Nexar
Population health intelligence — RAPIDS GPU analytics + NeMo predictive risk scoring over encounter data at health-system scale.
Explore NexarScribara is a clinical intelligence platform built on GPU-accelerated AI — population health, chronic care monitoring, synthetic data, revenue analytics, and clinical trial matching, all sharing one encounter corpus and one NVIDIA compute backbone.
Five products · one shared clinical record · one GPU-accelerated backbone
Each product solves a distinct, high-value problem. Together they form a compounding clinical intelligence ecosystem.
Population health intelligence — RAPIDS GPU analytics + NeMo predictive risk scoring over encounter data at health-system scale.
Explore NexarReal-time chronic care monitoring — Holoscan sensor streaming + Jetson edge inference + NeMo exacerbation prediction between visits. [ASPIRATIONAL]
Explore VigilSynthetic clinical data — NeMo generative LLM + Clara/MONAI diffusion models for privacy-safe specialty encounter and imaging datasets. [ASPIRATIONAL]
Explore AugurRevenue intelligence — RAPIDS GPU denial-risk scoring and NeMo undercoding detection on every claim before submission.
Explore RevixClinical trial matching — NeMo Retriever semantic matching + RAPIDS GPU similarity connects specialty patients to eligible active trials.
Explore OrbisAll five products share one encounter corpus, one NVIDIA compute substrate, and one clinical audit trail — so each product compounds the others.
See the platformRAPIDS GPU cohort analysis over 100M+ encounter records in under 2 minutes. NeMo predictive models score readmission risk and care gaps at health-system scale.
Holoscan sensor streaming + Jetson edge inference detects biometric anomalies in under 500 ms — alerting clinicians 24–72 hours before an acute event. [ASPIRATIONAL]
NeMo generates statistically faithful synthetic specialty encounters; Clara/MONAI synthesizes clinical images. Every sample validated by a clinical accuracy classifier. [ASPIRATIONAL]
NeMo denial-risk classifier scores every outgoing claim in under 100 ms. RAPIDS scans a 5-year, 10M-claim history for systematic denial patterns in under 5 minutes.
NeMo Retriever + RAPIDS cuML matches 100k+ patients against 200+ active trials in under 2 minutes — the workload takes 90 minutes on CPU.
Every product trains and serves on the same accelerated-compute substrate.
Clinical AI model training, RAG retrieval, and GPU-accelerated data pipelines.
Specialty medical-imaging models — training and synthetic image generation.
Low-latency, multi-model inference serving for all five products.
HIPAA on-prem deployment — PHI stays inside the health system boundary.
Training-grade compute for Augur's generative models and Vigil's predictors.
Edge inference and real-time sensor streaming for Vigil's physical AI layer.
Outcomes are reported by design partners and marked illustrative [PLACEHOLDER].
"Nexar showed us which patients were slipping through — that's revenue and outcomes in one dashboard."
"Revix paid for itself in the first quarter — just from codes we were systematically leaving on the table."
Healthcare buys on trust. Scribara is built to clear procurement, security review, and compliance.
PHI handled under a Business Associate Agreement.
[ASPIRATIONAL] — roadmap to Type II within 12 months.
Run owned models inside your datacenter via NVIDIA AI Enterprise. [ASPIRATIONAL]
Every agent action logged, reversible, tied to the clinician's signature.
US, EU, UK regions; per-tenant keys (BYOK option). [ASPIRATIONAL]
AI management-system conformity on the roadmap. [ASPIRATIONAL]
We'll map the right product to your specialty and your highest-value workflow.