**The detailed document has been attached. For the complete version of the submission is recommended to consider the attachment pdf, which included images and tables.
THE PROBLEM TO TACKLE
During the current pandemic, the healthcare flows of millions of people affected by chronic diseases have been dramatically altered. Moreover, most dangerous diseases that can be affected by coronavirus contagion are cancer, diabetes and hypertension. Social distancing during COVID spread is a function of space, time: the need for healthcare is thus increasing, but our capacity of providing care and assistance is decreasing. Especially in the case of over 50 diabetes and hypertension patients, going to and waiting in the hospitals represent a higher contagion risk and subsequent spread. Congestion in the hospitals due to the simple prevention and follow-up of those categories is another aspect, overall due to increasing chronic states by patients that cannot receive the same care as in standard times. This finally brings the additional problem of severe difficulties for those same patients categories in monitoring their clinical conditions. In order to have an idea about the impact that the COVID pandemic is having on this scenario, we try to provide some statistics about social diseases. Worldwide, 400 million people are affected by diabetes, 927 million of people suffer from hypertension, 196 million and 76 million of people are affected by Age-Related Macular Degeneration and Glaucoma respectively. In the cases of the mentioned diseases, each patient must perform a follow-up eye examination every 1/6 months or annually, besides urgent visits in case of ongoing complications. One of the most significant examination in the context of these pathologies is provided by the ocular fundus images. These last can be acquired by means of a tool which could be located at public or private medical clinics. Moreover, many factors such as quality of the image, technicians skills and the optic lenses, make the telemedicine distribution of this exam very difficult. The process to get the proper assessment through the ocular fundus examination, in many countries, involves several interactions between the patient and, for instance, general practitioners, ophthalmologists, cardiologists, diabetologists or other specialists. The complexity of this process leads patients to reach several different clinics. Summarizing, about 1 billion 700 million people worldwide, must interact with at least 3 doctors for each examination, the result is a flow of mobilization of 6.3 billion interactions. A conservative idea about the numbers in Europe is 120M people, 500M interactions.
THE PROBLEM ANALYSIS
**Tables and images are reported into the attached pdf detailed document
We studied the flows related to monitoring clinical conditions for those patients, trying to measure the impacts they have over hospital congestion, contagion risks and health system economy. According to various researches and studies, we estimated the people affected by social diseases such as diabetes, hypertension, glaucoma, AMD, all over Europe, in Italy and, as we’ll see later, in a specific focus for the south italian Regione Calabria;
As we know, those people are especially exposed to the risk of complications caused by possible infection of Covid-19. This risk increases considering the amount of check-ups those patients have to do, most of the times going to hospitals or private clinics, which are also the most exposed places to pandemic and contagious infections.
To quantize those flows, we studied the case of an Italian region, Calabria, thinking that it could be a good starting point for our experiment, because of the critical issues connected to the health care system congestion and lack of accessibility.
We estimated how many people affected by the three main social diseases analysed before (diabetes, hypertension, glaucoma) are in Calabria; after that, we identified and geolocalized the hospitals and public infrastructures equipped to make follow-up eye examinations, 9 in total.
Simply looking at the maps, we could already realise the amount of patients already cut out (in terms of accessibility) from the health infrastructures.
Our analysis says that of the over 1 million people affected by those social diseases, 800.000 of them are far from those structures, meaning that they have to move out from their municipality to check their conditions twice a year or more. On average those patients are 30 kilometers or more distant from their hospitals, an amount of space we can’t afford especially during a pandemic emergency.
In the specific, over 90 thousands of diabetes patients, 78% of all the diabetics in Calabria, are living in urban areas with no access to the health technologies they need; even more massive flows are connected to the hypertensives citizens, 665 thousands don’t have access to health care into the boundaries of their cities.
