Spaces:
Sleeping
Sleeping
Update index.html
Browse files- index.html +189 -177
index.html
CHANGED
|
@@ -434,57 +434,59 @@
|
|
| 434 |
</section>
|
| 435 |
|
| 436 |
<!-- Overview -->
|
| 437 |
-
<
|
| 438 |
-
<div class="
|
| 439 |
-
|
| 440 |
-
|
| 441 |
-
<
|
| 442 |
-
|
| 443 |
-
|
| 444 |
-
|
| 445 |
-
|
| 446 |
-
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
<div class="
|
| 450 |
-
<div class="
|
| 451 |
-
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 452 |
</div>
|
| 453 |
-
<
|
| 454 |
-
|
| 455 |
-
|
| 456 |
-
|
| 457 |
-
|
| 458 |
-
|
| 459 |
-
|
| 460 |
</div>
|
| 461 |
-
<
|
| 462 |
-
|
| 463 |
-
|
| 464 |
-
|
| 465 |
-
|
| 466 |
-
|
| 467 |
-
|
| 468 |
</div>
|
| 469 |
-
<p class="text-sm text-slate-300 mt-2">
|
| 470 |
-
Structured artifact log: retrieval, ranking, policies, safety checks.
|
| 471 |
-
</p>
|
| 472 |
</div>
|
| 473 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 474 |
</div>
|
| 475 |
-
|
| 476 |
-
<!-- Right: compact system summary -->
|
| 477 |
-
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 478 |
-
<div class="font-extrabold text-lg tracking-tight mb-3">Why MedSwin is different</div>
|
| 479 |
-
<ul class="space-y-3 text-sm text-slate-300">
|
| 480 |
-
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Refuses to answer when evidence is insufficient</li>
|
| 481 |
-
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Explicit EMR + CPG coverage requirements</li>
|
| 482 |
-
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Deterministic retrieval policies (no silent guessing)</li>
|
| 483 |
-
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Local-deployable, auditable by design</li>
|
| 484 |
-
</ul>
|
| 485 |
-
</aside>
|
| 486 |
</div>
|
| 487 |
-
</
|
| 488 |
|
| 489 |
<!-- Contributions -->
|
| 490 |
<section id="contributions" class="section">
|
|
@@ -863,161 +865,171 @@ sequenceDiagram
|
|
| 863 |
</section>
|
| 864 |
|
| 865 |
<!-- Retrieval -->
|
| 866 |
-
<
|
| 867 |
-
<div class="
|
| 868 |
-
<div class="lg:
|
| 869 |
-
<
|
| 870 |
-
|
| 871 |
-
|
| 872 |
-
|
| 873 |
-
|
| 874 |
-
|
| 875 |
-
|
| 876 |
-
|
| 877 |
-
|
| 878 |
-
<div class="
|
| 879 |
-
<div class="
|
| 880 |
-
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 881 |
</div>
|
| 882 |
-
|
| 883 |
-
|
| 884 |
-
|
| 885 |
-
|
| 886 |
-
|
| 887 |
-
|
| 888 |
-
|
| 889 |
-
|
| 890 |
</div>
|
| 891 |
-
|
| 892 |
-
|
| 893 |
-
|
| 894 |
-
|
| 895 |
-
|
| 896 |
-
|
| 897 |
-
|
| 898 |
-
|
| 899 |
</div>
|
| 900 |
-
<p class="text-sm text-slate-300 mt-2">
|
| 901 |
-
Final selection enforces EMR + guideline sufficiency, diversity (MMR-style), and a strict token budget.
