Detection of diabetic retinopathy using artificial intelligence: an exploratory systematic review

doi: 10.62486/latia2024112

SYSTEMATIC REVIEW

Detection of diabetic retinopathy using artificial intelligence: an exploratory systematic review

Detección de retinopatía diabética utilizando inteligencia artificial: una revisión sistemática exploratoria

Richard Injante¹ *, Marck Julca¹ *

¹Universidad Nacional de San Martín, Facultad de Ingeniería de Sistemas e Informática. Tarapoto, Perú.

Cite as: Injante R, Julca M. Detection of diabetic retinopathy using artificial intelligence: an exploratory systematic review. LatIA. 2024; 2:112. https://doi.org/10.62486/latia2024112

Submitted: 21-02-2024 Revised: 08-05-2024 Accepted: 04-09-2024 Published: 05-09-2024

Editor: Prof. Dr. Javier González Argote

Corresponding author: Richard Injante *

ABSTRACT

Diabetic retinopathy is a disease that can lead to vision loss and blindness in people with diabetes, so its early detection is important to prevent ocular complications. The aim of this study was to analyze the usefulness of artificial intelligence in the detection of diabetic retinopathy. For this purpose, an exploratory systematic review was performed, collecting 77 empirical articles from the Scopus, IEEE, ACM, SciELO and NIH databases. The results indicate that the most commonly used factors for the detection of diabetic retinopathy include changes in retinal vascularization, macular edema and microaneurysms. Among the most commonly applied algorithms for early detection are ResNet 101, CNN and IDx-DR. In addition, some artificial intelligence models are reported to have an accuracy ranging from 90 % to 95 %, although models with accuracies below 80 % have also been identified. It is concluded that artificial intelligence, and in particular deep learning, has been shown to be effective in the early detection of diabetic retinopathy, facilitating timely treatment and improving clinical outcomes. However, ethical and legal concerns arise, such as privacy and security of patient data, liability in case of diagnostic errors, algorithmic bias, informed consent, and transparency in the use of artificial intelligence.

Keywords: Deep Learning; CNN; Resnet 101; Blindness; Macular Edema.

RESUMEN

La retinopatía diabética es una enfermedad que puede llevar a la pérdida de visión y ceguera en personas con diabetes, por lo que su detección temprana es importante para prevenir complicaciones oculares. El objetivo de este estudio fue analizar la utilidad de la inteligencia artificial en la detección de la retinopatía diabética. Para ello, se realizó una revisión sistemática exploratoria, recopilando 77 artículos empíricos de las bases de datos Scopus, IEEE, ACM, SciELO y NIH. Los resultados indican que los factores más utilizados para la detección de retinopatía diabética incluyen cambios en la vascularización de la retina, edema macular y microaneurismas. Entre los algoritmos más aplicados para la detección temprana se encuentran ResNet 101, CNN e IDx-DR. Además, se reporta que algunos modelos de inteligencia artificial presentan una precisión que varía entre el 90 % y el 95 %, aunque también se han identificado modelos con precisiones inferiores al 80 %. Se concluye que la inteligencia artificial, y en particular el aprendizaje profundo, ha demostrado ser efectiva en la detección temprana de retinopatía diabética, facilitando un tratamiento oportuno y mejorando los resultados clínicos. Sin embargo, surgen preocupaciones éticas y legales, tales como la privacidad y seguridad de los datos del paciente, la responsabilidad en caso de errores de diagnóstico, el sesgo algorítmico, el consentimiento informado y la transparencia en el uso de la inteligencia artificial.

Palabras clave: Aprendizaje Profundo; CNN; Resnet 101; Ceguera; Edema Macular.

INTRODUCTION

According to ⁽¹⁾, Diabetic Retinopathy (DR) is an eye condition that causes vision loss and blindness in people with diabetes, as it affects the blood vessels of the retina. Therefore, detecting DR in its early stages is vital because timely treatment can prevent ocular complications, highlighting the importance of early detection in patients with diabetes.

One of the main issues is the need for accurate diagnosis of DR to mitigate its impact and reduce reliance on human resources. Artificial Intelligence (AI) offers advances in this field, improving pattern recognition and accurate diagnoses.⁽²⁾ However, the accessibility and costs of AI in low- and middle-income countries remain uncertain, as do the necessary regulatory approvals.⁽¹⁾ In addition, the lack of specialized resources for early diagnosis highlights the need for accurate solutions and real-time monitoring that AI can provide.⁽³⁾

Implementing AI for this purpose faces limitations, including the need for extensive resources, workflow issues, and technology integration and privacy issues.⁽⁴⁾ Mobile devices present image quality issues compared with nonportable retinal cameras, and AI algorithms must be adapted to national DR detection initiatives.⁽⁵⁾ Improving data interpretation and overcoming the lack of training data is critical, especially in rural areas and developing countries.^(6,7) Assessing biases in AI algorithms is essential to ensure representativeness and applicability.⁽⁸⁾ In addition, technological evolution requires constant updates in DR detection programs.⁽⁹⁾ The lack of standards and demographic variability in AI studies highlight the need for comparative evaluations and legal and ethical considerations.⁽¹⁰⁾Although AI has the potential to improve the diagnosis and treatment of DR significantly, it faces barriers in terms of accessibility, costs, data quality, and assessment of biases and standards.^(4,11,12)

Systematic review studies on DR screening have demonstrated how AI has revolutionized the diagnosis of this and other retinal diseases⁽²⁾ and highlighted that the ability of AI to recognize patterns and compare them to standards has driven early detection and timely treatment, promoting efficiency and reducing disease burden. Furthermore, the research in ⁽¹⁾ shows that AI in DR detection exhibits high sensitivities and specificities; however, more research on regulatory approvals and improvements in health outcomes is required before large-scale implementation.

In this framework, deep learning, an AI technique, enables early and accurate detection of DR, facilitating personalized treatments and real-time monitoring, thus improving patients’ quality of life.⁽³⁾ Despite its efficacy,⁽⁴⁾ noted that AI implementation is limited to integration into healthcare systems and ethical and economic issues.⁽⁵⁾ also mentions that, although AI shows high sensitivity, its clinical implementation is limited due to legal and technical obstacles.

Advances in AI, specifically with deep learning algorithms and the Inception-v3 model, demonstrated high efficacy in diagnosing ocular diseases, comparable to human experts with less training data^(6,7) emphasize that AI is fundamental for early detection and classification of DR, addressing future areas of research in ophthalmology. However, there are concerns about bias and lack of representativeness in studies, as well as the need for economic studies to support its implementation.⁽⁸⁾Therefore, although AI has improved the early and effective detection of DR, further research is required to overcome current barriers and ensure its efficacy and safety in clinical practice.⁽¹⁰⁾

The present study aimed to analyze the detection of DR by AI, identifying the factors and variables considered by the models, the most commonly used algorithms, and the accuracy of these algorithms. The aim is to examine the technical limitations in detection and the ethical and legal challenges associated with its implementation.

METHOD

The present study was based on an exploratory systematic review, whose main objective is to analyze and synthesize DR detection using AI. This type of review allows us to obtain a complete and rigorous understanding of the current state of knowledge by identifying, evaluating, and critically synthesizing the results of previous studies. Systematic review is essential to identify gaps in knowledge and establish a solid foundation for future research and informed decisions.

Research questions

In order to answer the proposed objective, the following research questions were formulated:

Q1: What are the factors or variables considered by the models in DR detection?

Q2: What are the most effective algorithms for early detection of DR using AI?

Q3: What is the accuracy of AI models in DR detection?

Q4: What technical or implementation limitations do AI systems face in DR detection?