As the last step of our impact analysis we estimated how much it costs to the national health care system to follow up examination of those diseases. We calculated the costs of the only ocular fundus exam and the frequency over a one year sample. In a context where the emergency has turned social health systems upside down, it is mandatory to redesign the availability of hardware and software technologies for both patients and medical personnel and invest in the prevention of the same pathologies. Citizens may be lazy about their examination or simply unaware of the beginning of some troubling condition. In this emergency state and considering the high risks they face, we see these problems in the pandemic panorama of great importance: we cannot ignore these individuals just because our healthcare system is congested.
Acronyms
- Diabetic retinopathy (DR)
- Age related macular degeneration (AMD)
- Ocular Fundus Photograph (OCF)
- Artificial Intelligence (AI)
Terminology
The system : the whole solution (processes, concept and technology) Social diseases : Diabetes, Glaucoma, Hypertension, AMD Device or Lens : physical tool used along with the smartphone, in order to capture the OCF Primary prevention : large scale health protection to avoid the spread of pathologies Secondary prevention: Screening of health conditions of risk factor related subjects Screening : pre pathology control in order to keep under control the patient’s risks factors Diagnosis : once the signs and symptoms are present, the diagnosis is the definition of the pathology which causes those symptoms Follow UP : continuous check of patients with diagnosed pathologies conditions in order to avoid further degenerations of the pathology
THE SOLUTION
The PREVYNET project aims at simplifying the availability of healthcare and prevention, during the COVID-19 crisis as well as for the future to come.
INTUITION
All of the four social diseases we analysed, turn out to have a common denominator: the examination!
Preventing a huge number of fragile people from moving to execute routine examinations, could bring a massive impact on their health during this pandemic emergency, protecting them by the contagion.
We thought about an examination process to tackle the huge problem of social diseases and their side effects, with the purpose of considerably reducing the patients' flows towards physical clinics and clinicians.
We want to leverage the following results obtained by research and reported by literature: AI capability of detecting social diseases (Glaucoma, Diabetes, Hypertension, Age-Related Macular Degeneration) AI capability in images quality assessments Ocular fundus images acquired by independent lenses and smartphones
WHAT IT IS
PREVYNET is a service that consists of a set of tools defined as follow:
PREVYAPP : patient smartphone application to acquire OCF, and send it to the physician
PREVYMED : Platform Medical Interface for physicians to ease the diagnosis
PREVYAGENT : qualified technician that support patient examination when the latter can not acquire the OCF autonomously, or with the help of a caregiver
PREVYPOINT : strategically located place where PREVIAGENT helps patients with the OCF examination
PREVYPlatform : software cloud infrastructure in charge of managing data streams
PREVYBASE: global European - privacy compliant - research enabler database
HOW IT WORKS
Patients will be able to acquire their ocular fundus images by means of a hardware tool, such as a lens, and a smartphone. Images and clinical data will be sent to the PREVYPlatform which is in charge of: Delivering the screening to clinicians, enabling remote periodical control “from home”! Collecting data on to the PREVYBASE Training an AI system for diseases detection, and supporting clinicians diagnosis
Let’s imagine: Jimmy is 55 years old. It has diabetes and he used to attend an ocular fundus examination once every six months. Jimmy decides to use the PREVYAPP to capture and send the ocular fundus image to his Ophthalmologist. The physician, supported by an artificial intelligence can make his diagnosis thanks to the prevynet PREVIMED Interface, WITHOUT ANY CONTACT with the patient.
Teresa is 72 years old. She is affected by hypertension and she used to do periodical control by means of the ocular fundus examination. Due to the pandemic, she has no chance of getting a reservation for the examination in any of her trusted clinics! Teresa decides to use the PREVYNET service, but she is not so confident in using the smartphone camera. So she looks for the nearest PREVYPOINT. She finds one nearby her house and she gets there in 5 minutes by walking. Teresa is provided with the proper support in acquiring a OCF by a PREVYAGENT. Teresa comes back home waiting for the diagnosis made by her trusted doctor, using the PREVYMED interface with the AI support. Teresa gets her follow up examination and diagnosis, AVOIDING clinics thus a HIGH-RISK environment!
HOW WE BUILD IT
Hardware
Something more on PREVYPOINT The concept behind the prevypoint is providing patients of a simple and accessible system to receive the proper assistance; reducing significantly the physical flow of the patient to Covid-risk spots. To do it, we need to distribute fundus photography technology in strategically located points in the neighborhood.