|
| 902 |
-
</p>
|
| 903 |
</div>
|
| 904 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 905 |
</div>
|
| 906 |
-
|
| 907 |
-
<!-- Right: sufficiency gate -->
|
| 908 |
-
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 909 |
-
<div class="font-extrabold text-lg tracking-tight mb-3">Evidence acceptance gate</div>
|
| 910 |
-
<ul class="space-y-3 text-sm text-slate-300">
|
| 911 |
-
<li class="flex gap-2"><i data-lucide="book-open-check" class="icon"></i> Required guideline recommendations present</li>
|
| 912 |
-
<li class="flex gap-2"><i data-lucide="file-heart" class="icon"></i> Patient-specific EMR signals included</li>
|
| 913 |
-
<li class="flex gap-2"><i data-lucide="layers" class="icon"></i> Redundancy reduced under budget</li>
|
| 914 |
-
<li class="flex gap-2"><i data-lucide="shield-alert" class="icon"></i> Safety critic approves synthesis</li>
|
| 915 |
-
</ul>
|
| 916 |
-
</aside>
|
| 917 |
</div>
|
| 918 |
-
</
|
|
|
|
|
|
|
|
|
|
| 919 |
|
| 920 |
<!-- Training -->
|
| 921 |
-
<
|
| 922 |
-
<div class="
|
| 923 |
-
<div class="lg:
|
| 924 |
-
<
|
| 925 |
-
|
| 926 |
-
|
| 927 |
-
|
| 928 |
-
|
| 929 |
-
|
| 930 |
-
|
| 931 |
-
|
| 932 |
-
|
| 933 |
-
<div class="
|
| 934 |
-
<div class="
|
| 935 |
-
|
| 936 |
-
|
| 937 |
-
|
| 938 |
-
|
| 939 |
-
|
| 940 |
-
<div class="
|
| 941 |
-
|
| 942 |
-
|
| 943 |
-
|
| 944 |
-
|
| 945 |
-
|
| 946 |
-
<div class="
|
| 947 |
-
|
| 948 |
-
|
| 949 |
-
|
|
|
|
|
|
|
| 950 |
</div>
|
| 951 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 952 |
</div>
|
| 953 |
-
|
| 954 |
-
<!-- Right: why KD -->
|
| 955 |
-
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 956 |
-
<div class="font-extrabold text-lg tracking-tight mb-3">Why distillation?</div>
|
| 957 |
-
<ul class="space-y-3 text-sm text-slate-300">
|
| 958 |
-
<li class="flex gap-2"><i data-lucide="cpu" class="icon"></i> Enables local inference on modest GPUs</li>
|
| 959 |
-
<li class="flex gap-2"><i data-lucide="shield-check" class="icon"></i> Preserves calibrated reasoning behaviour</li>
|
| 960 |
-
<li class="flex gap-2"><i data-lucide="settings" class="icon"></i> PEFT-friendly (LoRA / QLoRA)</li>
|
| 961 |
-
<li class="flex gap-2"><i data-lucide="lock" class="icon"></i> Institution-controlled deployment</li>
|
| 962 |
-
</ul>
|
| 963 |
-
</aside>
|
| 964 |
</div>
|
| 965 |
-
</
|
|
|
|
|
|
|
|
|
|
| 966 |
|
| 967 |
<!-- Evaluation -->
|
| 968 |
-
<
|
| 969 |
-
<div class="
|
| 970 |
-
<div class="lg:
|
| 971 |
-
<
|
| 972 |
-
|
| 973 |
-
|
| 974 |
-
|
| 975 |
-
|
| 976 |
-
|
| 977 |
-
|
| 978 |
-
|
| 979 |
-
|
| 980 |
-
<div class="
|
| 981 |
-
<div class="
|
| 982 |
-
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 983 |
</div>
|
| 984 |
-
<
|
| 985 |
-
|
| 986 |
-
|
| 987 |
-
|
| 988 |
-
|
| 989 |
-
|
| 990 |
-
|
| 991 |
</div>
|
| 992 |
-
<
|
| 993 |
-
|
| 994 |
-
|
| 995 |
-
|
| 996 |
-
|
| 997 |
-
|
| 998 |
-
|
| 999 |
</div>
|
| 1000 |
-
<p class="text-sm text-slate-300 mt-2">
|
| 1001 |
-
Final answers remain grounded in cited evidence only.
|
| 1002 |
-
</p>
|
| 1003 |
</div>
|
| 1004 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1005 |
</div>
|
| 1006 |
-
|
| 1007 |
-
<!-- Right: safety behaviour -->
|
| 1008 |
-
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 1009 |
-
<div class="font-extrabold text-lg tracking-tight mb-3">Runtime safety behaviour</div>
|
| 1010 |
-
<ul class="space-y-3 text-sm text-slate-300">
|
| 1011 |
-
<li class="flex gap-2"><i data-lucide="help-circle" class="icon"></i> Clarifies when evidence is missing</li>
|
| 1012 |
-
<li class="flex gap-2"><i data-lucide="quote" class="icon"></i> Enforces citation-required answers</li>
|
| 1013 |
-
<li class="flex gap-2"><i data-lucide="shield-alert" class="icon"></i> Safety critic checks contraindications</li>
|
| 1014 |
-
<li class="flex gap-2"><i data-lucide="users" class="icon"></i> Designed for human-in-the-loop use</li>
|
| 1015 |
-
</ul>
|
| 1016 |
-
</aside>
|
| 1017 |
</div>
|
| 1018 |
-
</
|
| 1019 |
-
|
| 1020 |
-
|
| 1021 |
|
| 1022 |
<!-- Team -->
|
| 1023 |
<section id="team" class="section">
|
|
|
|
| 434 |
</section>
|
| 435 |
|
| 436 |
<!-- Overview -->
|
| 437 |
+
<section id="overview" class="section">
|
| 438 |
+
<div class="container">
|
| 439 |
+
<div class="grid lg:grid-cols-5 gap-8 items-start">
|
| 440 |
+
<!-- Left: narrative -->
|
| 441 |
+
<div class="lg:col-span-3" data-aos="fade-up">
|
| 442 |
+
<h2 class="text-3xl lg:text-4xl font-extrabold tracking-tight">Overview</h2>
|
| 443 |
+
<p class="mt-4 text-slate-300 leading-relaxed">
|
| 444 |
+
MedSwin frames clinical QA as an <span class="font-semibold text-slate-100">evidence-constrained decision pipeline</span>.