Q5: What ethical and legal challenges are posed by the use of AI algorithms in the diagnosis and treatment of DR?

Search strategy

The key term “diabetic retinopathy,” translated into Spanish as “retinopatía diabética,” was used. In addition, AND and OR connectors were used to introduce related terms such as “artificial intelligence.” The Scopus, ACM, SciELO, and NIH databases were selected because of their broad international and regional coverage, as well as the ease of applying advanced search filters and ensuring the academic quality of the contributions.

A selection process was implemented in five phases: i) First classification consisted of applying the search strings using the advanced tools of each database, mainly using the title fields; ii) Second classification, filters were applied, taking into consideration the manually applicable inclusion and exclusion criteria such as the coverage of the years, language and type of document. iii) Third classification: Metadata was imported and organized in Excel, where duplicates were purified using the Scopus database as a reference. iv) Fourth classification: Filter articles focused on AI for DR detection by reading the title and abstract. v) Fifth classification, a final selection of downloadable articles relevant to the study was made.

Table 1 shows the number of articles obtained in each classification:

Table 1. Number of items obtained by classification
Database	Ranking
Database	1st	2nd	3rd	4th	5th
Scopus	195	102	99	96	74
IEEE	11	11	9	9	0
ACM	24	20	20	-	-
SciELO	2	2	2	2	2
NIH	2	2	2	2	1
Total	234	137	129	109	77

Inclusion and exclusion criteria

The definition of inclusion and exclusion criteria helps to guarantee the objectivity and reproducibility of the review, ensuring that the studies are relevant and meet quality standards.⁽¹³⁾ Therefore, the following inclusion criteria were defined: published between 2019 and 2024; Spanish and English language; original articles in IEEE, SCOPUS, ACM, and SciELO; focused on the application of AI, specifically Deep Learning, for DR detection. Regarding exclusion criteria, the following were established: review articles or other types of secondary sources; duplicate publications; articles without access to the full text or not downloadable.

Data extraction

After the final selection of the 77 articles, we proceeded to the extraction of relevant data, including the identification of factors and variables considered by the models, the evaluation of the most effective algorithms, the accuracy of these models, and the technical, ethical, and legal limitations facing the implementation of artificial intelligence in the detection of diabetic retinopathy. This methodology ensures a thorough and rigorous review, providing a solid basis for evaluating the efficacy and challenges of AI in the diagnosis and treatment of DR.

RESULTS AND DISCUSSION

After selecting relevant articles and conducting a thorough review of each article, we proceeded to answer the research questions posed.

Q1: What are the factors or variables considered by the models in the detection of DR?

This question refers to the different elements that AI models analyze to detect DR. It is crucial to understand which factors and variables, such as clinical characteristics, imaging attributes, and demographic parameters, are used, as these determine the accuracy and effectiveness of the diagnosis. Knowing these aspects helps to improve the models and their application in various clinical settings.

Table 2. Factors or variables considered by the models in the detection of diabetic retinopathy
Factor/Variable	Articles
Neovascularization	^{(14); (15); (16); (17); (18); (19); (20); (8); (21); (22); (23); (24); (25); (26); (27); (28); (29); (30); (31); (32); (33); (34); (35); (36); (37); (38); (39)}
Macular edema	^{(40); (41); (42); (43); (44); (45); (46); (47); (15); (48); (16); (17); (49); (18); (50); (20); (8); (51); (52); (24); (26); (28); (53); (54); (55); (56); (57); (35); (58); (59); (60); (61); (62); (38); (39)}
Changes in retinal vascularization	^{(40); (4); (41); (63); (64); (65); (66); (42); (43); (14); (44); (45); (67); (46); (47); (15); (3); (68); (69); (17); (49); (18); (1); (70); (19); (20); (71); (8); (51); (72); (22); (73); (25); (28); (53); (31); (32); (54); (33); (74); (35); (58); (60); (75); (61); (76); (62); (37); (38); (77)}
Optic disc size	^{(78); (4); (41); (63); (64); (65); (42); (67); (15); (79); (48); (69); (17); (49); (20); (8); (80); (23); (24); (26); (28); (29); (54); (34); (75); (61); (36); (37)}
Microaneurysms	^{(78); (4); (41); (81); (63); (64); (65); (66); (42); (43); (14); (82); (44); (67); (46); (15); (79); (48); (16); (3); (68); (69); (17); (49); (18); (1); (70); (50); (83); (19); (20); (84); (71); (8); (51); (52); (21); (80); (22); (23); (24); (73); (25); (26); (28); (53); (29); (30); (32); (54); (85); (56); (33); (34); (74); (57); (35); (58); (59); (60); (75); (86); (61); (36); (76); (62); (37); (38); (77); (39)}
Exudates	^{(78); (4); (41); (63); (64); (65); (66); (42); (14); (44); (67); (46); (15); (79); (48); (16); (68); (69); (17); (49); (18); (1); (70); (50); (83); (19); (20); (84); (71); (8); (51); (21); (80); (22); (23); (24); (73); (25); (26); (28); (53); (29); (30); (32); (54); (85); (56); (33); (34); (74); (57); (35); (59); (60); (75); (86); (61); (36); (76); (62); (37); (38); (77)}

Table 2 provides a detailed overview of the different factors or variables considered by the models in the detection of DR. The results indicate that microaneurysms are the most recurrently addressed factor, with a presence in 71 of the 77 articles collected. This suggests that microaneurysms are a prominent and relevant feature in the detection and diagnosis of DR, according to the literature reviewed. The high incidence of microaneurysms in the models may indicate their importance as an early marker of DR. These small dilations of capillaries are indicative of retinal vascular damage, some associated with diabetes mellitus. However, other factors are also highlighted in the literature, such as macular edema and changes in retinal vascularization, which are present in 35 and 53 articles, respectively. These findings suggest that models also pay attention to the presence of macular edema and vascular changes as key indicators of disease progression. The inclusion of different factors in the models reflects the diversity of DR manifestations and the importance of a comprehensive approach to its detection and clinical management. The focus on exudates and changes in vascularization reflects the need for AI models to address both early signs and those indicative of advanced disease.

Q2: What are the most effective algorithms for early detection of DR using AI?

This question seeks to identify the specific algorithms that have been shown to be most effective in the early detection of DR. It is important to know these algorithms to understand which ones provide the best results in terms of sensitivity, specificity, and diagnostic accuracy, which is crucial to optimize the use of AI in eye care.

Table 3. Algorithms for early detection of DR with AI
Algorithms	Article(s)
ResNet 101	^{(40), (16), (30), (24), (59)}
CNN	^{(78), (68), (21), (62)}
IDx-DR	^{(4), (45), (8), (25), (60)}
Retinalyze	⁽⁴⁾
VeriSee DR	⁽⁶⁵⁾
Deep Learning (DL)	^{(66), (79), (39)}
EyeWisdom	⁽¹⁴⁾
DART (TeleDx)	^{(82), (77)}
AI general	^{(44), (47), (73), (85)}
Ensemble AI	⁽⁴⁶⁾
SqueezeNet	⁽⁵⁰⁾
AlexNet	⁽¹⁹⁾
Modified GoogLeNet	⁽²⁰⁾
Grey Level Co-occurrence Matrix	⁽⁸⁴⁾
VGG-16	⁽⁷¹⁾
EyeArt™	^{(51), (31)}
EyeCheckup AI	⁽⁵²⁾
Deep Convolutional Neural Network	⁽⁶⁹⁾
Support Vector Machine (SVM)	^{(69), (68)}
PNN	⁽⁶⁸⁾
ExplAIn	⁽¹⁷⁾
Selena + Aurora	⁽⁵³⁾
THEIA™	⁽⁵⁵⁾
EBM y XGBoost	⁽³²⁾
Medios AI	^{(57), (75)}
PhelcomNet	⁽⁸⁶⁾
Phoebus Algorithm	⁽⁶¹⁾
RetCAD AI	⁽³³⁾
DeepDR	⁽⁷⁴⁾
ResNet-50	⁽³⁸⁾
DLA	⁽⁷⁶⁾

Table 3 shows a variety of algorithms used in the early detection of DR using AI. The algorithms included in the list are representative of the different approaches and approaches used in the detection of this disease. Among the most commonly used algorithms are those based on convolutional neural networks (CNN) and those based on deep learning (DL). These algorithms have proven to be effective in DR detection due to their ability to learn patterns in images and detect changes in the retina. Another group of algorithms includes those based on supervised learning machines, such as SVM and PNN. These algorithms have also been shown to be effective in DR detection, as they can learn to recognize patterns in images and classify them according to their correspondence with the disease. The inclusion of algorithms such as EyeArt™, EyeCheckup AI, and Medios AI in the list suggests that early detection of DR can also be performed using machine learning tools and computer vision systems.