There are different ways to do it. One option is to use existing hardware such as mobile phones, equipped with a lens mounted on a device.
Smartphone fundus photography is a simple technique to obtain ocular fundus pictures using a smartphone camera and a conventional handheld indirect ophthalmoscopy lens. The main advantage of this technique is the widespread availability of smartphones that allows documentation of macula and optic nerve changes in many settings that was not previously possible.
There are several solutions on the market:
- Ophthalmoscopes (No imaging or image transfer capability) : Welch Allyn; Heine; Keeler Ophthalmic Fundus Cameras (Expensive, not portable): Centervue; Optovue; Topcon; Kowa
- Smartphone/Lens Attachments : Welch Allyn iExaminer (poor image, $900 price) Peek Vision (no application . Only for use with dilated eyes) • Digisight (Indirect Ophthalmoscopy. Only for use with dilated eyes) • Volk iNview (Indirect Ophthalmoscopy. Only for use with dilated eyes) • D-EYE Retinal Imaging System mixes the ophthalmoscope optical qualities with fundus camera digital capabilities and with connectivity of a smartphone (only for clinicians).
By entering into a partnership with one or more of these companies, we hope to be able to reach the appropriate number of devices to be distributed in the cities.
The study on the distribution of devices in places that we call Previnet is underway. But we have identified at least one place that is not "hot" in terms of turnout (ex pharmacies, clinics, etc.) or the photographers' shops.
We chose this place as an example for two reasons: These devices to photograph the ocular fundus have a non-trivial learning curve, and the photographers could learn faster than other professionals.
In addition, the photographer's market is in crisis due to digitalization, so we would like to bring an activity that can increase their business.
We will take care of distributing the devices to these photographers, who will create the first PREVYPOINTs, and we will take care of doing the first training to PREVYAGENT.
The impact that this operation will have on the flows of patients over 65 will have to be calculated based on the arrangement of the Previpoints, but from the numbers it is clear that we are talking about important flows. Try to work with this numbers: 31.948.548 patients In Italy, 392.200.600 in Europe, and in the case study for a total of 1.029.963 patients in a single region 799.917,30 will move from one city to another one to access to ocular fundus image exam.
The strategy with which these previpoints are positioned will take place in collaboration with the containment plans of the individual city or region.
Software
PREVYAPP Our friend Mr. Rossi must perform an examination of the fundus. The health system, through the Health prevention campaigns, has provided Prevyapp to all patients enrolled in the social disease program. By logging into Prevyapp, the app recommends the closest place to go for the exam. Once the Prevypoint has been found, Mr. Rossi will be helped by PREVYAGENT to acquire the exam. Mr. Rossi will use the app with the device that provides prevypoint. Once the image has been acquired, the app will automatically send the image to the ophthalmologist, who will report the exam, and automatically this report will be sent to the specialists who follow the patient. With Previapp We provide patients an accessible system to receive the proper assistance by creating a distributed health network which links patients and all the doctors involved in the prevention process. With this simple solution we can reduce of ⅓ the amount of patients who need to perform ocular fundus examinations for prevention. 31.948.548 patients In Italy, means 95845644 displacement of the fragile patient in hospital places at high risk of covid infection. With Previapp we reduce virtually an amount of 63897096 iteration in the risk spot.
Artificial Intelligence
During the last years, ophthalmology has been receiving a massive contribution from artificial intelligence to provide diagnostic support with extremely high accuracy, and some of these AI systems have been approved by the FDA for the prevention of social diseases.
While most of the studies are based on data and images acquired by means of the proper instruments, the state of the art reports several investigations about AI models trained using smartphone-based image acquisition, which results are characterized by high sensitivity.
