of Deep Learning Models in Diabetic Retinopathy Screening: Nationwide Real-World Validation Study

BACKGROUND

For medical artificial intelligence (AI) training and validation, human expert labels are considered the gold standard that represents the correct answers or desired outputs for a given data set. These labels serve as a reference or benchmark against which the model's predictions are compared.

OBJECTIVE

This study aimed to assess the accuracy of a custom deep learning (DL) algorithm on classifying diabetic retinopathy (DR) and further demonstrate how label errors may contribute to this assessment in a nationwide DR-screening program.

METHODS

Fundus photographs from the Lifeline Express, a nationwide DR-screening program, were analyzed to identify the presence of referable DR using both (1) manual grading by National Health Service England-certificated graders and (2) a DL-based DR-screening algorithm with validated good lab performance. To assess the accuracy of labels, a random sample of images with disagreement between the DL algorithm and the labels was adjudicated by ophthalmologists who were masked to the previous grading results. The error rates of labels in this sample were then used to correct the number of negative and positive cases in the entire data set, serving as postcorrection labels. The DL algorithm's performance was evaluated against both pre- and postcorrection labels.

RESULTS

The analysis included 736,083 images from 237,824 participants. The DL algorithm exhibited a gap between the real-world performance and the lab-reported performance in this nationwide data set, with a sensitivity increase of 12.5% (from 79.6% to 92.5%, P<.001) and a specificity increase of 6.9% (from 91.6% to 98.5%, P<.001). In the random sample, 63.6% (560/880) of negative images and 5.2% (140/2710) of positive images were misclassified in the precorrection human labels. High myopia was the primary reason for misclassifying non-DR images as referable DR images, while laser spots were predominantly responsible for misclassified referable cases. The estimated label error rate for the entire data set was 1.2%. The label correction was estimated to bring about a 12.5% enhancement in the estimated sensitivity of the DL algorithm (P<.001).

CONCLUSIONS

Label errors based on human image grading, although in a small percentage, can significantly affect the performance evaluation of DL algorithms in real-world DR screening.

Artificial intelligence integration in the drug lifecycle and in regulatory science: policy implications, challenges and opportunities

Artificial intelligence tools promise transformative impacts in drug development. Regulatory agencies face challenges in integrating AI while ensuring reliability and safety in clinical trial approvals, drug marketing authorizations, and post-market surveillance. Incorporating these technologies into the existing regulatory framework and agency practices poses notable challenges, particularly in evaluating the data and models employed for these purposes. Rapid adaptation of regulations and internal processes is essential for agencies to keep pace with innovation, though achieving this requires collective stakeholder collaboration. This article thus delves into the need for adaptations of regulations throughout the drug development lifecycle, as well as the utilization of AI within internal processes of medicine agencies.

Novel Technologies in Artificial Intelligence and Telemedicine for Glaucoma Screening

PURPOSE

To provide an overview of novel technologies in telemedicine and artificial intelligence (AI) approaches for cost-effective glaucoma screening.

METHODS/RESULTS

A narrative review was performed by summarizing research results, recent developments in glaucoma detection and care, and considerations related to telemedicine and AI in glaucoma screening. Telemedicine and AI approaches provide the opportunity for novel glaucoma screening programs in primary care, optometry, portable, and home-based settings. These approaches offer several advantages for glaucoma screening, including increasing access to care, lowering costs, identifying patients in need of urgent treatment, and enabling timely diagnosis and early intervention. However, challenges remain in implementing these systems, including integration into existing clinical workflows, ensuring equity for patients, and meeting ethical and regulatory requirements. Leveraging recent work towards standardized data acquisition as well as tools and techniques developed for automated diabetic retinopathy screening programs may provide a model for a cost-effective approach to glaucoma screening.

CONCLUSION

Leveraging novel technologies and advances in telemedicine and AI-based approaches to glaucoma detection show promise for improving our ability to detect moderate and advanced glaucoma in primary care settings and target higher individuals at high risk for having the disease.