Q3: How accurate are AI models in DR detection?

This question examines how accurate AI models are in detecting DR. It is essential to assess the accuracy of these models, as this determines their feasibility and usefulness in clinical practice. Knowing the accuracy allows healthcare professionals to rely on these systems for diagnosis and management of the disease.

Table 4. Accuracy of AI models in DR detection
Accuracy in % of algorithm	Items (Algorithms, Percentage)
< 80 %	⁽⁵⁹⁾ (ResNet 101, 79 %), ⁽⁴¹⁾ (CNN, 76 %), ⁽⁸⁶⁾ (Medios AI, 78 %)
80 - 85 %	⁽¹⁶⁾ (ResNet 101, 83 %), ⁽⁷⁸⁾ (CNN, 84 %), ⁽⁶⁹⁾(DL, 82 %), ⁽⁶⁸⁾(SVM, 84 %), ⁽⁶⁹⁾ (SVM, 83 %), ⁽¹⁴⁾ (EyeWisdom, 82 %), ⁽⁵⁰⁾(SqueezeNet, 84 %)
85 - 90 %	⁽³⁰⁾ (ResNet 101, 89 %), ⁽⁶⁸⁾ (CNN, 87 %), ⁽⁵³⁾ (Selena + Aurora, 86 %), ⁽⁸⁾ (IDx-DR, 88 %), ⁽⁸²⁾ (DART, 88 %), ⁽⁷³⁾ (AI general, 87 %), ⁽⁷⁵⁾ (Medios AI, 87 %), ⁽¹⁹⁾ (AlexNet, 88 %), ⁽⁷⁹⁾ (DL, 85 %), ⁽³²⁾ (EBM y XGBoost, 89 %)
90 - 95 %	⁽⁴⁰⁾ (ResNet 101, 98 %), ⁽⁶²⁾ (CNN, 92 %), ⁽⁴⁾ (IDx-DR, 95 %), ⁽⁴⁵⁾ (IDx-DR, 92 %), ⁽²⁵⁾ (IDx-DR, 93 %), ⁽⁶⁰⁾ (IDx-DR, 91 %), ⁽⁵¹⁾ (EyeArt™, 95 %), ⁽³¹⁾ (EyeArt™, 90 %), ⁽⁵²⁾ (EyeCheckup AI, 92 %), ⁽⁸⁴⁾ (Grey Level Co-occurrence Matrix, 85 %), ⁽⁷¹⁾ (VGG-16, 91 %), ⁽²⁰⁾ (Modified GoogLeNet, 90 %), ⁽⁷⁴⁾ (DeepDR, 89 %), ⁽⁴⁴⁾ (AI general, 94 %), ⁽⁴⁶⁾ (Ensemble AI, 92 %), A76 (DART, 90 %)
95 - 100 %	⁽⁶⁵⁾ (VeriSee DR, 96 %), ⁽²⁹⁾ (AI general, 97 %), ⁽⁴¹⁾ (CNN, 76 %), ⁽²⁰⁾ (Modified GoogLeNet, 90 %), ⁽⁶¹⁾ (Phoebus Algorithm, 95 %), ⁽³⁶⁾ (ResNet-50, 90 %), ⁽³⁷⁾ (DLA, 92 %), ⁽⁸⁰⁾ (EyeCheckup AI, 93 %)

Table 4 shows the accuracy of different AI algorithms in DR detection. Examination of the data shows a varied distribution of accuracy among the different algorithms, with a wide range of percentages from less than 80 % to more than 95 %. Algorithms with an accuracy of less than 80 % could be considered less reliable for implementation in clinical practice, as there is a greater margin of error. However, it is important to recognize that some of these algorithms, such as ResNet 101 and CNN, approach the 80 % threshold, suggesting a potential for improvement with additional adjustments or training.

On the other hand, algorithms with accuracy in the 90 % to 95 % range, such as IDx-DR and EyeArt™, show remarkably strong performance in DR detection. These results support the feasibility of using these algorithms in clinical practice, as they offer consistently high accuracy that inspires confidence in their diagnostic ability. Algorithms with 95-100 % accuracy, such as VeriSee DR and general AI, are especially promising, as they offer high accuracy in DR detection. These results suggest that these algorithms have the potential to be highly reliable and accurate tools for healthcare professionals in detecting and diagnosing the disease.

Q4: What technical or implementation limitations do AI systems face in DR detection?

This question addresses the technical and implementation barriers that hinder the effective use of AI systems in DR detection. Identifying these limitations is critical to developing solutions that improve the integration and performance of these systems in real clinical settings.

Table 5. Limitations in DR detection using AI systems
Technical or implementation limitations	Artículos
Dependence on image quality	^{(40), (4), (41), (81), (64), (65), (42), (43), (14), (44), (45), (46), (47), (79), (3), (68), (69), (17), (49), (18), (70), (50), (83), (19), (20), (71), (51), (72), (21), (22), (73), (28), (53), (31), (54), (55), (85), (33), (74), (58), (60), (61), (76), (62), (37), (38), (77)}
Variability in interpretation of retinal features	^{(40), (78), (4), (41), (81), (63), (64), (65), (42), (43), (82), (45), (67), (47), (79), (3), (68), (69), (17), (49), (18), (1), (70), (50), (19), (20), (71), (8), (51), (72), (21), (22), (73), (25), (28), (53), (29), (31), (32), (54), (55), (85), (33), (74), (57), (35), (58), (60), (75), (61), (36), (62), (37), (38), (77)}
Implementation and maintenance costs	^{(78), (41), (81), (63), (64), (65), (42), (43), (14), (44), (45), (46), (47), (15), (79), (3), (68), (69), (17), (49), (18), (1), (70), (50), (83), (19), (20), (8), (51), (72), (21), (80), (22), (73), (25), (28), (53), (31), (32), (55), (85), (33), (74), (57), (35), (58), (60), (75), (61), (36), (76), (62), (37), (38), (77)}
Limited access to quality data	^{(40), (4), (81), (64), (66), (43), (14), (82), (44), (45), (46), (47), (15), (79), (3), (68), (69), (17), (49), (1), (70), (83), (19), (20), (8), (51), (72), (21), (80), (22), (24), (73), (25), (28), (53), (31), (32), (54), (55), (85), (33), (74), (57), (35), (58), (59), (60), (61), (76), (62), (37), (38), (77)}
Lack of integration with electronic medical record systems	^{(40), (78), (4), (81), (63), (64), (65), (42), (43), (14), (82), (45), (46), (47), (15), (79), (3), (68), (69), (17), (49), (18), (1), (70), (50), (83), (19), (20), (71), (8), (51), (72), (21), (80), (22), (73), (25), (28), (53), (31), (32), (54), (55), (85), (33), (74), (57), (35), (58), (60), (61), (76), (62), (37), (38), (77)}

According to table 5, major limitations include dependence on image quality, variability in the interpretation of retinal features, and implementation and maintenance costs. The quality of retinal images is crucial for diagnostic accuracy, and the lack of high-quality images may reduce the effectiveness of AI systems in DR detection. In addition, variability in the interpretation of retinal features can arise due to differences in algorithm training and human interpretation of image features, generating inconsistent and potentially erroneous results.