The PREVYAPP will be provided with the proper AI system for images quality assessment . The Platform will integrate an AI system for social pathologies detection (glaucoma, diabetic retinopathy, age-related macular degeneration, hypertension) which aims at supporting doctors diagnoses. Besides the original ocular fundus photograph, for the purpose, doctors will be provided of:
- The AI diagnosis
- A quantitative information about the certainty of the AI diagnosis
- A graphical representation of the meaningful parts of the ocular fundus images that most affected the AI decision (e.g. heatmap)
LEGAL ASPECTS
The report on the State of Health in the EU concluded that only by fundamentally rethinking our health and care systems can we ensure that they remain fit-for-purpose. This means systems which aim to continue to promote health, prevent disease and provide patient-centred care that meets citizens' needs. Health and care systems require reforms and innovative solutions to become more resilient, accessible and effective in providing quality care to European citizens. Digital solutions for health and care can increase the well-being of millions of citizens and radically change the way health and care services are delivered to patients, if designed purposefully and implemented in a cost-effective way. Our project is designed to be compliant to the UE Legal framework. According to the articles 56 and 57 of the Treaty on the Functioning of the European Union (TFEU), telemedicine is a free service, a hybrid between health service and an information service (an electronical service provided for a fee at a distance). Therefore, both the regulations relating to health care and the regulations relating to information society services are applicable. The platform is designed to be compliant to the: Directive on patients' rights in cross-border healthcare established the eHealth network to advance the interoperability of eHealth solutions. (Directive 2011/24/EU of the European Parliament and of the Council of 9 March 2011 on the application of patients’ rights in cross-border healthcare, OJ L 88 of 4.4.2011) EU legislation on medical devices, (Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, OJ L 117 of 5.5.2017; Regulation (EU) 2017/746 of the European Parliament and of the Council of 5 April 2017 on in vitro diagnostic medical devices, OJ L 117 of 5.05 2017) data protection (Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC, OJ L 119/1 of 4.05.2016) electronic identification (Regulation (EU) No 910/2014 of the European Parliament and of the Council of 23 July 2014 on electronic identification and trust services for electronic transactions in the internal market and repealing Directive 1999/93/EC, OJ L 257 of 28.8.2014) security of network and information systems (Directive (EU) 2016/1148 of the European Parliament and of the Council of 6 July 2016 concerning measures for a high common level of security of network and information systems across the Union, OJ L 194 of 19.7.2016) Before proceeding with data collection, it is necessary to inform the patient. The platform is designed to be compliant to the GDPR: health data of the patients are subject to a general prohibition of dissemination, as well as enhanced protection (article 4 of GDPR) and according to the European regulation health data can be used only for purposes related to health (treatment purposes). According to article 24 of the GDPR …“ the controller shall implement appropriate technical and organisational measures to ensure and to be able to demonstrate that processing is performed in accordance with this Regulation” (for for example through data anonymization and the use of cryptographic codes for the transmission). The security, integrity, and availability of data (including backup and archiving) are the tasks of the IT staff.
Business Model
The business revenue model can be divided in three main streams:
- Public health Today, in Italy, in order to perform only ocular fundus examination, Public health spends 7,75 € each follow up visit. It means that the national health system spends only for the prevention of diabetes complications 54.758.592€. If we add the amount of the other social disease we take care, we arrive at an amount of 54.758.592 €. Previnet fits into this market.
- Data sourcing Data will be accessible in two different way based on the user and the collaboration needed: Annual/monthly subscription → Big pharma company Project basis: this revenue stream is created in order to create a deep relationship with the client. Project charges are customized based on the need of the customer. → University for research
- Private Clinicians Private Clinicians, that at the beginning will have access to all the AI features on diagnosis support. Once that the Application reaches a critical mass, it will be changed with a monthly fee. Accomplishments that we’re proud of
From the beginning we believed in the possibility of facing such an important problem as a pandemic.
The idea behind this project was born only one day before the start of the hackathon. Thanks to the union of deep knowledge in ophthalmology and AI we have intercepted a powerful union with urban planning and design. We managed to develop the work presented in just 2 days, but we have in mind what there is to do.
