Inspiration/Context Healthcare management has rapidly shifted online, but patient portals are notoriously difficult to navigate. For millions of individuals—including seniors, people with visual or motor disabilities, and those with low health literacy—booking an appointment, checking lab results, or understanding a medical bill can feel like an impossible task. Missing an appointment or misunderstanding dosage instructions isn't just a minor inconvenience; it’s a real health risk.
We were inspired to build AccessMed to ensure that digital healthcare is an accessible right for everyone, regardless of physical capability or medical knowledge.
What it does AccessMed is a client-side browser extension (Manifest V3) that acts as an intelligent accessibility and plain-language layer over any patient portal (such as Epic, MyChart, or Athenahealth).
Visual Enhancements: Auto-detects and injects missing aria-labels and alt text for icon buttons, offers customizable high-contrast color modes, and allows instant font scaling.
Motor & Navigation Support: Overrides strict form timeouts that stress users with mobility limitations and optimizes keyboard navigation across complex multi-step menus and date pickers.
Health Literacy & Plain Language Toggle: Automatically scans the web page for dense medical jargon and confusing insurance terms (like deductible, benign, or hypertension), replacing or inline-defining them with clear, everyday language.
How we built it Extension Framework: Built using Manifest V3 to ensure modern security compliance, fast performance, and cross-browser compatibility (Chrome and Firefox).
DOM Manipulation & Accessibility Injection: JavaScript content scripts utilize dynamic MutationObserver instances to continuously audit single-page application (SPA) updates, injecting ARIA attributes and styling overrides on the fly.
Text Analysis Engine: Developed a lightweight DOM tree-walking algorithm paired with custom regular expression mappings to parse text nodes without breaking the underlying layout or form functions of the hospital's portal.
User Interface: Clean HTML/CSS popup menu providing one-click toggles for high-contrast mode, font size sliders, and feature switches.
Challenges we ran into Dynamic Web Apps (SPAs): Modern patient portals frequently re-render DOM elements dynamically. Ensuring our ARIA injection and text simplifiers re-ran without causing lag or infinite re-render loops required careful use of mutation observers and debouncing.
Preserving Functional Layouts: Overriding inline styles for high contrast and font scaling without breaking responsive grid layouts or hidden form fields was a tricky CSS exercise.
Balancing Medical Accuracy with Simplicity: Simplifying medical terminology requires extreme accuracy. We had to carefully curate dictionary definitions so terms remained medically accurate while remaining simple enough for a non-expert.
Accomplishments that we're proud of Zero Backend Requirement: We succeeded in making AccessMed 100% client-side, meaning it works immediately on third-party sites without needing hospital IT departments to modify their code or APIs.
Seamless Real-Time Parsing: Achieving smooth DOM text node replacement for plain-language definitions without visible page flicker or UI disruption.
Universal Impact: Creating a lightweight solution that solves visual, motor, and cognitive barriers simultaneously in a single tool.
What we learned Deepened our understanding of WCAG 2.1 guidelines and how critical programmatic aria-labels are for screen reader users.
Learned the nuances of working with Manifest V3 service workers and content script isolation.
Gained valuable insight into health literacy challenges and how design choices directly impact patient outcomes.
What's next for AccessMed AI-Powered Translation & Summarization: Integrate a lightweight local AI model or LLM API to summarize complex multi-page PDF medical records into 3 bullet points.
Voice Command Suite: Add Web Speech API capabilities so users with severe motor impairments can navigate portals entirely hands-free using voice commands like "Read lab results" or "Click messages."
Localized Multilingual Support: Expand the Plain Language engine to automatically translate medical summaries into multiple languages for non-native English speakers.
Log in or sign up for Devpost to join the conversation.