ABSTRACT
Developing countries like Nepal face challenges in accessing health services due to sparse distribution in communities, difficult geographic terrain, limited transportation, poverty, and lack of health human expertise in rural areas. The COVID-19 pandemic added woes to the wound. To address this gap, the Hospital for Children, Eye, ENT, and Rehabilitation Services adopted an innovative approach to remote rural patient care using telehealth and artificial intelligence in close coordination with IT professionals and healthcare professionals. We developed a deep learning-based disease prediction model that incorporates telemedicine with AI for screening and diagnosing Eye and ENT diseases using nonspecialist health workers. Deep learning-based disease prediction models in Diabetic Retinopathy (DR) and Glaucoma added quality specialized services to telehealth. This paper presents the adoption of digital innovations and the incorporation of telehealth to tackle various diseases. To predict DR, 61,458 colorful retinal photographs from fundus photography and 1500 for Glaucoma were used. To reduce the biases, EyePACS data sets were also incorporated. Inception V3 transfer learning model was used for DR and employed DenseNet architecture for Glaucoma. An accuracy of more than 90 %in both models was achieved. Accurate specialized diagnosis, better medical care, patient monitoring, limited specialized hospital visits, and easier with shorter wait times are now possible. In the future, this successful model can be replicated nationally and in other developing countries. © 2022 IEEE.