CONTEXT & OVERVIEW
CoDoctor is a white-box AI assistant for attending physicians, standardizing unstructured clinical notes and producing transparent diagnostic hypotheses with 0% generative distortion.
PLATFORM — WEB / DESKTOP, CLINIC WORKSTATIONS
TOOLS — FIGMA
01 — PROBLEM STATEMENT
Up to 80% of clinically relevant information lives in handwritten notes, fragmented EHR templates, and inconsistent terminology — easy to miss under everyday time pressure.
Success meant reducing time-to-diagnosis and time-to-insight while improving extraction accuracy and traceability for EHDS-ready interoperability in 2027.
02 — DISCOVERY & RESEARCH

The audience included general practitioners, primary-care physicians, and specialists. Competitive review of Linguamatics and Savana clarified the need for error robustness, zero-shot domain support, and open reasoning.
Clinicians wanted ranked hypotheses and the next tests to order — not an opaque answer. Every step had to remain visible and reviewable.
03 — IDEATION & ARCHITECTURE

USER FLOW — CASE OPENING AND ANALYSIS
Find or select a patient → open the patient card → add history by typing or upload → run AI analysis → validate extracted entities → review ranked results → form appointments.
The information architecture connects Home, Patient List, Cases and Case Log, Patient Card, and Disease Diagnosis Results into one continuous workflow.
04 — ITERATION & VALIDATION

The structure began as paper wireframes and card-layout sketches, then progressed through Home Page V2, Cases V2, Case Log V2, and Patient List V2.
Diagnostic-result layouts and the Add Anamnesis modal were tested with and without guidance, refining the balance between speed and clinician support.
05 — UI & DESIGN SYSTEM

System building blocks — Header with global search and user profile; Table, Header Table Row, Table Row, Table Item, Accordion; Text Field with validation and error states; Upload Area with progress.
Core product screens — Patient Profile / Card with history, active and archived cases, and vitals analysis; Disease Diagnosis Results grouped as Top Hits, Probable, and Unlikely with ICD coding and symptom triggers.
06 — OUTCOMES & IMPACT

Q1 2026: prototype and closed beta. Q2–Q3: standalone API and EHR pilots, with UX/UI metric tracking and clinic feedback. Q4 2026–2027: strategic integrations and scaling. The atomized Figma design system, component states, and detailed user-flow diagrams are ready for engineering handoff.


