Generalizability assessment of AI models across hospitals in a low-middle and high income country | Synapse