Toward Realizing the Promise of AI in Precision Health Across the Spectrum of Care

Significant progress has been made in augmenting clinical decision-making using artificial intelligence (AI) in the context of secondary and tertiary care at large academic medical centers. For such innovations to have an impact across the spectrum of care, additional challenges must be addressed, including inconsistent use of preventative care and gaps in chronic care management. The integration of additional data, including genomics and data from wearables, could prove critical in addressing these gaps, but technical, legal, and ethical challenges arise. On the technical side, approaches for integrating complex and messy data are needed. Data and design imperfections like selection bias, missing data, and confounding must be addressed. In terms of legal and ethical challenges, while AI has the potential to aid in leveraging patient data to make clinical care decisions, we also risk exacerbating existing disparities. Organizations implementing AI solutions must carefully consider how they can improve care for all and reduce inequities.

Evaluation of an AI algorithm trained on an ethnically diverse dataset to screen a previously unseen population for diabetic retinopathy

PURPOSE

This study aimed to determine the generalizability of an artificial intelligence (AI) algorithm trained on an ethnically diverse dataset to screen for referable diabetic retinopathy (RDR) in the Armenian population unseen during AI development.

METHODS

This study comprised 550 patients with diabetes mellitus visiting the polyclinics of Armenia over 10 months requiring diabetic retinopathy (DR) screening. The Medios AI-DR algorithm was developed using a robust, diverse, ethnically balanced dataset with no inherent bias and deployed offline on a smartphone-based fundus camera. The algorithm here analyzed the retinal images captured using the target device for the presence of RDR (i.e., moderate non-proliferative diabetic retinopathy (NPDR) and/or clinically significant diabetic macular edema (CSDME) or more severe disease) and sight-threatening DR (STDR, i.e., severe NPDR and/or CSDME or more severe disease). The results compared the AI output to a consensus or majority image grading of three expert graders according to the International Clinical Diabetic Retinopathy severity scale.

RESULTS

On 478 subjects included in the analysis, the algorithm achieved a high classification sensitivity of 95.30% (95% CI: 91.9%-98.7%) and a specificity of 83.89% (95% CI: 79.9%-87.9%) for the detection of RDR. The sensitivity for STDR detection was 100%.

CONCLUSION

The study proved that Medios AI-DR algorithm yields good accuracy in screening for RDR in the Armenian population. In our literature search, this is the only smartphone-based, offline AI model validated in different populations.

Retinal imaging and Alzheimer's disease: a future powered by Artificial Intelligence

Alzheimer's disease (AD) is a neurodegenerative condition that primarily affects brain tissue. Because the retina and brain share the same embryonic origin, visual deficits have been reported in AD patients. Artificial Intelligence (AI) has recently received a lot of attention due to its immense power to process and detect image hallmarks and make clinical decisions (like diagnosis) based on images. Since retinal changes have been reported in AD patients, AI is being proposed to process images to predict, diagnose, and prognosis AD. As a result, the purpose of this review was to discuss the use of AI trained on retinal images of AD patients. According to previous research, AD patients experience retinal thickness and retinal vessel density changes, which can occasionally occur before the onset of the disease's clinical symptoms. AI and machine vision can detect and use these changes in the domains of disease prediction, diagnosis, and prognosis. As a result, not only have unique algorithms been developed for this condition, but also databases such as the Retinal OCTA Segmentation dataset (ROSE) have been constructed for this purpose. The achievement of high accuracy, sensitivity, and specificity in the classification of retinal images between AD and healthy groups is one of the major breakthroughs in using AI based on retinal images for AD. It is fascinating that researchers could pinpoint individuals with a positive family history of AD based on the properties of their eyes. In conclusion, the growing application of AI in medicine promises its future position in processing different aspects of patients with AD, but we need cohort studies to determine whether it can help to follow up with healthy persons at risk of AD for a quicker diagnosis or assess the prognosis of patients with AD.

