United StatesHealthcare and life sciences2025 – ongoing

Care Access

Medical-record analysis with Amazon Bedrock prompt caching

Care Access cut Bedrock costs by 86% by reusing medical-record context

WorkingVendor case studyVendor ↗

Problem

Care Access repeatedly processed the same medical-record content for different analysis questions. Rising record intake increased model costs and processing time.

Approach

Care Access built an LLM medical-history review workflow on Amazon Bedrock to help match participants with clinical trials. It cached each record's static content while changing the analysis questions. The system combined answers into structured records for downstream analysis.

Results

86%reduction in Amazon Bedrock costs reported by AWSVendor case study[1]

66%reduction in processing time per medical record reported by AWSVendor case study[1]

Takeaway

The savings concerned repeated model processing. They did not establish better clinical-trial matching accuracy.

Sources

How Care Access achieved 86% data processing cost reductions and 66% faster data processing with Amazon Bedrock prompt caching
AWS20 Nov 2025Vendor publication

Source summaries

[1]AWS20 Nov 2025

AWS and Care Access staff described a deployed medical-record analysis workflow. They reported 86% lower Bedrock costs and 66% less processing time per record after prompt caching.

Full details are in the original sources.

Added 1 Nov 2026 · Updated 1 Nov 2026