In addition, implementation and maintenance costs are an important factor to consider. The investment required to develop, implement, and maintain AI systems in clinical settings can be significant, limiting their adoption in areas with limited resources. Limited access to quality data and lack of integration with electronic medical record systems also emerge as challenges. Therefore, it is critical to develop solutions that address these limitations to improve the efficacy and adoption of artificial intelligence systems in diabetic retinopathy screening.

Q5: What ethical and legal challenges are posed by the use of AI algorithms in the diagnosis and treatment of DR?

This question explores the ethical and legal issues associated with the use of AI algorithms in the diagnosis and treatment of DR. It is important to consider these challenges to ensure that the implementation of AI is safe, fair, and compliant with legal regulations, thereby protecting patients and healthcare professionals.

Table 6. Challenges of using AI algorithms in the diagnosis and treatment of DR
Challenge	Articles
Privacy and security of patient data	^{(40), (4), (41), (81), (64), (65), (42), (43), (14), (44), (45), (46), (47), (79), (3), (68), (69), (17), (49), (18), (70), (50), (83), (19), (20), (71), (51), (72), (21), (22), (73), (28), (53), (31), (54), (55), (85), (33), (74), (58), (60), (61), (76), (62), (37), (38), (77)}
Liability for diagnostic errors	^{(40), (78), (4), (41), (81), (63), (64), (65), (42), (43), (82), (45), (67), (47), (79), (3), (68), (69), (17), (49), (18), (1), (70), (50), (19), (20), (71), (8), (51), (72), (21), (22), (73), (25), (28), (53), (29), (31), (32), (54), (55), (85), (33), (74), (57), (35), (58), (60), (75), (61), (36), (62), (37), (38), (77)}
Algorithmic bias in medical care	^{(40), (78), (41), (81), (63), (64), (65), (42), (43), (14), (82), (45), (67), (47), (15), (79), (48), (3), (68), (17), (49), (18), (70), (50), (19), (20), (84), (71), (8), (51), (52), (21), (80), (23), (73), (25), (28), (29), (30), (31), (54), (85), (33), (74), (57), (59), (60), (75), (86), (36), (62), (37), (38), (77)}
Informed consent and transparency in the use of AI	^{(78), (4), (41), (81), (64), (65), (42), (43), (14), (82), (45), (67), (46), (47), (15), (79), (48), (3), (68), (69), (17), (49), (18), (1), (70), (50), (83), (20), (84), (71), (8), (51), (72), (52), (21), (80), (23), (73), (25), (28), (29), (30), (31), (54), (85), (33), (74), (57), (35), (59), (75), (86), (36), (62), (38), (77)}
Regulations on the use of medical data for research and development purposes.	^{(40), (78), (4), (41), (81), (63), (64), (65), (42), (43), (14), (82), (44), (45), (67), (46), (47), (15), (79), (48), (3), (68), (69), (17), (49), (18), (1), (70), (50), (83), (19), (20), (84), (71), (8), (51), (72), (52), (21), (80), (23), (73), (25), (28), (29), (30), (31), (54), (85), (33), (74), (57), (58), (59), (75), (86), (61), (36), (62), (38), (39)}

According to table 6, the main concern revolves around the privacy and security of patient data, highlighting the need to protect sensitive information. In addition, liability in the case of diagnostic errors is a critical issue that requires a clear definition of the roles and responsibilities of healthcare professionals and algorithm developers. Algorithmic bias is also a significant concern, as a lack of diversity in training data can lead to erroneous medical decisions. Addressing this challenge is, therefore, essential to ensure equitable and fair medical care for all patients.

Transparency and informed consent are important elements in building a trusting relationship between patients and AI systems. Patients need to understand how their data is used and the implications of the use of AI algorithms in their medical care, highlighting the need for clear communication by healthcare professionals. Finally, regulations on the use of medical data for research and development purposes are necessary to ensure that the use of AI in healthcare is ethical and responsible. These regulations should establish clear standards to protect patient’s rights and ensure integrity and ethics in medical research.

CONCLUSIONS

A systematic review has shown that AI, specifically deep learning, is an effective technological alternative for DR detection. Various algorithms, such as ResNet, CNN, IDx-DR, and EyeArt™, have reported high accuracy, with percentages above 90 % in identifying DR features. Also, AI models consider factors such as microaneurysms, macular edema, changes in retinal vascularity, and optic disc size. These algorithms have proven effective in the early detection of DR, which facilitates timely treatment and improves clinical outcomes.

However, there are technical and implementation limitations that need to be addressed. Dependence on image quality, variability in the interpretation of retinal features, implementation and maintenance costs, limited access to quality data, and lack of integration with electronic medical record systems are some of the challenges identified. In addition, ethical and legal concerns arise, such as privacy and security of patient data, liability in case of diagnostic errors, algorithmic bias, informed consent and transparency in the use of AI, and regulations on the use of medical data for research and development purposes.

Although the systematic review has provided relevant results, it is important to recognize some limitations. The search was restricted to specific databases and to a particular range of years, which could have excluded relevant studies. Future research should focus on broadening the data sources over several decades, using additional databases, and selecting specific AI techniques for more detailed analysis.

REFERENCES

1. Cleland CR, Bascaran C, Makupa W, Shilio B, Sandi FA, Philippin H, et al. Artificial intelligence-supported diabetic retinopathy screening in Tanzania: rationale and design of a randomised controlled trial. BMJ Open [Internet]. 2024 Jan;14(1):e075055. Available from: https://bmjopen.bmj.com/content/bmjopen/14/1/e075055.full.pdf

2. Padhy S, Takkar B, Chawla R, Kumar A. Artificial intelligence in diabetic retinopathy: A natural step to the future. Indian J Ophthalmol [Internet]. 2019;67(7):1004. Available from: https://journals.lww.com/10.4103/ijo.IJO_1989_18

3. Huang XM, Yang BF, Zheng WL, Liu Q, Xiao F, Ouyang PW, et al. Cost-effectiveness of artificial intelligence screening for diabetic retinopathy in rural China. BMC Health Serv Res [Internet]. 2022 Feb 25;22(1):260. Available from: https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-022-07655-6

4. Grzybowski A, Brona P. Analysis and Comparison of Two Artificial Intelligence Diabetic Retinopathy Screening Algorithms in a Pilot Study: IDx-DR and Retinalyze. J Clin Med [Internet]. 2021 May 27;10(11):2352. Available from: https://www.mdpi.com/2077-0383/10/11/2352

5. Grauslund J. Diabetic retinopathy screening in the emerging era of artificial intelligence. Diabetologia [Internet]. 2022 Sep;65(9):1415–23. Available from: https://link.springer.com/10.1007/s00125-022-05727-0

6. Sheng B, Chen X, Li T, Ma T, Yang Y, Bi L, et al. An overview of artificial intelligence in diabetic retinopathy and other ocular diseases. Front Public Heal [Internet]. 2022 Oct 28;10. Available from: https://www.frontiersin.org/articles/10.3389/fpubh.2022.971943/full

