Healios
Interactive Clinical AI Showcase
V2 Architecture

Production-Ready Clinical AI & Telemetry Platform

This showcase enables engineers, clinical leads, and recruiters to test our end-to-end multimodal systems: in-browser Edge CV gatekeeping, Grad-CAM saliency heatmaps, color-space tissue morphology, and our EHR-integrated ERAS recovery agent.

1. Multi-Tenant Clinical Persona Switchboard

Select a persona to immediately recalibrate the entire portal’s recovery day, surgical notes, vitals history, and medication courses.

Day 5 Post-Op
warning

Elena Rostova

Laparoscopic Appendectomy

Abdominal surgery recovery with low-grade fever and mild incision erythema.

Rx: Amoxicillin-Clavulanate 875mgACTIVE
Day 11 Post-Op
nominal

Marcus Vance

Total Knee Arthroplasty (Left)

20cm midline incision healing on schedule with active DVT blood-thinner prophylaxis.

Rx: Enoxaparin (Lovenox) 40mg SCSELECT →
Day 2 Post-Op
critical

Sophia Chen

Robotic Cholecystectomy

Acute Day 2 recovery with 101.1°F fever spike and severe incision tenderness.

Rx: Cefazolin 500mg + Zofran 4mgSELECT →

2. Computer Vision & Explainable AI (Grad-CAM) Sandbox

Test the automated diagnostic pipeline using preloaded high-fidelity clinical wound cases. Demonstrates browser Edge CV gatekeeping, convolutional inference, Grad-CAM saliency heatmaps, and HSV color-space tissue segmentation.

Normal
ACTIVE TEST

Sample A: Nominal Post-Op Incision

Clean, well-approximated wound edges with minimal reactive erythema and healthy granulation bed.

Laparoscopic Appendectomy Select Case →
Warning

Sample B: Early Periwound Erythema & Slough

Noticeable 4mm erythema margin with localized fibrinous yellowish slough along the inferior staple line.

Total Knee Arthroplasty Select Case →
Critical

Sample C: Acute High-Risk Dehiscence / Infection

Spreading fiery erythema > 10mm with non-viable dark eschar and tension along margin.

Robotic Cholecystectomy Select Case →

Multimodal Computer Vision Diagnostic Sandbox

Testing Sample A: Nominal Post-Op Incision (Laparoscopic Appendectomy (Port 1))

Ready for Multimodal Neural Inference

Triggers Edge CV validation, deep convolutional classification, Grad-CAM saliency heatmaps, and HSV color-space tissue morphology.

3. Clinical ERAS Hybrid RAG & Diagnostic EHR Agent Playground

Our multi-turn recovery agent inspects the active patient’s real-time EHR telemetry (vitals, medication adherence, wound score) and grounds answers in validated surgical protocols.

Pre-Configured Clinical Scenarios (Click to Execute)

Currently Consulting for: (Day 4 • )

4. Verified Systems Engineering Benchmarks

Grad-CAM Latency
8.09 ms

NumPy Jet colormap (No matplotlib locks)

Tissue Segmentation
2.00 ms

Vectorized HSV granulation & slough

RAG Retrieval
0.14 ms

Hybrid BM25 + TF-IDF cosine similarity

Edge Quality Gate
< 35 ms

HTML5 Canvas Laplacian blur filter