|
| 445 |
+
Every answer is gated by evidence sufficiency, bounded by a strict context budget, and accompanied by a
|
| 446 |
+
replayable trace suitable for audit and safety review.
|
| 447 |
+
</p>
|
| 448 |
+
|
| 449 |
+
<div class="mt-6 grid sm:grid-cols-3 gap-4">
|
| 450 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 451 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 452 |
+
<i data-lucide="message-square" class="icon"></i> Answer
|
| 453 |
+
</div>
|
| 454 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 455 |
+
Clinically phrased, uncertainty-aware output generated only when evidence gates are satisfied.
|
| 456 |
+
</p>
|
| 457 |
</div>
|
| 458 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 459 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 460 |
+
<i data-lucide="files" class="icon"></i> Evidence bundle
|
| 461 |
+
</div>
|
| 462 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 463 |
+
Compact EMR + guideline passages selected under token and diversity constraints.
|
| 464 |
+
</p>
|
| 465 |
</div>
|
| 466 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 467 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 468 |
+
<i data-lucide="route" class="icon"></i> Trace
|
| 469 |
+
</div>
|
| 470 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 471 |
+
Structured artifact log: retrieval, ranking, policies, safety checks.
|
| 472 |
+
</p>
|
| 473 |
</div>
|
|
|
|
|
|
|
|
|
|
| 474 |
</div>
|
| 475 |
</div>
|
| 476 |
+
|
| 477 |
+
<!-- Right: compact system summary -->
|
| 478 |
+
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 479 |
+
<div class="font-extrabold text-lg tracking-tight mb-3">Why MedSwin is different</div>
|
| 480 |
+
<ul class="space-y-3 text-sm text-slate-300">
|
| 481 |
+
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Refuses to answer when evidence is insufficient</li>
|
| 482 |
+
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Explicit EMR + CPG coverage requirements</li>
|
| 483 |
+
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Deterministic retrieval policies (no silent guessing)</li>
|
| 484 |
+
<li class="flex gap-2"><i data-lucide="check" class="icon text-emerald-300"></i> Local-deployable, auditable by design</li>
|
| 485 |
+
</ul>
|
| 486 |
+
</aside>
|
| 487 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 488 |
</div>
|
| 489 |
+
</section>
|
| 490 |
|
| 491 |
<!-- Contributions -->
|
| 492 |
<section id="contributions" class="section">
|
|
|
|
| 865 |
</section>
|
| 866 |
|
| 867 |
<!-- Retrieval -->
|
| 868 |
+
<section id="retrieval" class="section">
|
| 869 |
+
<div class="container">
|
| 870 |
+
<div class="grid lg:grid-cols-5 gap-8 items-start">
|
| 871 |
+
<div class="lg:col-span-3" data-aos="fade-up">
|
| 872 |
+
<h2 class="text-3xl lg:text-4xl font-extrabold tracking-tight">
|
| 873 |
+
Two-Stage Retrieval & Calibrated Reranking
|
| 874 |
+
</h2>
|
| 875 |
+
<p class="mt-4 text-slate-300 leading-relaxed">
|
| 876 |
+
Evidence selection is separated into recall-oriented candidate generation and precision-oriented reranking.
|
| 877 |
+
This avoids early truncation while enabling deterministic, policy-aware inclusion decisions.
|
| 878 |
+
</p>
|
| 879 |
+
|
| 880 |
+
<div class="mt-6 space-y-4">
|
| 881 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 882 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 883 |
+
<i data-lucide="scan-search" class="icon"></i> Stage 1 — Candidate generation
|
| 884 |
+
</div>
|
| 885 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 886 |
+
Dense ANN retrieval is unioned with BM25 to preserve rare clinical terms, abbreviations, and lab-specific phrasing.