Artificial intelligence for diabetes care: current and future prospects

Artificial intelligence (AI) use in diabetes care is increasingly being explored to personalise care for people with diabetes and adapt treatments for complex presentations. However, the rapid advancement of AI also introduces challenges such as potential biases, ethical considerations, and implementation challenges in ensuring that its deployment is equitable. Ensuring inclusive and ethical developments of AI technology can empower both health-care providers and people with diabetes in managing the condition. In this Review, we explore and summarise the current and future prospects of AI across the diabetes care continuum, from enhancing screening and diagnosis to optimising treatment and predicting and managing complications.

Data-driven decisions about individual patients: The case of medical AI

There are high hopes that clinical decisions can be improved by adopting algorithms trained to estimate the likelihood that a patient suffers a condition C. Introducing work on the epistemic value of purely statistical evidence in legal epistemology I show that a certain type of AI devices for making medical decisions about persons rely on purely statistical evidence and that it raises an important question about the appropriateness of relying on such devices for allocating health resources. If the argument I present is sound, then it suggests a radical rethinking of the use of prevalent types of AI devices as well as the use of statistical evidence in medical practice more generally.

Automatic Determination of Endothelial Cell Density From Donor Cornea Endothelial Cell Images

Purpose

To determine endothelial cell density (ECD) from real-world donor cornea endothelial cell (EC) images using a self-supervised deep learning segmentation model.

Methods

Two eye banks (Eversight, VisionGift) provided 15,138 single, unique EC images from 8169 donors along with their demographics, tissue characteristics, and ECD. This dataset was utilized for self-supervised training and deep learning inference. The Cornea Image Analysis Reading Center (CIARC) provided a second dataset of 174 donor EC images based on image and tissue quality. These images were used to train a supervised deep learning cell border segmentation model. Evaluation between manual and automated determination of ECD was restricted to the 1939 test EC images with at least 100 cells counted by both methods.

Results

The ECD measurements from both methods were in excellent agreement with rc of 0.77 (95% confidence interval [CI], 0.75-0.79; P < 0.001) and bias of 123 cells/mm2 (95% CI, 114-131; P < 0.001); 81% of the automated ECD values were within 10% of the manual ECD values. When the analysis was further restricted to the cropped image, the rc was 0.88 (95% CI, 0.87-0.89; P < 0.001), bias was 46 cells/mm2 (95% CI, 39-53; P < 0.001), and 93% of the automated ECD values were within 10% of the manual ECD values.

Conclusions

Deep learning analysis provides accurate ECDs of donor images, potentially reducing analysis time and training requirements.

Translational Relevance

The approach of this study, a robust methodology for automatically evaluating donor cornea EC images, could expand the quantitative determination of endothelial health beyond ECD.

Current Applications and Future Perspectives of Artificial and Biomimetic Intelligence in Vascular Surgery and Peripheral Artery Disease

Artificial Intelligence (AI) made its first appearance in 1956, and since then it has progressively introduced itself in healthcare systems and patients' information and care. AI functions can be grouped under the following headings: Machine Learning (ML), Deep Learning (DL), Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Computer Vision (CV). Biomimetic intelligence (BI) applies the principles of systems of nature to create biological algorithms, such as genetic and neural network, to be used in different scenarios. Chronic limb-threatening ischemia (CLTI) represents the last stage of peripheral artery disease (PAD) and has increased over recent years, together with the rise in prevalence of diabetes and population ageing. Nowadays, AI and BI grant the possibility of developing new diagnostic and treatment solutions in the vascular field, given the possibility of accessing clinical, biological, and imaging data. By assessing the vascular anatomy in every patient, as well as the burden of atherosclerosis, and classifying the level and degree of disease, sizing and planning the best endovascular treatment, defining the perioperative complications risk, integrating experiences and resources between different specialties, identifying latent PAD, thus offering evidence-based solutions and guiding surgeons in the choice of the best surgical technique, AI and BI challenge the role of the physician's experience in PAD treatment.