7. Bidwai P, Gite S, Pahuja K, Kotecha K. A Systematic Literature Review on Diabetic Retinopathy Using an Artificial Intelligence Approach. Big Data Cogn Comput [Internet]. 2022 Dec 8;6(4):152. Available from: https://www.mdpi.com/2504-2289/6/4/152

8. Nakayama LF, Ribeiro LZ, Malerbi FK, Regatieri CVS. Ophthalmology and Artificial Intelligence: Present or Future? A Diabetic Retinopathy Screening Perspective of the Pursuit for Fairness. Front Ophthalmol [Internet]. 2022 May 10;2. Available from: https://www.frontiersin.org/articles/10.3389/fopht.2022.898181/full

9. Huemer J, Wagner SK, Sim DA. The Evolution of Diabetic Retinopathy Screening Programmes: A Chronology of Retinal Photography from 35 mm Slides to Artificial Intelligence. Clin Ophthalmol [Internet]. 2020 Jul;Volume 14:2021–35. Available from: https://www.dovepress.com/the-evolution-of-diabetic-retinopathy-screening-programmes-a-chronolog-peer-reviewed-article-OPTH

10. Rajesh AE, Davidson OQ, Lee CS, Lee AY. Artificial Intelligence and Diabetic Retinopathy: AI Framework, Prospective Studies, Head-to-head Validation, and Cost-effectiveness. Diabetes Care [Internet]. 2023 Oct 1;46(10):1728–39. Available from: https://diabetesjournals.org/care/article/46/10/1728/153626/Artificial-Intelligence-and-Diabetic-Retinopathy

11. Senapati A, Tripathy HK, Sharma V, Gandomi AH. Artificial intelligence for diabetic retinopathy detection: A systematic review. Informatics Med Unlocked [Internet]. 2024;45:101445. Available from: https://linkinghub.elsevier.com/retrieve/pii/S2352914824000017

12. Li S, Zhao R, Zou H. Artificial intelligence for diabetic retinopathy. Chin Med J (Engl) [Internet]. 2022 Feb 5;135(3):253–60. Available from: https://journals.lww.com/10.1097/CM9.0000000000001816

13. Kitchenham B, Charters S. Guidelines for performing Systematic Literature Reviews in Software Engineering. Keele Univ y Univ Durham [Internet]. 2007; Available from: https://legacyfileshare.elsevier.com/promis_misc/525444systematicreviewsguide.pdf

14. Pei X, Yao X, Yang Y, Zhang H, Xia M, Huang R, et al. Efficacy of artificial intelligence-based screening for diabetic retinopathy in type 2 diabetes mellitus patients. Diabetes Res Clin Pract [Internet]. 2022 Feb;184(109190):109190. Available from: https://www.diabetesresearchclinicalpractice.com/article/S0168-8227(22)00002-X/fulltext

15. Lee AY, Yanagihara RT, Lee CS, Blazes M, Jung HC, Chee YE, et al. Multicenter, Head-to-Head, Real-World Validation Study of Seven Automated Artificial Intelligence Diabetic Retinopathy Screening Systems. Diabetes Care [Internet]. 2021 May 1;44(5):1168–75. Available from: https://diabetesjournals.org/care/article/44/5/1168/138752/Multicenter-Head-to-Head-Real-World-Validation

16. Gomez Rossi J, Rojas-Perilla N, Krois J, Schwendicke F. Cost-effectiveness of Artificial Intelligence as a Decision-Support System Applied to the Detection and Grading of Melanoma, Dental Caries, and Diabetic Retinopathy. JAMA Netw Open [Internet]. 2022 Mar 15;5(3):e220269.

17. Quellec G, Al Hajj H, Lamard M, Conze PH, Massin P, Cochener B. ExplAIn: Explanatory artificial intelligence for diabetic retinopathy diagnosis. Med Image Anal [Internet]. 2021 Aug;72:102118. Available from: https://linkinghub.elsevier.com/retrieve/pii/S136184152100164X

18. Wolf RM, Channa R, Liu TYA, Zehra A, Bromberger L, Patel D, et al. Autonomous artificial intelligence increases screening and follow-up for diabetic retinopathy in youth: the ACCESS randomized control trial. Nat Commun [Internet]. 2024 Jan 11;15(1):421. Available from: https://www.nature.com/articles/s41467-023-44676-z

19. Salma A, Bustamam A, Yudantha A, Victor A, Mangunwardoyo W. Artificial Intelligence Approach in Multiclass Diabetic Retinopathy Detection Using Convolutional Neural Network and Attention Mechanism. Int J Adv Soft Comput its Appl [Internet]. 2021 Dec 30;13(3):101–14. Available from: http://ijasca.zuj.edu.jo/PapersUploaded/2021.3.8.pdf

20. Takahashi H, Tampo H, Arai Y, Inoue Y, Kawashima H. Applying artificial intelligence to disease staging: Deep learning for improved staging of diabetic retinopathy. Mori K, editor. PLoS One [Internet]. 2017 Jun 22;12(6):e0179790. Available from: https://dx.plos.org/10.1371/journal.pone.0179790

21. Rajamani S, Sasikala S. Artificial Intelligence Approach for Diabetic Retinopathy Severity Detection. Informatica [Internet]. 2023 Jan 5;46(8). Available from: https://www.informatica.si/index.php/informatica/article/view/4425

22. Qian X, Jingying H, Xian S, Yuqing Z, Lili W, Baorui C, et al. The effectiveness of artificial intelligence-based automated grading and training system in education of manual detection of diabetic retinopathy. Front Public Heal [Internet]. 2022 Nov 7;10. Available from: https://www.frontiersin.org/articles/10.3389/fpubh.2022.1025271/full

23. Hao Z, Cui S, Zhu Y, Shao H, Huang X, Jiang X, et al. Application of non-mydriatic fundus examination and artificial intelligence to promote the screening of diabetic retinopathy in the endocrine clinic: an observational study of T2DM patients in Tianjin, China. Ther Adv Chronic Dis [Internet]. 2020 Jan 10;11:204062232094241. Available from: http://journals.sagepub.com/doi/10.1177/2040622320942415

24. Liu YF, Fei FQ, Chen NM, Zhu ZT, Fei XZ. Research progress in artificial intelligence assisted diabetic retinopathy diagnosis. Int J Ophthalmol [Internet]. 2023 Sep 18;16(9):1395–405. Available from: http://ies.ijo.cn/gjyken/ch/reader/view_abstract.aspx?file_no=20230905&flag=1

25. Cuadros J. The Real-World Impact of Artificial Intelligence on Diabetic Retinopathy Screening in Primary Care. J Diabetes Sci Technol [Internet]. 2021 May 24;15(3):664–5. Available from: http://journals.sagepub.com/doi/10.1177/1932296820914287

26. Han R, Yu W, Chen H, Chen Y. Using artificial intelligence reading label system in diabetic retinopathy grading training of junior ophthalmology residents and medical students. BMC Med Educ [Internet]. 2022 Dec 9;22(1):258. Available from: https://bmcmededuc.biomedcentral.com/articles/10.1186/s12909-022-03272-3

27. Li H, Li G, Li N, Liu C, Yuan Z, Gao Q, et al. Cost-effectiveness analysis of artificial intelligence-based diabetic retinopathy screening in rural China based on the Markov model. Raafat KA, editor. PLoS One [Internet]. 2023 Nov 16;18(11):e0291390. Available from: https://dx.plos.org/10.1371/journal.pone.0291390

28. Xie Y, Gunasekeran D V., Balaskas K, Keane PA, Sim DA, Bachmann LM, et al. Health Economic and Safety Considerations for Artificial Intelligence Applications in Diabetic Retinopathy Screening. Transl Vis Sci Technol [Internet]. 2020 Apr 13;9(2):22. Available from: https://tvst.arvojournals.org/article.aspx?articleid=2764675