|
| 887 |
+
</p>
|
| 888 |
</div>
|
| 889 |
+
|
| 890 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 891 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 892 |
+
<i data-lucide="badge-check" class="icon"></i> Stage 2 — Long-context reranking
|
| 893 |
+
</div>
|
| 894 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 895 |
+
A biomedical LLM reranker scores each passage and outputs calibrated probabilities usable as policy thresholds.
|
| 896 |
+
</p>
|
| 897 |
</div>
|
| 898 |
+
|
| 899 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 900 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 901 |
+
<i data-lucide="sliders" class="icon"></i> Policy-aware selection
|
| 902 |
+
</div>
|
| 903 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 904 |
+
Final selection enforces EMR + guideline sufficiency, diversity (MMR-style), and a strict token budget.
|
| 905 |
+
</p>
|
| 906 |
</div>
|
|
|
|
|
|
|
|
|
|
| 907 |
</div>
|
| 908 |
</div>
|
| 909 |
+
|
| 910 |
+
<!-- Right: sufficiency gate -->
|
| 911 |
+
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 912 |
+
<div class="font-extrabold text-lg tracking-tight mb-3">Evidence acceptance gate</div>
|
| 913 |
+
<ul class="space-y-3 text-sm text-slate-300">
|
| 914 |
+
<li class="flex gap-2"><i data-lucide="book-open-check" class="icon"></i> Required guideline recommendations present</li>
|
| 915 |
+
<li class="flex gap-2"><i data-lucide="file-heart" class="icon"></i> Patient-specific EMR signals included</li>
|
| 916 |
+
<li class="flex gap-2"><i data-lucide="layers" class="icon"></i> Redundancy reduced under budget</li>
|
| 917 |
+
<li class="flex gap-2"><i data-lucide="shield-alert" class="icon"></i> Safety critic approves synthesis</li>
|
| 918 |
+
</ul>
|
| 919 |
+
</aside>
|
| 920 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 921 |
</div>
|
| 922 |
+
</section>
|
| 923 |
+
|
| 924 |
+
<!-- Section spacer -->
|
| 925 |
+
<div class="h-20 lg:h-28"></div>
|
| 926 |
|
| 927 |
<!-- Training -->
|
| 928 |
+
<section id="training" class="section">
|
| 929 |
+
<div class="container">
|
| 930 |
+
<div class="grid lg:grid-cols-5 gap-8 items-start">
|
| 931 |
+
<div class="lg:col-span-3" data-aos="fade-up">
|
| 932 |
+
<h2 class="text-3xl lg:text-4xl font-extrabold tracking-tight">
|
| 933 |
+
Data, Training & Distillation
|
| 934 |
+
</h2>
|
| 935 |
+
<p class="mt-4 text-slate-300 leading-relaxed">
|
| 936 |
+
MedSwin’s deployable 7B model is trained for reliability rather than raw scale,
|
| 937 |
+
combining large-scale augmentation, supervised fine-tuning, and knowledge distillation.
|
| 938 |
+
</p>
|
| 939 |
+
|
| 940 |
+
<div class="mt-6 space-y-4">
|
| 941 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 942 |
+
<div class="font-extrabold">A · Data augmentation</div>
|
| 943 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 944 |
+
Paraphrasing, formatting variants, deduplication, and medical consistency checks expand coverage without semantic drift.
|
| 945 |
+
</p>
|
| 946 |
+
</div>
|
| 947 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 948 |
+
<div class="font-extrabold">B · Supervised fine-tuning</div>
|
| 949 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 950 |
+
Aligns the student to clinical instruction style, tone control, and structured answers.
|
| 951 |
+
</p>
|
| 952 |
+
</div>
|
| 953 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 954 |
+
<div class="font-extrabold">C · Knowledge distillation</div>
|
| 955 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 956 |
+
Hard labels expand task coverage; soft labels preserve calibration and uncertainty from a larger instructor.