Hybrid deep learning models for the screening of Diabetic Macular Edema in optical coherence tomography volumes

Several studies published so far used highly selective image datasets from unclear sources to train computer vision models and that may lead to overestimated results, while those studies conducted in real-life remain scarce. To avoid image selection bias, we stacked convolutional and recurrent neural networks (CNN-RNN) to analyze complete optical coherence tomography (OCT) cubes in a row and predict diabetic macular edema (DME), in a real-world diabetic retinopathy screening program. A retrospective cohort study was carried out. Throughout 4-years, 5314 OCT cubes from 4408 subjects who attended to the diabetic retinopathy (DR) screening program were included. We arranged twenty-two (22) pre-trained CNNs in parallel with a bidirectional RNN layer stacked at the bottom, allowing the model to make a prediction for the whole OCT cube. The staff of retina experts built a ground truth of DME later used to train a set of these CNN-RNN models with different configurations. For each trained CNN-RNN model, we performed threshold tuning to find the optimal cut-off point for binary classification of DME. Finally, the best models were selected according to sensitivity, specificity, and area under the receiver operating characteristics curve (AUROC) with their 95% confidence intervals (95%CI). An ensemble of the best models was also explored. 5188 cubes were non-DME and 126 were DME. Three models achieved an AUROC of 0.94. Among these, sensitivity, and specificity (95%CI) ranged from 84.1-90.5 and 89.7-93.3, respectively, at threshold 1, from 89.7-92.1 and 80-83.1 at threshold 2, and from 80.2-81 and 93.8-97, at threshold 3. The ensemble model improved these results, and lower specificity was observed among subjects with sight-threatening DR. Analysis by age, gender, or grade of DME did not vary the performance of the models. CNN-RNN models showed high diagnostic accuracy for detecting DME in a real-world setting. This engine allowed us to detect extra-foveal DMEs commonly overlooked in other studies, and showed potential for application as the first filter of non-referable patients in an outpatient center within a population-based DR screening program, otherwise ended up in specialized care.

Autonomous artificial intelligence for diabetic eye disease increases access and health equity in underserved populations

Diabetic eye disease (DED) is a leading cause of blindness in the world. Annual DED testing is recommended for adults with diabetes, but adherence to this guideline has historically been low. In 2020, Johns Hopkins Medicine (JHM) began deploying autonomous AI for DED testing. In this study, we aimed to determine whether autonomous AI implementation was associated with increased adherence to annual DED testing, and how this differed across patient populations. JHM primary care sites were categorized as "non-AI" (no autonomous AI deployment) or "AI-switched" (autonomous AI deployment by 2021). We conducted a propensity score weighting analysis to compare change in adherence rates from 2019 to 2021 between non-AI and AI-switched sites. Our study included all adult patients with diabetes (>17,000) managed within JHM and has three major findings. First, AI-switched sites experienced a 7.6 percentage point greater increase in DED testing than non-AI sites from 2019 to 2021 (p < 0.001). Second, the adherence rate for Black/African Americans increased by 12.2 percentage points within AI-switched sites but decreased by 0.6% points within non-AI sites (p < 0.001), suggesting that autonomous AI deployment improved access to retinal evaluation for historically disadvantaged populations. Third, autonomous AI is associated with improved health equity, e.g. the adherence rate gap between Asian Americans and Black/African Americans shrank from 15.6% in 2019 to 3.5% in 2021. In summary, our results from real-world deployment in a large integrated healthcare system suggest that autonomous AI is associated with improvement in overall DED testing adherence, patient access, and health equity.

Screening for diabetic retinopathy with artificial intelligence: a real world evaluation

AIM

Periodic screening for diabetic retinopathy (DR) is effective for preventing blindness. Artificial intelligence (AI) systems could be useful for increasing the screening of DR in diabetic patients. The aim of this study was to compare the performance of the DAIRET system in detecting DR to that of ophthalmologists in a real-world setting.

METHODS

Fundus photography was performed with a nonmydriatic camera in 958 consecutive patients older than 18 years who were affected by diabetes and who were enrolled in the DR screening in the Diabetes and Endocrinology Unit and in the Eye Unit of ULSS8 Berica (Italy) between June 2022 and June 2023. All retinal images were evaluated by DAIRET, which is a machine learning algorithm based on AI. In addition, all the images obtained were analysed by an ophthalmologist who graded the images. The results obtained by DAIRET were compared with those obtained by the ophthalmologist.

