India is actively promoting a digital health ecosystem to support its quest for universal healthcare. Quality-graded health data can provide exceptional decision-making support in this endeavor. The biggest hurdle to developing artificial intelligence (AI)-enabled healthcare technologies in India is the availability of high-quality annotated data for South Asians, especially Indians, that can be used to train and validate AI applications.

The large pool of multimodal data collected during patient treatment journeys or disease-specific surveillance provides unique archival opportunities for large medical datasets. Once annotated, medical images such as X-ray, computerized tomography (CT), magnetic resonance imaging (MRI), ultrasound, and whole-slide images acquired during an individual’s visit to healthcare institutions are a rich resource for training AI algorithms. Such endeavors include the Medical Imaging Databank of the Valencia Region, PathLAKE, The Cancer Imaging Archive¹, and the Medical AI Data for All initiative². Some of the existing medical imaging datasets from India are the IN-CXR (an open dataset of chest radiography for tuberculosis), the Indian Diabetic Retinopathy Image Dataset (IDRiD)³, Chákṣu (a glaucoma-specific fundus image database⁴), and CHAVI-RO (a comprehensive digital archive of cancer imaging from radiation oncology⁵). However, India has no centrally coordinated and discoverable repository for medical imaging data with established interoperable standards.

印度正在积极推广数字健康生态系统，以支持其对全民医疗保健的追求。质量分级的健康数据可以在这项工作中提供卓越的决策支持。在印度开发人工智能 （AI） 支持的医疗保健技术的最大障碍是为南亚人（尤其是印度人）提供高质量的注释数据，这些数据可用于训练和验证 AI 应用程序。

在患者治疗旅程或特定疾病监测期间收集的大量多模式数据为大型医疗数据集提供了独特的存档机会。注释后，X 射线、计算机断层扫描 （CT）、磁共振成像 （MRI）、超声和个人访问医疗机构期间获取的全玻片图像等医学图像是训练 AI 算法的丰富资源。这些努力包括瓦伦西亚地区的医学成像数据库、PathLAKE、癌症成像档案¹ 和全民医学 AI 数据计划²。印度现有的一些医学成像数据集包括 IN-CXR（结核病胸部 X 光检查的开放数据集）、印度糖尿病视网膜病变图像数据集 （IDRiD）³、Chákṣu（青光眼特异性眼底图像数据库⁴）和 CHAVI-RO（放射肿瘤学癌症成像的综合数字档案⁵）。但是，印度没有具有既定可互操作标准的集中协调和可发现的医学成像数据存储库。