|
| 957 |
+
</p>
|
| 958 |
+
</div>
|
| 959 |
</div>
|
| 960 |
</div>
|
| 961 |
+
|
| 962 |
+
<!-- Right: why KD -->
|
| 963 |
+
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 964 |
+
<div class="font-extrabold text-lg tracking-tight mb-3">Why distillation?</div>
|
| 965 |
+
<ul class="space-y-3 text-sm text-slate-300">
|
| 966 |
+
<li class="flex gap-2"><i data-lucide="cpu" class="icon"></i> Enables local inference on modest GPUs</li>
|
| 967 |
+
<li class="flex gap-2"><i data-lucide="shield-check" class="icon"></i> Preserves calibrated reasoning behaviour</li>
|
| 968 |
+
<li class="flex gap-2"><i data-lucide="settings" class="icon"></i> PEFT-friendly (LoRA / QLoRA)</li>
|
| 969 |
+
<li class="flex gap-2"><i data-lucide="lock" class="icon"></i> Institution-controlled deployment</li>
|
| 970 |
+
</ul>
|
| 971 |
+
</aside>
|
| 972 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 973 |
</div>
|
| 974 |
+
</section>
|
| 975 |
+
|
| 976 |
+
<!-- Section spacer -->
|
| 977 |
+
<div class="h-20 lg:h-28"></div>
|
| 978 |
|
| 979 |
<!-- Evaluation -->
|
| 980 |
+
<section id="evaluation" class="section">
|
| 981 |
+
<div class="container">
|
| 982 |
+
<div class="grid lg:grid-cols-5 gap-8 items-start">
|
| 983 |
+
<div class="lg:col-span-3" data-aos="fade-up">
|
| 984 |
+
<h2 class="text-3xl lg:text-4xl font-extrabold tracking-tight">
|
| 985 |
+
Evaluation & Safety
|
| 986 |
+
</h2>
|
| 987 |
+
<p class="mt-4 text-slate-300 leading-relaxed">
|
| 988 |
+
MedSwin evaluates clinical QA systems beyond answer accuracy, focusing on evidence quality,
|
| 989 |
+
guideline compliance, and runtime safety behaviour.
|
| 990 |
+
</p>
|
| 991 |
+
|
| 992 |
+
<div class="mt-6 grid sm:grid-cols-3 gap-4">
|
| 993 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 994 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 995 |
+
<i data-lucide="search" class="icon"></i> Retrieval quality
|
| 996 |
+
</div>
|
| 997 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 998 |
+
Evidence relevance and coverage under a fixed token budget.
|
| 999 |
+
</p>
|
| 1000 |
</div>
|
| 1001 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 1002 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 1003 |
+
<i data-lucide="book-open-check" class="icon"></i> Guideline coverage
|
| 1004 |
+
</div>
|
| 1005 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 1006 |
+
Presence of actionable recommendations and contraindications.
|
| 1007 |
+
</p>
|
| 1008 |
</div>
|
| 1009 |
+
<div class="rounded-2xl border border-white/10 bg-slate-950/40 p-4">
|
| 1010 |
+
<div class="font-extrabold inline-flex items-center gap-2">
|
| 1011 |
+
<i data-lucide="check-check" class="icon"></i> Faithfulness
|
| 1012 |
+
</div>
|
| 1013 |
+
<p class="text-sm text-slate-300 mt-2">
|
| 1014 |
+
Final answers remain grounded in cited evidence only.
|
| 1015 |
+
</p>
|
| 1016 |
</div>
|
|
|
|
|
|
|
|
|
|
| 1017 |
</div>
|
| 1018 |
</div>
|
| 1019 |
+
|
| 1020 |
+
<!-- Right: safety behaviour -->
|
| 1021 |
+
<aside class="lg:col-span-2 rounded-3xl border border-white/10 bg-slate-950/40 p-6" data-aos="fade-up">
|
| 1022 |
+
<div class="font-extrabold text-lg tracking-tight mb-3">Runtime safety behaviour</div>
|
| 1023 |
+
<ul class="space-y-3 text-sm text-slate-300">
|
| 1024 |
+
<li class="flex gap-2"><i data-lucide="help-circle" class="icon"></i> Clarifies when evidence is missing</li>
|
| 1025 |
+
<li class="flex gap-2"><i data-lucide="quote" class="icon"></i> Enforces citation-required answers</li>
|
| 1026 |
+
<li class="flex gap-2"><i data-lucide="shield-alert" class="icon"></i> Safety critic checks contraindications</li>
|
| 1027 |
+
<li class="flex gap-2"><i data-lucide="users" class="icon"></i> Designed for human-in-the-loop use</li>
|
| 1028 |
+
</ul>
|
| 1029 |
+
</aside>
|
| 1030 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1031 |
</div>
|
| 1032 |
+
</section>
|
|
|
|
|
|
|
| 1033 |
|
| 1034 |
<!-- Team -->
|
| 1035 |
<section id="team" class="section">
|