EHR Dashboard
Less Admin,
More Care
An optimized dashboard designed to restore face-to-face time for outpatient psychiatrists.
Glance EHR
3 months
October 2025
Product Design
CONTEXT
Outpatient psychiatrists face a critical dilemma: the more time they spend managing data, the less time they have for patients. Interviews revealed that administrative tasks are consuming valuable clinical minutes, fracturing the therapeutic connection essential for mental health treatment.
EHR-related tasks make up ~54% of total visit time
EHR-related work per visit: ~18.6 min
Face-to-face time: ~16.5 min
Time-motion study, Family Medicine, 2018
OBSERVING
SUBTLE CUES
CONTINUITY
OVER TIME
NARRATIVE
UNDERSTANDING
LISTENING
GOOD
CARE
JAMA Psychiatry, APA, Psychiatric Times, Lancet Psychiatry
For psychiatrists, conversation and connection are the core of treatment, face time directly impacts care quality.
PROBLEM STATEMENT
Clinical data is currently fragmented across long narrative notes and scattered tabs. Psychiatrists are forced to function as "data clerks" during appointments, sacrificing face-to-face interaction to hunt for information.
GOAL
Information
Visibility
Make essential patient information easy to locate
Documentation
Support
Assist documentation with LLM-powered summaries and drafts.
Clinical Decision
Support
Support decision making with supervised ML medication recommendations and alerts.
Restore clinical focus by designing a dashboard that centralizes critical patient history, automates low-level documentation, and surfaces risks by reducing cognitive load during the session.
RESEARCH
Due to HIPAA regulations and enterprise licensing, accessing competitor platforms for analysis was impossible. Instead of relying on market assumptions, I pivoted to primary research to understand the psychiatrist's mental model.
Method 1: Interviews & Journey Mapping
I interviewed five outpatient psychiatrists to map their end-to-end workflow from pre-visit prep to post-visit documentation. The journey map revealed that the highest friction occurs during the patient interaction, where multitasking is critical.
Key Finding
Regardless of the specific tool, every psychiatrist followed the same mental checklist before entering the room:
1
Identification
Patient name, DOB, Medical ID#
2
The Narrative
Last visit note & follow-up summary
3
The Context
Recent medication changes
4
The Team
Care plans from therapists/social workers
Due to the limited access to competitor platforms, I conducted a card-sorting session where psychiatrists grouped EHR data points to reflect their preferred information structure.
Method 2: Card Sorting
To translate this "ritual" into a UI structure, I conducted a card-sorting session. Psychiatrists grouped data points based on priority, revealing a hierarchy that prioritizes clinical safety over administrative details.
RESEARCH INSIGHTS
Cognitive Overload
Dashboards show everything at once, forcing doctors to filter signal from noise manually
Fragmented Medication History
Recent changes are scattered across different tabs, making it risky and slow to reconstruct the timeline.
Visual Clutter
Old labs and administrative details compete for space with urgent clinical data.
Buried Action Items
Next steps are often hidden deep within narrative paragraphs, requiring scroll-and-hunt behavior.
IDEATION
Derived from user insights, the new structure surfaces high-frequency data instantly, allowing clinicians to review patient history without leaving the dashboard.
TESTING & ITERATION
I tested the prototype with five clinicians to validate decision speed and clarity. The goal: Ensure critical data could be located in under 10 seconds.
Enhancing Scannability
Mid-Fidelity
Flat visual hierarchy made it difficult to distinguish between sections, slowing down the 'Pre-Visit Ritual
Refined Design
Introduced distinct cards and clearer grouping to allow rapid scanning between patient appointments
Optimizing Documentation Smoother Workflow
Mid-Fidelity
Dense layouts slowed scanning and made visit categorization and note completion unclear.
Refined Design
Visual grouping, visit-type tags, autosave feedback, and clear CTAs improve clarity, organization, and confidence.
Clarifying AI Intent
Mid-Fidelity
The command 'Rewrite with AI' caused hesitation. Clinicians worried the model might alter medical facts or the clinical meaning
Refined Design
Renamed to 'Magic Polish' to clearly signal that the tool only cleans up grammar and typos—ensuring the doctor's original intent remains untouched.
FINAL DESIGN
Designed for Clincial Speed
Glance surfaces only the essentials—reducing navigation, cognitive load, and time-to-action for high-pressure environments.
One Click Navigation
Intuitive entry points allow instant access to the patient dashboard without menu-diving.
Search
Schedule
ID Tags
1 CLICK
Quick access to key information
Signal Over Noise
Surfaces only high-value data: key patient details, concise summaries, and critical alerts. No clutter, no distraction.
Viusal Visit Timeline
A color-coded, chronological view allows clinicians to scan patterns and history in seconds—no need to open individual notes.
Anchored Patient Identity
Critical identifiers (Name, DOB, ID) remain visible at the top to prevent errors and ensure immediate orientation.
Alternative Options
Intelligently suggests alternatives based on side-effect profiles and insurance constraints.
Similar Medication
Supervised ML suggests alternative medications based on recent notes, side-effect profiles, and insurance coverage constraints.
Healthy Friction
Requires explicit clinician approval for all AI edits. This confirmation step ensures 100% medical accuracy and legal accountability before signing.
Predictive Scheduling
Suggests follow-ups based on visit type and historical cadence, reducing manual administrative friction.
Clear Disclaimers
Visual badges ensure users know when content is machine-generated, requiring final human sign-off for safety.
IMPACT
Reducing Interaction Cost
A task analysis comparing the standard 'Pre-Visit Ritual' workflow against the Glance prototype.
Clinician Confidence
In usability testing, 5/5 psychiatrists successfully located risk factors in under 10 seconds, reporting significantly lower cognitive effort compared to their current tools.
REFLECTION
Limited Resources…
Accessing competitor platforms was impossible due to HIPAA regulations and enterprise licensing. Instead of guessing, I pivoted to primary research: conducting extensive user interviews to deconstruct the clinical workflow and card-sorting sessions to uncover the psychiatrists' mental models.
This constraint actually led to a more user-centered result. The final architecture was built on their actual logic, rather than simply copying existing (and often flawed) market standards.
Other Projects