29. Barakat AA, Mobarak O, Javaid HA, Awad MR, Hamweyah K, Ouban A, et al. The application of artificial intelligence in diabetic retinopathy screening: a Saudi Arabian perspective. Front Med [Internet]. 2023 Nov 22;10. Available from: https://www.frontiersin.org/articles/10.3389/fmed.2023.1303300/full

30. Hao Z, Xu R, Huang X, Ren X, Li H, Shao H. Application and observation of artificial intelligence in clinical practice of fundus screening for diabetic retinopathy with non-mydriatic fundus photography: a retrospective observational study of T2DM patients in Tianjin, China. Ther Adv Chronic Dis [Internet]. 2022 Jan 19;13:204062232210973. Available from: http://journals.sagepub.com/doi/10.1177/20406223221097335

31. Ipp E, Liljenquist D, Bode B, Shah VN, Silverstein S, Regillo CD, et al. Pivotal Evaluation of an Artificial Intelligence System for Autonomous Detection of Referrable and Vision-Threatening Diabetic Retinopathy. JAMA Netw Open [Internet]. 2021 Nov 15;4(11):e2134254. Available from: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2786132

32. Yagin FH, Yasar S, Gormez Y, Yagin B, Pinar A, Alkhateeb A, et al. Explainable Artificial Intelligence Paves the Way in Precision Diagnostics and Biomarker Discovery for the Subclass of Diabetic Retinopathy in Type 2 Diabetics. Metabolites [Internet]. 2023 Dec 18;13(12):1204. Available from: https://www.mdpi.com/2218-1989/13/12/1204

33. Komatsu K, Sano K, Fukai K, Nakagawa R, Nakagawa T, Tatemichi M, et al. Associated factors of diabetic retinopathy by artificial intelligence evaluation of fundus images in Japan. Sci Rep [Internet]. 2023 Nov 13;13(1):19742. Available from: https://www.nature.com/articles/s41598-023-47270-x

34. Hu W, Joseph S, Li R, Woods E, Sun J, Shen M, et al. Population impact and cost-effectiveness of artificial intelligence-based diabetic retinopathy screening in people living with diabetes in Australia: a cost effectiveness analysis. eClinicalMedicine [Internet]. 2024 Jan;67:102387. Available from: https://linkinghub.elsevier.com/retrieve/pii/S2589537023005643

35. Scanzera AC, Beversluis C, Potharazu A V., Bai P, Leifer A, Cole E, et al. Planning an artificial intelligence diabetic retinopathy screening program: a human-centered design approach. Front Med [Internet]. 2023 Jul 7;10. Available from: https://www.frontiersin.org/articles/10.3389/fmed.2023.1198228/full

36. Ursin F, Timmermann C, Orzechowski M, Steger F. Diagnosing Diabetic Retinopathy With Artificial Intelligence: What Information Should Be Included to Ensure Ethical Informed Consent? Front Med [Internet]. 2021 Jul 21;8. Available from: https://www.frontiersin.org/articles/10.3389/fmed.2021.695217/full

37. Wong TY, Sabanayagam C. Strategies to Tackle the Global Burden of Diabetic Retinopathy: From Epidemiology to Artificial Intelligence. Ophthalmologica [Internet]. 2020;243(1):9–20. Available from: https://karger.com/OPH/article/doi/10.1159/000502387

38. Shahsuvaryan ML. Is it time to consider teleophthalmology as a game-changer in the management of diabetic retinopathy? Rev Bras Oftalmol [Internet]. 2023 Sep 1;82. Available from: https://www.rbojournal.org/en/article/is-it-time-to-consider-teleophthalmology-as-a-game-changer-in-the-management-of-diabetic-retinopathy/

39. Zhelev Z, Peters J, Rogers M, Allen M, Kijauskaite G, Seedat F, et al. Test accuracy of artificial intelligence-based grading of fundus images in diabetic retinopathy screening: A systematic review. J Med Screen [Internet]. 2023 Sep 9;30(3):97–112. Available from: http://journals.sagepub.com/doi/10.1177/09691413221144382

40. Liu R, Li Q, Xu F, Wang S, He J, Cao Y, et al. Application of artificial intelligence-based dual-modality analysis combining fundus photography and optical coherence tomography in diabetic retinopathy screening in a community hospital. Biomed Eng Online [Internet]. 2022 Dec 20;21(1):47. Available from: https://biomedical-engineering-online.biomedcentral.com/articles/10.1186/s12938-022-01018-2#Sec1

41. Lin S, Ma Y, Xu Y, Lu L, He J, Zhu J, et al. Artificial Intelligence in Community-Based Diabetic Retinopathy Telemedicine Screening in Urban China: Cost-effectiveness and Cost-Utility Analyses With Real-world Data. JMIR Public Heal Surveill [Internet]. 2023 Feb 23;9(9):e41624. Available from: https://pubmed.ncbi.nlm.nih.gov/36821353/

42. Wolf RM, Liu TYA, Thomas C, Prichett L, Zimmer-Galler I, Smith K, et al. The SEE Study: Safety, Efficacy, and Equity of Implementing Autonomous Artificial Intelligence for Diagnosing Diabetic Retinopathy in Youth. Diabetes Care [Internet]. 2021 Mar 1;44(3):781–7. Available from: https://diabetesjournals.org/care/article/44/3/781/138584/The-SEE-Study-Safety-Efficacy-and-Equity-of

43. Scheetz J, Koca D, McGuinness M, Holloway E, Tan Z, Zhu Z, et al. Real-world artificial intelligence-based opportunistic screening for diabetic retinopathy in endocrinology and indigenous healthcare settings in Australia. Sci Rep [Internet]. 2021 Aug 4;11(1):15808. Available from: https://www.nature.com/articles/s41598-021-94178-5

44. He J, Cao T, Xu F, Wang S, Tao H, Wu T, et al. Artificial intelligence-based screening for diabetic retinopathy at community hospital. Eye [Internet]. 2020 Mar 27;34(3):572–6. Available from: https://www.nature.com/articles/s41433-019-0562-4#citeas

45. Shah A, Clarida W, Amelon R, Hernaez-Ortega MC, Navea A, Morales-Olivas J, et al. Validation of Automated Screening for Referable Diabetic Retinopathy With an Autonomous Diagnostic Artificial Intelligence System in a Spanish Population. J Diabetes Sci Technol [Internet]. 2021 May 16;15(3):655–63. Available from: http://journals.sagepub.com/doi/10.1177/1932296820906212

46. Bellemo V, Lim ZW, Lim G, Nguyen QD, Xie Y, Yip MYT, et al. Artificial intelligence using deep learning to screen for referable and vision-threatening diabetic retinopathy in Africa: a clinical validation study. Lancet Digit Heal [Internet]. 2019 May;1(1):e35–44. Available from: https://www.thelancet.com/journals/landig/article/PIIS2589-7500(19)30004-4/fulltext#

47. Dong X, Du S, Zheng W, Cai C, Liu H, Zou J. Evaluation of an Artificial Intelligence System for the Detection of Diabetic Retinopathy in Chinese Community Healthcare Centers. Front Med [Internet]. 2022 Apr 11;9:883462. Available from: https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2022.883462/full

48. Wang H, Meng X, Tang Q, Hao Y, Luo Y, Li J. Development and Application of a Standardized Testset for an Artificial Intelligence Medical Device Intended for the Computer‐Aided Diagnosis of Diabetic Retinopathy. Xu Y, editor. J Healthc Eng [Internet]. 2023 Jan 8;2023(1). Available from: https://onlinelibrary.wiley.com/doi/epdf/10.1155/2023/7139560

