📋
Changelog
Platform updates and improvements.
Current: v0.5.1 Beta
v0.5.1feature
Pilot readiness, feedback system, legal pages
- ✓Added beta feedback modal (footer + floating button on desktop)
- ✓Added /pilot page for physician and hospital conversations
- ✓Added /privacy and /terms pages (plain-language)
- ✓Added /feedback-dashboard (protected) for beta feedback review
- ✓Expanded /analytics dashboard with specialty breakdown, zero-result tracking
- ✓Reduced "AI-powered" hype language throughout; increased healthcare navigation framing
- ✓Added "For Healthcare" nav link and "Send Feedback" footer link
v0.5feature
SEO foundation — condition pages, JSON-LD schema, sitemap
- ✓Added 12 condition landing pages at /conditions/[slug] with FAQ schema
- ✓Added /conditions index with specialty-grouped cards
- ✓Added JSON-LD schema: Organization, WebSite (SearchAction), FAQPage, MedicalWebPage, BreadcrumbList
- ✓Added canonical tags, unique metadata, and OG tags to all pages
- ✓Dynamic sitemap.ts with lastmod timestamps (17 URLs)
- ✓Security headers: X-Content-Type-Options, X-Frame-Options, Referrer-Policy, Permissions-Policy
- ✓Updated robots.txt: allows /conditions/, /pilot, /privacy, /terms
- ✓Updated condition pills to link to SEO landing pages
v0.4.1improvement
Specialty-aware semantic prefiltering + token budget fixes
- ✓Created lib/specialties.ts: 8-specialty patient + trial detection
- ✓Created lib/semanticTags.ts: 50+ regex rules across oncology, psychiatry, derm, GI, neuro
- ✓Created lib/exclusions.ts: domain-specific penalty heuristics (-6 to -60)
- ✓Integrated semantic pipeline into preRankTrials (base score unchanged)
- ✓Fixed medium/slow stage token budget overflow (was causing silent 0-result failures)
- ✓Switched runBatched to parallel batch execution for lower latency
- ✓Added ?debug=1 to /api/trials for per-trial debug breakdown
v0.4feature
Progressive hydration — 4-stage AI enrichment
- ✓Split monolithic /api/explain into 3 parallel stages: fast, medium, slow
- ✓Stage 2 (fast): fitScore + summary in ~3–6s
- ✓Stage 3 (medium): matchFactors, concerns, medicationFactors in ~8–15s
- ✓Stage 4 (slow): coordinatorQuestions, washoutConflicts, priorTreatmentRelevance in ~15–30s
- ✓Section-level skeleton loading states in TrialCard
- ✓Post-completion KV cache write via /api/explain/cache
- ✓Cache hits serve full enriched result in <300ms (8× speedup verified)
v0.3.9improvement
Two-stage ranking, provisional badges, score animations
- ✓Provisional "top match" amber badge before AI scores arrive
- ✓Score entrance animation on fit score circle
- ✓Soft reorder with DOM opacity flicker when AI scores arrive
- ✓Phase 1 penalty (-14) to suppress PK/first-in-human trials
- ✓Phase 3 bonus (+8) to surface efficacy trials above safety studies
v0.3.1feature
International travel support, location intelligence
- ✓Travel range selector: local / state / US / international
- ✓Location categorized into local, US, international site buckets
- ✓International search bypasses location filter on ClinicalTrials.gov
- ✓Site counts and recruiting countries displayed per trial card
v0.3feature
Medication intelligence, screening complexity, match factors
- ✓Medication class detection (biologic, JAK, topical, immunosuppressant)
- ✓Screening complexity badge (Simple / Moderate / Complex)
- ✓matchFactors and concerns pill tags per trial
- ✓Washout conflict detection and orange warning card
- ✓Print-ready match report at /report
- ✓Saved trials (localStorage) with bookmark icon