We want to report the words of one of us, for which this has been the first hackathon:
As long as we are united, there will be no defeat. We learned how to develop an idea and bring it among us until we can touch it. Engineers, lawyers, doctors, graphic designers and computer scientists interacted with all their knowledge. Every professional has abandoned his clothes to go back to being a student: engineers who listened to pathologies, economists who were looking for the best graphic solution, doctors who think about canvas. We started from afar, each of us used different languages but each of us had the same desire to bring his help against the covid. We fought and believed that we can combine our knowledge, going from books to magazines, from trivial programs to algorithms on artificial intelligence, from flows of a few people to millions of people. What we imagined is already here, and we are a part of it!
What's next for prevynet
- Short term: Calabria test
Strategic study on PREVIPOINTS distribution Create a network Ophthalmologists Find a partner for lenses distribution Design and develop the software infrastructure Measures the actual impact of PREVINET during the covid emetìrgency Calculate the breaking with actual revenue stream and cos
- Mid/Long term Large scale distribution AI System Integration
THE NECESSITIES IN ORDER TO CONTINUE THE PROJECT
We need financing and technical support for our research, find data about populations affected by social diseases, in order to analyze flows of other urban contexts.
Establish Network of professionals in order to certify the testing with medical doctors.
Have connection support: talking with the best drug brands (Nova Nordisk for diabetics, Novartis for hypertension, etc.), opinion leaders, key testimonials in the main Diabetes, Glaucoma and Hypertension social pathologies.
Legal support to our legal department to prepare licenses for every partner and interact with every European nation Healthcare representative for interacting with all the different healthcare systems in order to spread and disseminate the service
Telemedicine equipment ad certification The equipment should be compliant to the CEI EN 60601-1 (electro-medical devices for the diagnosis and treatment of patients). The basis of this legislative framework is the Consumer Protection Act of 1987 dealing with the general responsibility for artisanal products and applied to the teleconsultations equipment. The device must show the European conformity (CE) mark indicating that they comply with the appropriate safety, quality, and performance standards.
THE VALUE OF YOUR SOLUTION(S) AFTER THE CRISIS
Our solution will impact consistently on the healthcare system in emergency context as well as for the normal one. One of the key aspects of our idea is to act on social diseases, which represent a very threat during a pandemic, but they will not disappear with the passing of the crisis!
Another aspect of the PREVYNET project that will assign value to the service, has to do with the data economy. The service will collect clinical data that will be available on the PREVYBASE in order to enable AI and research for universities, research centers and other European bodies.
The PREVINET team may catch the highest risk COVID patients, reduce risks, congestion and costs, adopt the most innovative adaptive technology applicable to clinician and assistance practice, making it resilient to the virus and ready for the future of AI based telehealth.
Bibliography
Rajalakshmi, R., Subashini, R., Anjana, R.M. et al. Automated diabetic retinopathy detection in smartphone-based fundus photography using artificial intelligence. Eye 32, 1138–1144 (2018)
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. 2019;137(10):1182–1188. doi:10.1001/jamaophthalmol.2019.2923
King H, Aubert RE, Herman WH. Global burden of diabetes, 1995-2025:prevalence, numerical estimates, and projections.Diabetes Care. 1998 Sep;21(9):1414-31. PubMed PMID: 9727886.
Wong WL, Su X, Li X, Cheung CM, Klein R, Cheng CY, Wong TY. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob Health. 2014Feb;2(2):e106-16. doi:10.1016/S2214-109X(13)70145-1. Epub 2014 Jan 3. Review. PubMed PMID: 25104651.
Bloch MJ. Worldwide prevalence of hypertension exceeds 1.3 billion. J Am Soc Hypertens. 2016 Oct;10(10):753-754. doi: 10.1016/j.jash.2016.08.006. Epub 2016 Sep 19. PubMed PMID: 27660007.
Lacruz ME, Kluttig A, Hartwig S, Löer M, Tiller D, Greiser KH, Werdan K,Haerting J. Prevalence and Incidence of Hypertension in the General Adult Population: Results of the CARLA-Cohort Study. Medicine (Baltimore). 2015 Jun;94(22):e952. doi: 10.1097/MD.0000000000000952. PubMed PMID: 26039136; PubMed Central PMCID: PMC4616348.
