Embedded Files
Minimising cognitive load to ensure patient safety in high-stress clinical workflows.
A zero-memory-dependency interface designed for rapid, error-free documentation.
Leveraging the exact strategic frameworks used for leading global clients, this project serves as a comprehensive demonstration of my end-to-end consulting workflow, from strategic discovery to validation planning.
MediFlow - 4K.movDemo Video
THE CHALLENGE
Designing a EMR Ecosystem from Scratch
Medical professionals are often overwhelmed by complex, outdated administrative tools. The objective was to conceptualise the core 'Patient Chart' module for MediFlow Manager, a new practice management system aiming to disrupt the market.
THE MISSION
With no existing legacy constraints (Greenfield Project), the mission is to architect a high-efficiency workflow that reduces cognitive load and ensures maximum patient safety, translating the vision from abstract requirements to final high-fidelity designs.
THE LOGIC : Double Diamond Process
User-centred, data-driven, and technically validated.
GOAL:
Minimising development risk & maximising patient safety
1. UNDERSTAND
Requirement Analysis
Key Question
“Which tasks must/should MediFlow Manager perform?"
Goal
Validating initial ideas.
Real-world Workflow Analysis
Key Question
"How do doctors and medical assistants currently work, and what are their pain points/needs?"
Goal
Understanding the "As-Is" state.
Legacy-Check
Key Question
"How is the data currently structured?”
Goal
Understanding the legacy "As-Is" state and avoiding Design Debt.
2. DEFINE
Writing User Stories
Key Question
"What do doctors and medical assistants (MFAs) expect from MediFlow Manager?"
Goal
Defining the product vision and direction from the users' perspective.
Defining Information Architecture
Key Question
"What information should be included in MediFlow Manager, and in what hierarchy/order?"
Goal
Establishing a clear and intuitive information structure.
Core KPI festlegen
Key Question
“Which strategic goals best reflect our success?”
Goal
To evaluate operational performance and task execution efficiency.
3. IDEATION
Sketching & Ideation
Key Question
"What is the fastest way to visually solve the requirements?"
Goal
Generating a wide range of solution approaches without getting lost in the details.
Low-fidelity Wireframe
Key Question
"Is the information flow logical? Does it support the physician's mental model and the SOAP note framework?"
Goal
Defining structure and hierarchy before investing time into pixel-perfect design.
Feasibility Check
Key Question
Is this concept viable considering legacy data and current system performance?"
Goal
Avoiding Design Waste by identifying technical hurdles (e.g., loading times for large patient lists) early with developers.
4. VALIDATION
High-Fidelity Prototype
Key Question
"Does the interaction feel realistic? Are all states (including error states) covered?"
Goal
Providing a testable artifact that is as close to the final product as possible.
Iterative Usability Testing
Key Question
"Can the physician enter a diagnosis quickly and without errors? Are there any risks to patient safety?"
Goal
Identifying usability issues and potential use errors (medical errors) before the development phase.
Developer Collaboration & Handover
Key Question
"Do the developers have all the information needed to implement the design without misinterpretations?"
Goal
Avoiding a traditional "waterfall" handover; instead, maintaining a collaborative partnership until release to ensure design quality in the code.
CROSS-FUNCTIONAL ORCHESTRATION
UX design acts as the strategic nexus between Business Viability (Scope), Technical Feasibility (Legacy Data), and Clinical Safety (User Needs).
By establishing a continuous feedback loop with the 'Three Amigos' and medical experts, I ensure that every design decision is realistic, profitable, and safe.
USER JOURNEY MAP
I deconstruct behavioral sequences through in-depth qualitative analysis, simulating high-probability edge cases and non-linear trajectories to ensure absolute logical integrity and systemic coherence.
INSIGHTS
1. Reducing Cognitive Load
Physicians are constant multitaskers. The UI must filter information and only display what is relevant to the current context (Progressive Disclosure).
