Voice · Ask Liya

Case Study 01 · Healthcare AI

Visit Live Site

Healium Sono

AI-powered remote ultrasound and teleguidance for cardiology and OB/GYN

Dereje Seifu4 min read

Intro

Client: Healium Intelliscan Corporation

Industry: HealthTech / Telemedicine

Location: United States

Role: Lead Full-stack engineer responsible for architecture, AI service integration, real-time video, and deployment.

What Made This a Good Bet

Specialist cardiologists and OB/GYN physicians are concentrated in urban centers, while rural clinics often have ultrasound hardware but no real-time interpretation expertise.

Patient transfer and delayed referral increased clinical risk and operational costs.

The platform required remote specialist guidance in real time, with AI-assisted analysis running in parallel.

Misread or delayed OB/GYN ultrasound can lead to missed fetal abnormalities or undetected high-risk pregnancies.

Cardiology patients in low-resource settings often waited days for specialist review, delaying life-saving interventions.

Traditional teleconsultation tools offered video, but lacked a specialized guidance layer or AI decision support.

The target market included dozens of rural clinics with no on-site specialist access.

What I Built

Built a full teleguidance workflow with encrypted low-latency specialist sessions.

Integrated Zoom Video SDK for specialist-tech collaboration with live annotation and voice instruction.

Created an AI inference pipeline with FastAPI on AWS Lambda for structured ultrasound findings.

Implemented secure storage and clinic-level data isolation with Supabase Row-Level Security.

Solved timestamp drift and unstable network handling with session recovery and adaptive quality.

The Stack

Frontend
Next.js (React)
Backend / API
FastAPI (Python)
Database
Supabase (PostgreSQL + RLS)
Real-time Video
Zoom Video SDK
AI Inference
AWS Lambda + FastAPI
Cloud
AWS (Lambda, S3, API Gateway)
Auth
Supabase Auth
Storage
Supabase Storage / S3

Compliance & Data Security

Patient data encrypted at rest and in transit (TLS 1.3 + AES-256).

Compliance standards met: HIPAA and local health data regulations.

Audit logging: All specialist actions and session modifications logged for clinical accountability.

Beyond The Headline Metrics

Remote specialists guided 1,200 sessions across 45 clinics in the first year.

Average scan-to-specialist finding time reduced from 48 hours to 15 minutes.

92% of flagged cases were confirmed by subsequent specialist review.

Physician feedback indicates high reliability in low-bandwidth rural environments.

Other case studies

See all case studies