项目概况
与美国知名医疗软件公司 Epic 合作，基于敏捷开发与精益创业 (Lean Startup) 方法，从 0 到 1 设计并交付了一款面向患者门户平台 Epic MyChart 的 AI 插件，用于简化复杂医疗信息、提供医疗数据总结与健康建议。54 名用户参与的可用性测试显示，该方案使用户满意度提升了 18.3%，且 83% 的用户表示愿意长期使用。
Epic
09/24 - 12/24
用户研究员、
用户体验设计师
由 Epic Systems 开发的 Epic MyChart 是美国广泛使用的患者门户，为用户提供对病历、检查结果和预约安排等医疗信息的安全数字化访问。
尽管这类患者门户提升了医疗信息透明度，但其中大量内容——例如医生记录和检验结果——本质上仍是面向医疗专业人员撰写的，因此患者往往难以真正理解和解读。
我与 MBA 战略方向成员及工程团队协作开展了用户研究。在访谈了 Epic 团队成员以及 14 位 Epic MyChart 用户后，我们梳理并界定了核心痛点，并将目标用户划分为两类：
新用户
长期用户（如慢性病患者、需要持续照护的患者）
通过深入研究，我们识别出三个用户痛点最集中的使用场景，分别是检验结果、医生记录和药物信息：
检验结果
医生记录 / 病程记录
药物信息
我们总结了三个主要的用户痛点如下：
医疗术语晦涩难懂
患者常遇到晦涩难懂的医学术语，难以理解自身的诊疗情况，也无法做出明智的决策。
检验数据含义不明
患者面对化验结果（如 TSH）时常常感到困惑，不清楚其含义，也不知道下一步该怎么做。
病历资料繁杂难理
慢性病患者或需要就诊多位专科医生的人，常因数据与病历资料过多而理不清头绪，不知从何看起。
核心设计目标（HMW问题陈述）
愿景陈述
我们的愿景是借助 AI 能力，帮助 MyChart 中的患者看懂自己的健康数据——通过简化医学术语、图表和报告，将复杂信息转化为可操作的指导，让患者能够做出明智的决策，并真正参与到自己的诊疗过程中。
我们通过头脑风暴提出了多个想法，并通过点票法和 PUGH 矩阵进行筛选，在综合评估可行性、用户影响力和商业价值后，最终确定了四个最能满足用户需求、提升整体体验的核心功能。
原型 1.0（低保真）
原型 1.0 V.S 原型 2.0
The second prototype was evaluated at UW Health Hospital, involving 24 patients who were Epic MyChart users.
User Experience
Enhance AI elements in visual design.
Improve data visualization features.
Enhance discoverability and affordance of core features.
Privacy Concern
Some participants prefer sensitive information like medication details to be hidden by default instead of arising when hovering on the AI assistant to prevent exposure in public settings.
Refining AI Assistant's Visual Identity
The human-like appearance of the AI assistant led some users to mistake it for a real doctor representative, causing confusion about the source of medical advice.
Prototypes 3.0
Building upon insights and usability testing results from prototype 2.0, I refined the design to better align with user needs.
Solving Key Pain Points with AI-Enhanced Features
Patients often encounter confusing medical terms, making it hard to understand their care and make informed decisions.
Select text with your mouse to automatically see a popup explaining medical jargon.
Patients often struggle with lab results or values (like TSH), unsure what they mean or what to do next.
The final prototype was evaluated with 15 Epic MyChart users by remote video. Usability testing revealed a 18.3% increase in user satisfaction compared with first-round testing. And it indicated that participants universally experienced improved comprehension, and 83% expressed a willingness to use the plugin regularly.
Additionally, the System Usability Scale (SUS) score improved by 12 points, reaching 72, further confirming a high level of usability and user acceptance. The tech team is is now continuing to iterate and implement the design.
Usability Testing Record
Given more time, we would continue refining and iterating on the design. While overall feedback was positive, many users primarily rely on the MyChart mobile app for daily health management. Integrating medication reminders with mobile devices would improve accessibility and ensure the AI assistant seamlessly fits into users' existing routines.