2. Efficiency through Prediction
The system should anticipate the next step (e.g., Findings -> suggest matching diagnosis) to minimise unnecessary clicks.
3. Error prevention over aesthetic
High contrast and unambiguous labels are more important than modern design. Safety takes precedence.
USER STORIES
The defined key insights and user stories formed the strategic foundation for our design decisions. Based on this, low-fidelity wireframes were developed to translate theoretical requirements directly into a tangible structure.
As General Practitioner
I want to see critical alerts (CAVE) and the reason for visit immediately upon opening a patient record,
so that I can mentally prepare in split seconds and ensure patient safety.
Cognitive Load
As General Practitioner
I want to capture documentation using smart text modules,
so that I can maintain eye contact with the patient instead of losing it through extensive typing.
Cognitive Load AI Powered
As General Practitioner
I want a clear visual separation between subjective notes and objective findings,
so that I can quickly scan and understand the documentation later.
Visual Hierarchy
As General Practitioner
I want the system to suggest matching ICD codes based on entered symptoms,
so that I can save research time and avoid coding errors.
Error Prevention Recognition
STRATEGIC IDEATION & STRUCTURAL SCAFFOLDING
Before rendering a single pixel, I utilise low-fidelity sketching as a high-speed sandbox for systemic stress-testing. These artifacts represent where complex clinical workflows are deconstructed and reconfigured for optimal cognitive efficiency.
Architectural Exploration
I simulated multiple spatial configurations, including 3-pane and Tower layouts, to identify the most resilient structure for simultaneous data retrieval and documentation.
The goal was to minimise saccadic eye movement and cognitive friction during high-pressure clinical sessions.
Workflow Orchestration
The sketches map the transition from patient triage to the SOAP (Subjective, Objective, Assessment, Plan) documentation process.
I specifically prototyped a swappable panel system to maintain contextual continuity while managing high information density.
AI-Augmented Intelligence
A critical focus was the integration of Smart AI Recommendations for ICD-10 coding.
I designed the interaction to act as a feedforward control mechanism, assisting the physician’s decision-making process without interrupting the primary clinical trajectory.
Granular Precision
Even at this embryonic stage, metadata specifications (such as monospace typography for ICD codes to ensure legibility) were defined to ensure the final interface meets the highest standards of diagnostic accuracy.
STRUCTURAL X-RAY (Low-fidelity)
To ensure absolute systemic integrity, I utilise a Reverse-Deconstruction methodology.
By stripping away visual aesthetics, I expose the raw information architecture to perform a rigorous stress test on the functional hierarchy. This diagnostic layer allows for the preemptive neutralisation of potential system failures before high-fidelity rendering.
CORE PRINCIPLES:
1. Zero-Memory Dependency (Simultaneous Visibility)
Logic: Memory Load = 0
By projecting ① past records (Static/Read) and ② present documentation (Dynamic/Write) on a single plane, the system eliminates reliance on the physician's working memory, ensuring a seamless information flow without context loss.
2. Silent Canvas: Evolutionary Visual Prioritisation
Logic: MAX Signal-to-Noise Ratio
The UI remains monochromatic to minimise baseline cognitive load.
Colour is reserved exclusively for Pre-attentive Processing, inducing an immediate limbic system response before the brain even processes the text, triggering immediate visceral responses to safety critical alerts (e.g., ③ CAVE/Allergies).
3. Mental Model Transfer: The 3-Pane Architecture
Logic: Jakob's Law
Utilising familiar UI patterns (Outlook Principle) reduces onboarding time and cognitive friction.
The clear separation between [④ Global Control - ⑤ Situational Awareness - Focus Zone] allows the physician to maintain spatial orientation during high-pressure clinical sessions.
4. Semantic Chunking & Gestalt Mapping
Logic: Miller's Law (7±2)
Raw data is encapsulated into logical visual containers.
This encourages Block Scanning rather than line-by-line reading, allowing for near instantaneous context capture and faster decision making.