49. Xie Q, Liu Y, Huang H, Hong B, Wang J, Han H, et al. An innovative method for screening and evaluating the degree of diabetic retinopathy and drug treatment based on artificial intelligence algorithms. Pharmacol Res [Internet]. 2020 Sep;159:104986. Available from: https://linkinghub.elsevier.com/retrieve/pii/S1043661820312949

50. Obayya M, Nemri N, Nour MK, Al Duhayyim M, Mohsen H, Rizwanullah M, et al. Explainable Artificial Intelligence Enabled TeleOphthalmology for Diabetic Retinopathy Grading and Classification. Appl Sci [Internet]. 2022 Aug 31;12(17):8749. Available from: https://www.mdpi.com/2076-3417/12/17/8749

51. Rajalakshmi R, Subashini R, Anjana RM, Mohan V. Automated diabetic retinopathy detection in smartphone-based fundus photography using artificial intelligence. Eye [Internet]. 2018 Jun 9;32(6):1138–44. Available from: https://www.nature.com/articles/s41433-018-0064-9

52. Doğan ME, Bilgin AB, Sari R, Bulut M, Akar Y, Aydemir M. Head to head comparison of diagnostic performance of three non-mydriatic cameras for diabetic retinopathy screening with artificial intelligence. Eye [Internet]. 2024 Jun 11;38(9):1694–701. Available from: https://www.nature.com/articles/s41433-024-03000-9

53. Lupidi M, Danieli L, Fruttini D, Nicolai M, Lassandro N, Chhablani J, et al. Artificial intelligence in diabetic retinopathy screening: clinical assessment using handheld fundus camera in a real-life setting. Acta Diabetol [Internet]. 2023 May 8;60(8):1083–8. Available from: https://link.springer.com/10.1007/s00592-023-02104-0

54. Korn Malerbi F, Barreto Melo G. Feasibility of screening for diabetic retinopathy using artificial intelligence, Brazil. Bull World Health Organ [Internet]. 2022 Oct 1;100(10):643–7. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9511671/pdf/BLT.22.288580.pdf

55. Vaghefi E, Yang S, Xie L, Hill S, Schmiedel O, Murphy R, et al. THEIATM development, and testing of artificial intelligence‐based primary triage of diabetic retinopathy screening images in New Zealand. Diabet Med [Internet]. 2021 Apr 27;38(4). Available from: https://onlinelibrary.wiley.com/doi/10.1111/dme.14386

56. Penha FM, Priotto BM, Hennig F, Przysiezny B, Wiethorn BA, Orsi J, et al. Single retinal image for diabetic retinopathy screening: performance of a handheld device with embedded artificial intelligence. Int J Retin Vitr [Internet]. 2023 Jul 10;9(1):41. Available from: https://journalretinavitreous.biomedcentral.com/articles/10.1186/s40942-023-00477-6

57. Natarajan S, Jain A, Krishnan R, Rogye A, Sivaprasad S. Diagnostic Accuracy of Community-Based Diabetic Retinopathy Screening With an Offline Artificial Intelligence System on a Smartphone. JAMA Ophthalmol [Internet]. 2019 Oct 1;137(10):1182. Available from: https://jamanetwork.com/journals/jamaophthalmology/fullarticle/2747315

58. Kemp O, Bascaran C, Cartwright E, McQuillan L, Matthew N, Shillingford-Ricketts H, et al. Real-world evaluation of smartphone-based artificial intelligence to screen for diabetic retinopathy in Dominica: a clinical validation study. BMJ Open Ophthalmol [Internet]. 2023 Dec 21;8(1):e001491. Available from: https://bmjophth.bmj.com/lookup/doi/10.1136/bmjophth-2023-001491

59. Surya J, Garima, Pandy N, Hyungtaek Rim T, Lee G, Priya MS, et al. Efficacy of deep learning-based artificial intelligence models in screening and referring patients with diabetic retinopathy and glaucoma. Indian J Ophthalmol [Internet]. 2023 Aug;71(8):3039–45. Available from: https://journals.lww.com/10.4103/IJO.IJO_11_23

60. Grzybowski A, Rao DP, Brona P, Negiloni K, Krzywicki T, Savoy FM. Diagnostic Accuracy of Automated Diabetic Retinopathy Image Assessment Softwares: IDx-DR and Medios Artificial Intelligence. Ophthalmic Res [Internet]. 2023;66(1):1286–92. Available from: https://karger.com/ORE/article/doi/10.1159/000534098

61. Ruan S, Hu WT, Jia HX, Wang SS, Song ML, Shen MX, et al. A new handheld fundus camera combined with visual artificial intelligence facilitates diabetic retinopathy screening. Int J Ophthalmol [Internet]. 2022 Apr 18;15(4):620–7. Available from: http://ies.ijo.cn/gjyken/ch/reader/view_abstract.aspx?file_no=20220416&flag=1

62. Shah P, Mishra D, Shanmugam M, Doshi B, Jayaraj H, Ramanjulu R. Validation of Deep Convolutional Neural Network-based algorithm for detection of diabetic retinopathy – Artificial intelligence versus clinician for screening. Indian J Ophthalmol [Internet]. 2020;68(2):398. Available from: https://journals.lww.com/ijo/Fulltext/2020/68020/Validation_of_Deep_Convolutional_Neural.32.aspx

63. Dow ER, Chen KM, Zhao C, Knapp A, Phadke A, Weng K, et al. Artificial Intelligence Improves Patient Follow-Up in a Diabetic Retinopathy Screening Program. Clin Ophthalmol [Internet]. 2023 Nov;Volume 17:3323–30. Available from: https://www.dovepress.com/artificial-intelligence-improves-patient-follow-up-in-a-diabetic-retin-peer-reviewed-fulltext-article-OPTH

64. Kanagasingam Y, Xiao D, Vignarajan J, Preetham A, Tay-Kearney ML, Mehrotra A. Evaluation of Artificial Intelligence–Based Grading of Diabetic Retinopathy in Primary Care. JAMA Netw Open [Internet]. 2018 Sep 28;1(5):e182665.

65. Tsai MJ, Hsieh YT, Tsai CH, Chen M, Hsieh AT, Tsai CW, et al. Cross-Camera External Validation for Artificial Intelligence Software in Diagnosis of Diabetic Retinopathy. Li T, editor. J Diabetes Res [Internet]. 2022 Mar 9;2022(1):1–5. Available from: https://www.hindawi.com/journals/jdr/2022/5779276/

66. Zhang Y, Shi J, Peng Y, Zhao Z, Zheng Q, Wang Z, et al. Artificial intelligence-enabled screening for diabetic retinopathy: a real-world, multicenter and prospective study. BMJ Open Diabetes Res Care [Internet]. 2020 Oct 21;8(1):e001596. Available from: https://drc.bmj.com/content/8/1/e001596

67. Wewetzer L, Held LA, Goetz K, Steinhäuser J. Determinants of the implementation of artificial intelligence-based screening for diabetic retinopathy—a cross-sectional study with general practitioners in Germany. Digit Heal [Internet]. 2023 Jan 30;9:205520762311766. Available from: https://journals.sagepub.com/doi/10.1177/20552076231176644

68. Al-hazaimeh OM, Abu-Ein A, Tahat N, Al-Smadi M, Al-Nawashi M. Combining Artificial Intelligence and Image Processing for Diagnosing Diabetic Retinopathy in Retinal Fundus Images. Int J Online Biomed Eng [Internet]. 2022 Oct 19;18(13):131–51. Available from: https://online-journals.org/index.php/i-joe/article/view/33985/12091