Gulshan V, Peng L, Coram M, et al. Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. JAMA. 2016;316(22):2402–2410. doi:10.1001/jama.2016.17216
Ting DSW, Cheung CY, Lim G, et al. Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes. JAMA. 2017;318(22):2211–2223. doi:10.1001/jama.2017.18152
Sayres R, Taly A, Rahimy E, Blumer K, Coz D, Hammel N, Krause J, Narayanaswamy A, Rastegar Z, Wu D, Xu S, Barb S, Joseph A, Shumski M, Smith J, Sood AB, Corrado GS, Peng L, Webster DR. Using a Deep Learning Algorithm and Integrated Gradients Explanation to Assist Grading for Diabetic Retinopathy. Ophthalmology. 2019 Apr;126(4):552-564. doi: 10.1016/j.ophtha.2018.11.016. Epub 2018 Dec 13. PubMed PMID: 30553900
Burlina PM, Joshi N, Pacheco KD, Liu TYA, Bressler NM. Assessment of Deep Generative Models for High-Resolution Synthetic Retinal Image Generation of Age-Related Macular Degeneration. JAMA Ophthalmol. 2019 Mar 1;137(3):258-264 doi: 10.1001/jamaophthalmol.2018.6156. Erratum in: JAMA Ophthalmol. 2019 Feb 14; PubMed PMID: 30629091; PubMed Central PMCID: PMC6439721
Matsuba S, Tabuchi H, Ohsugi H, Enno H, Ishitobi N, Masumoto H, Kiuchi Y. Accuracy of ultra-wide-field fundus ophthalmoscopy-assisted deep learning, a machine-learning technology, for detecting age-related macular degeneration. Int Ophthalmol. 2019 Jun;39(6):1269-1275. doi: 10.1007/s10792-018-0940-0. Epub 2018 May 9. PubMed PMID: 29744763
Peng Y, Dharssi S, Chen Q, Keenan TD, Agrón E, Wong WT, Chew EY, Lu Z. DeepSeeNet: A Deep Learning Model for Automated Classification of Patient-based Age-related Macular Degeneration Severity from Color Fundus Photographs. Ophthalmology. 2019 Apr;126(4):565-575. doi: 10.1016/j.ophtha.2018.11.015. Epub 2018 Nov 22. PubMed PMID: 30471319; PubMed Central PMCID: PMC6435402
Yoo TK, Choi JY, Seo JG, Ramasubramanian B, Selvaperumal S, Kim DW. The possibility of the combination of OCT and fundus images for improving the diagnostic accuracy of deep learning for age-related macular degeneration: a preliminary experiment. Med Biol Eng Comput. 2019 Mar;57(3):677-687. doi: 10.1007/s11517-018-1915-z. Epub 2018 Oct 22. PubMed PMID: 30349958
Burlina PM, Joshi N, Pekala M, Pacheco KD, Freund DE, Bressler NM. Automated Grading of Age-Related Macular Degeneration From Color Fundus Images Using Deep Convolutional Neural Networks. JAMA Ophthalmol. 2017;135(11):1170–1176. doi:10.1001/jamaophthalmol.2017.3782
Treder, M., Lauermann, J.L. & Eter, N. Deep learning-based detection and classification of geographic atrophy using a deep convolutional neural network classifier. Graefes Arch Clin Exp Ophthalmol 256, 2053–2060 (2018)
Sandhu HS, El-Baz A, Seddon JM. Progress in Automated Deep Learning for Macular Degeneration. JAMA Ophthalmol. 2018;136(12):1366–1367. doi:10.1001/jamaophthalmol.2018.4108 An Automated Grading System for Detection of Vision-Threatening Referable Diabetic Retinopathy on the Basis of Color Fundus Photographs Zhixi Li, Stuart Keel, Chi Liu, Yifan He, Wei Meng, Jane Scheetz, Pei Ying Lee, Jonathan Shaw, Daniel Ting, Tien Yin Wong, Hugh Taylor, Robert Chang, Mingguang He Diabetes Care Dec 2018, 41 (12) 2509-2516; DOI: 10.2337/dc18-0147
