How Care Access achieved 86% data processing cost reductions and 66% faster data processing with Amazon Bedrock prompt caching

This post is co-written with Michelle Tat, Christopher Penrose, Rasmus Buchmann, and Daniel Hansen from Care Access.

Organizations operating in the healthcare industry face a critical challenge as they process large volumes of medical records and face a critical challenge: how to efficiently analyze thousands of medical records daily while maintaining strict security and compliance standards. LLM-based medical record analysis is a powerful solution but requires reprocessing substantial portions of medical data for each separate analysis question, leading to high operational costs and slower response times.

One organization that faced this challenge head-on was Care Access, a global health services and clinical research leader, who encountered this exact challenge when scaling their health screening program. Processing 300-500+ medical records daily required multiple separate prompts for each analysis, with each prompt reprocessing substantial portions of the medical record. With hundreds of new participants choosing to share their medical records daily, Care ...