RESULTS

We included 958 patients, but only 867 (90.5%) patients had retinal images sufficient for evaluation by a human grader. The sensitivity for detecting cases of moderate DR and above was 1 (100%), and the sensitivity for detecting cases of mild DR was 0.84 ± 0.03. The specificity of detecting the absence of DR was lower (0.59 ± 0.04) because of the high number of false-positives.

CONCLUSION

DAIRET showed an optimal sensitivity in detecting all cases of referable DR (moderate DR or above) compared with that of a human grader. On the other hand, the specificity of DAIRET was low because of the high number of false-positives, which limits its cost-effectiveness.

Implementation Mapping of the Collaborative University of California Teleophthalmology Initiative (CUTI): A Qualitative Study Using the Exploration, Preparation, Implementation, and Sustainment (EPIS) Framework

Background This study aimed to investigate the rationale, barriers, and facilitators of teleretinal camera implementation in primary care and endocrinology clinics for diabetic retinopathy (DR) screening across University of California (UC) health systems utilizing the Exploration, Preparation, Implementation, and Sustainment (EPIS) framework. Methodology Institutional representatives from UC Los Angeles, San Diego, San Francisco, and Davis participated in a series of focus group meetings to elicit implementation facilitators and barriers for teleophthalmology programs within their campuses. Site representatives also completed a survey regarding their program's performance over the calendar year 2022 in the following areas: DR screening camera sites, payment sources and coding, screening workflows including clinical, information technology (IT), reading, results, pathologic findings, and follow-up, including patient outreach for abnormal results. Focus group and survey results were mapped to the EPIS framework to gain insights into the implementation process of these programs and identify areas for optimization. Results Four UC campuses with 20 active camera sites screened 7,450 patients in the calendar year 2022. The average DR screening rate across the four campuses was 55%. Variations between sources of payment, turn-around time, image-grading structure, image-report characteristics, IT infrastructure, and patient outreach strategies were identified between sites. Closing gaps in IT integration between data systems, ensuring the financial sustainability of the program, and optimizing patient outreach remain primary challenges across sites and serve as good opportunities for cross-institutional learning. Conclusions Despite the potential for long-term cost savings and improving access to care, numerous obstacles continue to hinder the widespread implementation of teleretinal DR screening. Implementation science approaches can identify strategies for addressing these challenges and optimizing implementation.

Natural Language Processing in medicine and ophthalmology: A review for the 21st-century clinician

Natural Language Processing (NLP) is a subfield of artificial intelligence that focuses on the interaction between computers and human language, enabling computers to understand, generate, and derive meaning from human language. NLP's potential applications in the medical field are extensive and vary from extracting data from Electronic Health Records -one of its most well-known and frequently exploited uses- to investigating relationships among genetics, biomarkers, drugs, and diseases for the proposal of new medications. NLP can be useful for clinical decision support, patient monitoring, or medical image analysis. Despite its vast potential, the real-world application of NLP is still limited due to various challenges and constraints, meaning that its evolution predominantly continues within the research domain. However, with the increasingly widespread use of NLP, particularly with the availability of large language models, such as ChatGPT, it is crucial for medical professionals to be aware of the status, uses, and limitations of these technologies.

Big data to guide glaucoma treatment

Ophthalmology has been at the forefront of the medical application of big data. Often harnessed with a machine learning approach, big data has demonstrated potential to transform ophthalmic care, as evidenced by prior success on clinical tasks such as the screening of ophthalmic diseases and lesions via retinal images. With the recent establishment of various large ophthalmic datasets, there has been greater interest in determining whether the benefits of big data may extend to the downstream process of ophthalmic disease management. An area of substantial investigation has been the use of big data to help guide or streamline management of glaucoma, which remains a leading cause of irreversible blindness worldwide. In this review, we summarize relevant studies utilizing big data and discuss the application of the findings in the risk assessment and treatment of glaucoma.