Company or Institution
Coordinación General de Innovación del Gobierno de Jalisco
Industry
Health
Website
https://innovacion.jalisco.gob.mx/
Country
Mexico
Sustainable Development Goals (SDGs)
SDG 3: Good Health and Well-being
General description of the AI solution
Healthcare is one of the most dynamic and challenging sectors in Mexico and the Latin America and Caribbean (LAC). For instance, Diabetic Retinopathy (DR) faces three main problems in Mexico: i) High prevalence of diabetes, the World Health Organization (WHO) reported that the prevalence of diabetes in Mexico is around 10.4% in 2016 (and rising). ii) shortage of ophthalmologists, Mexico reports 42.5 ophthalmologists per million people (OPM), in contrast with other countries such as Spain with 105.5 OPM or Argentina with 103.6 OPM, Brazil with 67.4 OPM and iii) lack of eye illness detection services in primary health care. In this context, the Artificial Intelligence (AI) models (deep convolutional neural networks) could contribute to the early detection of Diabetic Retinopathy (DR), evaluating a considerable amount of Retinal Fundus Images (RFI) in a short period of time (from a couple of months to less than 10 min). In Jalisco’s Gov, Coordinación general de innovación developed a system of convolutional neural network models in line with the Mexican DR clinical guidelines. These models preprocess, evaluate the suitability/quality and classify the retinal fundus images with more than 90 % specificity and sensitivity on local data (never seen before by the model in the training or evaluation). Moreover, we develop an API to deploy these models in the cloud and we are working on a web application to collect the images. Regarding the AI ethical principles, note that this case (“AI-Based Referral System”) was selected by the Global Partnership of AI (GPAI) in 2020 as one responsible initiative of the AI international ecosystem.
Github, open data repository, prototype or working demo
https://gitlab.com/inteligencia-gubernamental-jalisco/retinopatia
http://34.82.227.100/docs
Publications
Suitability Classification of Retinal Fundus Images for Diabetic Retinopathy Using Deep Learning, 2022 https://www.mdpi.com/2079-9292/11/16/2564 ,
Exploración de resultados de modelos de IA para clasificar niveles de retinopatía diabética en imágenes de fondo de retina del piloto fAIr-LAC Jalisco, 2022
https://drive.google.com/file/d/15JYJFKF3B_8uYZUOB6PFom4ixercNqhU/view?usp=sharing
To Be fAIr or Not to Be: Using AI for the Good of Citizens, 2021. https://ieeexplore.ieee.org/document/9379032
AI-Based Referral System project 2020
https://thefuturesociety.org/wp-content/uploads/2021/01/TFS_GPAI-RAI-Final-Report.docx-1.pdf
Artificial Intelligence-Based Referral System for Patients With Diabetic Retinopathy
https://ieeexplore.ieee.org/document/9206433 , 2020
Deep learning models for diabetic retinopathy
screening program, 2019
https://research.latinxinai.org/papers/neurips/2019/pdf/Poster_Sanchez_Abraham.pdf
Needs
Funding
Personnel
Public Exposure
Mentorship Program