Grassmann F, Mengelkamp J, Brandl C, Harsch S, Zimmermann ME, Linkohr B, Peters A, Heid IM, Palm C, Weber BHF. A Deep Learning Algorithm for Prediction of Age-Related Eye Disease Study Severity Scale for Age-Related Macular Degeneration from Color Fundus Photography. Ophthalmology. 2018 Sep;125(9):1410-1420. doi:10.1016/j.ophtha.2018.02.037. Epub 2018 Apr 10. PubMed PMID: 2965386
Abràmoff MD, Lou Y, Erginay A, Clarida W, Amelon R, Folk JC, Niemeijer M. Improved Automated Detection of Diabetic Retinopathy on a Publicly Available Dataset Through Integration of Deep Learning. Invest Ophthalmol Vis Sci. 2016 Oct 1;57(13):5200-5206. doi: 10.1167/iovs.16-19964. PubMed PMID: 27701631
Gargeya R, Leng T. Automated Identification of Diabetic Retinopathy Using Deep Learning. Ophthalmology. 2017 Jul;124(7):962-969. doi: 10.1016/j.ophtha.2017.02.008. Epub 2017 Mar 27. PubMed PMID: 28359545
Takahashi H, Tampo H, Arai Y, Inoue Y, Kawashima H. Applying artificial intelligence to disease staging: Deep learning for improved staging of diabetic retinopathy. PLoS One. 2017 Jun 22;12(6):e0179790. doi: 10.1371/journal.pone.0179790. eCollection 2017. PubMed PMID: 28640840; PubMed Central PMCID: PMC5480986
Bajwa MN, Malik MI, Siddiqui SA, Dengel A, Shafait F, Neumeier W, Ahmed S. Two-stage framework for optic disc localization and glaucoma classification in retinal fundus images using deep learning. BMC Med Inform Decis Mak. 2019 Jul 17;19(1):136. doi: 10.1186/s12911-019-0842-8. Erratum in: BMC Med Inform Decis Mak. 2019 Jul 31;19(1):153. PubMed PMID: 31315618; PubMed Central PMCID:PMC6637616.
Shibata N, Tanito M, Mitsuhashi K, et al. Development of a deep residual learning algorithm to screen for glaucoma from fundus photography. Sci Rep. 2018;8(1):14665. Published 2018 Oct 2. doi:10.1038/s41598-018-33013-w
Ahn JM, Kim S, Ahn KS, Cho SH, Lee KB, Kim US. A deep learning model for the detection of both advanced and early glaucoma using fundus photography [published correction appears in PLoS One. 2019 Jan 25;14(1):e0211579]. PLoS One. 2018;13(11):e0207982. Published 2018 Nov 27. doi:10.1371/journal.pone.0207982
Christopher M, Belghith A, Bowd C, et al. Performance of Deep Learning Architectures and Transfer Learning for Detecting Glaucomatous Optic Neuropathy in Fundus Photographs. Sci Rep. 2018;8(1):16685. Published 2018 Nov 12. doi:10.1038/s41598-018-35044-9
Hemelings, R., Elen, B., Barbosa-Breda, J., Lemmens, S., Meire, M., Pourjavan, S., Vandewalle, E., Van de Veire, S., Blaschko, M.B., De Boever, P. and Stalmans, I. (2020), Accurate prediction of glaucoma from colour fundus images with a convolutional neural network that relies on active and transfer learning. Acta Ophthalmol, 98: e94-e100. doi:10.1111/aos.14193
Bajwa MN, Malik MI, Siddiqui SA, et al. Two-stage framework for optic disc localization and glaucoma classification in retinal fundus images using deep learning [published correction appears in BMC Med Inform Decis Mak. 2019 Jul 31;19(1):153]. BMC Med Inform Decis Mak. 2019;19(1):136. Published 2019 Jul 17. doi:10.1186/s12911-019-0842-8
Log in or sign up for Devpost to join the conversation.