Design how Epic should surface Cosmos-powered treatment insights to primary care clinicians in Hyperspace.

Introduction

Cosmos is Epic's de-identified dataset built from participating Epic customers. By Epic's 2026 user group meeting it covered about 320 million unique patients and 23 billion encounters. Epic already turns Cosmos into clinician tools such as Look-Alikes (finding similar patients for rare conditions), Cosmos Median Length of Stay and "Best Care Choices for My Patient", which shows how treatments worked for similar patients. The prompt asks us to design how Epic should surface Cosmos-powered treatment insights inside a primary care physician's workflow in Hyperspace. This is hard because it combines clinical safety, workflow design, AI explainability and data governance.

Step 1

Clarify Questions (3 minutes)

• Which decisions should the insights support?

• Hypothetical answer: Common, high-variation decisions in primary care, such as choosing a second diabetes medication or managing hypertension in patients with kidney disease.
• Impact: Focus on chronic disease decisions, not emergencies.

• Is this guidance or a recommendation?

• Hypothetical answer: Evidence summaries based on real-world outcomes, not orders or mandates.
• Impact: The design must be clear about uncertainty and not appear to replace guidelines.

• What regulatory constraints apply?

• Hypothetical answer: Clinical decision support must stay within FDA guidance for non-device software, which requires clinicians to be able to review the basis of recommendations.
• Impact: Transparency is a must-have.

• How will success be judged?

• Hypothetical answer: Usage by clinicians, influence on decisions, and patient outcomes over time.

Propose the Goal

Give primary care clinicians a fast, trustworthy view of how patients like theirs fared on different treatment options, inside the chart, at the moment of decision, so they can make better-informed choices with their patients.

Define the Scope

In scope: insight display in the Hyperspace chart and order workflow, transparency features, and patient sharing. Out of scope: automated ordering and specialist workflows.