HIGH-FIDELITY PROTOTYPE
This is not merely visual styling; it is the rigorous execution of the cognitive ergonomic mandates defined in the architecture phase.
The final interface adheres to a strict Silent Canvas protocol, ensuring that every pixel serves a functional purpose, minimising signal noise and maximising the physician's situational awareness.
Typography as Interface
Utilising monospace fonts for ICD-10 codes to prevent character ambiguity (e.g., '1' vs 'l'), ensuring zero error readability.
Visual Silence
The monochromatic foundation forces the user's focus solely on the Active Signal, the patient's critical data, while functional colours trigger immediate limbic responses only when necessary.
FEEDFORWARD AI CONTROL
Designed as a feedforward mechanism. The system pre-computes probable ICD-10 diagnostic paths in real-time, effectively shifting the physician's cognitive load from active Recall to passive Recognition.
This preemptive architecture neutralises potential syntax errors before they occur, while the use of Monospace typography ensures absolute character distinction (zero-ambiguity) for high-stakes alphanumeric data.
HYBRID INPUT PROTOCOL
Recognising that high-velocity clinical environments demand unbroken motor continuity, the system is engineered on a strict 'Keyboard-First' mandate.
Haptic Continuity
The entire navigational structure is mapped to keyboard shortcuts, allowing physicians to traverse lists, select codes, and finalise documentation without lifting their hands from the typing position.
Bidirectional Query Logic
To accommodate diverse cognitive retrieval patterns, the search engine utilizes a dual-index algorithm. It enables both semantic clinical terms (e.g., 'Hypertension') and raw ICD-10 alphanumeric codes (e.g., 'I10') search, ensuring flexibility between intuitive recall and precise coding.
ATOMIC DATA REFINEMENT
An integrated overlay for the rapid modulation of diagnostic precision. This micro-interaction allows immediate configuration of Certainty Status (G/V) and Spatial Localisation (L/R), ensuring absolute data granularity without forcing a context switch or disrupting the visual flow.
SIMULATED IMPACT & VALIDATION FRAMEWORK
Leveraging the exact strategic frameworks used for leading global clients, this project serves as a comprehensive demonstration of my end-to-end consulting workflow, from strategic discovery to validation planning.
The metrics below represent the calculated systemic impact of the proposed architecture.
~99%
Error Elimination
Neutralisation of ambiguity through 'Feedforward AI' validation. By shifting from open text entry to pre-computed selection (Recognition over Recall), the system theoretically eliminates 99% of syntax-based ICD-10 coding errors.
-40%
Documentation Time
Projected reduction in documentation time based on Keystroke Level Modeling (KLM). Eliminating Device Switching (Mouse ↔ Keyboard) and utilising predictive AI auto-completion drastically reduces motor operators per task.
Near-Zero
Training Latency
Leveraging Jakob’s Law via the '3-Pane Outlook Pattern'. By mirroring familiar mental models, the interface is designed to reduce the learning curve for new physicians to negligible levels, enabling immediate operational readiness.
*Metrics based on heuristic analysis and KLM projections
FIGMA STRUCTURE EVIDENCE
Design ends at code.
I maintain production-grade hygiene.
I use structured layering for immediate developer readability,
Auto-Layout for responsive scalability.
I adapt naming conventions to each dev team's standards.
I deliver files that require minimal explanation.
SELECTED B2B ENGAGEMENTS
The methodologies demonstrated above have been successfully deployed in complex B2B environments.
Due to strict Non-Disclosure Agreements (NDA), the following case studies are presented without visual artifacts, focusing exclusively on architectural strategy, team orchestration, and verified business impact.
Role: Lead Product Designer
Challenge: Streamlining rework in highly regulated workflows
Role: Lead Product Designer
Challenge: Replacing legacy maze with a spatial digital twin
*AI-generated conceptual images for visual context (NDA compliant)
Google Sites
Report abuse