69. Katada Y, Ozawa N, Masayoshi K, Ofuji Y, Tsubota K, Kurihara T. Automatic screening for diabetic retinopathy in interracial fundus images using artificial intelligence. Intell Med [Internet]. 2020 Dec;3–4(100024):100024. Available from: https://online-journals.org/index.php/i-joe/article/view/33985/12091

70. Vidal-Alaball J, Royo Fibla D, Zapata MA, Marin-Gomez FX, Solans Fernandez O. Artificial Intelligence for the Detection of Diabetic Retinopathy in Primary Care: Protocol for Algorithm Development. JMIR Res Protoc [Internet]. 2019 Feb 1;8(2):e12539. Available from: http://www.researchprotocols.org/2019/2/e12539/

71. Crane AB, Choudhry HS, Dastjerdi MH. Effect of simulated cataract on the accuracy of artificial intelligence in detecting diabetic retinopathy in color fundus photos. Indian J Ophthalmol [Internet]. 2024 Jan;72(Suppl 1):S42–5. Available from: https://journals.lww.com/10.4103/IJO.IJO_1163_23

72. Gopalakrishnan N, Joshi A, Chhablani J, Yadav NK, Reddy NG, Rani PK, et al. Recommendations for initial diabetic retinopathy screening of diabetic patients using large language model-based artificial intelligence in real-life case scenarios. Int J Retin Vitr [Internet]. 2024 Jan 24;10(1):11. Available from: https://journalretinavitreous.biomedcentral.com/articles/10.1186/s40942-024-00533-9

73. Xie Y, Nguyen QD, Hamzah H, Lim G, Bellemo V, Gunasekeran D V, et al. Artificial intelligence for teleophthalmology-based diabetic retinopathy screening in a national programme: an economic analysis modelling study. Lancet Digit Heal [Internet]. 2020 May;2(5):e240–9. Available from: https://linkinghub.elsevier.com/retrieve/pii/S2589750020300601

74. Tomić M, Vrabec R, Hendelja Đ, Kolarić V, Bulum T, Rahelić D. Diagnostic Accuracy of Hand-Held Fundus Camera and Artificial Intelligence in Diabetic Retinopathy Screening. Biomedicines [Internet]. 2023 Dec 22;12(1):34. Available from: https://www.mdpi.com/2227-9059/12/1/34

75. Sosale B, Sosale A, Murthy H, Sengupta S, Naveenam M. Medios– An offline, smartphone-based artificial intelligence algorithm for the diagnosis of diabetic retinopathy. Indian J Ophthalmol [Internet]. 2020;68(2):391. Available from: https://journals.lww.com/ijo/Fulltext/2020/68020/Medios__An_offline,_smartphone_based_artificial.30.aspx

76. Keel S, Lee PY, Scheetz J, Li Z, Kotowicz MA, MacIsaac RJ, et al. Feasibility and patient acceptability of a novel artificial intelligence-based screening model for diabetic retinopathy at endocrinology outpatient services: a pilot study. Sci Rep [Internet]. 2018 Mar 12;8(1):4330. Available from: https://www.nature.com/articles/s41598-018-22612-2

77. Ibáñez-bruron M c., Cruzat A, Órdenes-Cavieres G, Coria M. Exactitud de tamizaje de retinopatía diabética: inteligencia artificial versus tecnólogos médicos entrenados. Rev Med Chil [Internet]. 2021 Apr;149(4):493–500. Available from: http://www.scielo.cl/scielo.php?script=sci_arttext&pid=S0034-98872021000400493&lng=en&nrm=iso&tlng=en

78. Wang R, Zuo G, Li K, Li W, Xuan Z, Han Y, et al. Systematic bibliometric and visualized analysis of research hotspots and trends on the application of artificial intelligence in diabetic retinopathy. Front Endocrinol (Lausanne) [Internet]. 2022 Oct 31;13. Available from: https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2022.1036426/full

79. Morya AK, Gowdar J, Kaushal A, Makwana N, Biswas S, Raj P, et al. Evaluating the Viability of a Smartphone-Based Annotation Tool for Faster and Accurate Image Labelling for Artificial Intelligence in Diabetic Retinopathy. Clin Ophthalmol [Internet]. 2021 Mar;Volume 15:1023–39. Available from: https://www.dovepress.com/evaluating-the-viability-of-a-smartphone-based-annotation-tool-for-fas-peer-reviewed-article-OPTH

80. Jain A, Krishnan R, Rogye A, Natarajan S. Use of offline artificial intelligence in a smartphone-based fundus camera for community screening of diabetic retinopathy. Indian J Ophthalmol [Internet]. 2021;69(11):3150. Available from: https://journals.lww.com/ijo/Fulltext/2021/11000/Use_of_offline_artificial_intelligence_in_a.40.aspx

81. Wang Y, Liu C, Hu W, Luo L, Shi D, Zhang J, et al. Economic evaluation for medical artificial intelligence: accuracy vs. cost-effectiveness in a diabetic retinopathy screening case. npj Digit Med [Internet]. 2024 Feb 21;7(1):43. Available from: https://www.nature.com/articles/s41746-024-01032-9#citeas

82. Arenas-Cavalli JT, Abarca I, Rojas-Contreras M, Bernuy F, Donoso R. Clinical validation of an artificial intelligence-based diabetic retinopathy screening tool for a national health system. Eye [Internet]. 2022 Jan 11;36(1):78–85. Available from: https://www.nature.com/articles/s41433-020-01366-0

83. Held LA, Wewetzer L, Steinhäuser J. Determinants of the implementation of an artificial intelligence-supported device for the screening of diabetic retinopathy in primary care – a qualitative study. Health Informatics J [Internet]. 2022 Jul 3;28(3):146045822211128. Available from: http://journals.sagepub.com/doi/10.1177/14604582221112816

84. Cao K. Artificial intelligence on diabetic retinopathy diagnosis: an automatic classification method based on grey level co-occurrence matrix and naive Bayesian model. Int J Ophthalmol [Internet]. 2019 Jul 18;12(7):1158–62. Available from: http://www.ijo.cn/gjyken/ch/reader/view_abstract.aspx?file_no=20190717&flag=1

85. Sosale B, Aravind SR, Murthy H, Narayana S, Sharma U, Gowda SG V, et al. Simple, Mobile-based Artificial Intelligence Algo r ithm in the detection of Diabetic Retinopathy (SMART) study. BMJ Open Diabetes Res Care [Internet]. 2020 Jan 28;8(1):e000892. Available from: https://drc.bmj.com/lookup/doi/10.1136/bmjdrc-2019-000892

86. Malerbi FK, Andrade RE, Morales PH, Stuchi JA, Lencione D, de Paulo JV, et al. Diabetic Retinopathy Screening Using Artificial Intelligence and Handheld Smartphone-Based Retinal Camera. J Diabetes Sci Technol [Internet]. 2022 May 12;16(3):716–23. Available from: http://journals.sagepub.com/doi/10.1177/1932296820985567

FINANCING

The authors did not receive funding for the development of this research.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest.

AUTHORSHIP CONTRIBUTION

Conceptualization: Marck Julca and Richard Injante.

Data curation: Richard Injante.

Formal analysis: Marck A Julca and Richard Injante.

Research: Marck Julca and Richard Injante.

Methodology: Richard Injante.

Project management: Marck Julca and Richard Injante.

Resources: Richard Injante.

Supervision: Richard Injante.

Validation: Richard Injante.

Visualization: Marck Julca.

Writing - original draft: Marck Julca and Richard Injante.

Writing - revision and editing: Marck Julca and Richard Injante.