TED AI Hackathon Winner 2024 Portfolio Project Human-Centered AI

AI4Echolalia™

A human-centered AI concept exploring parent-mediated educational support for echolalia.

Classifies immediate vs delayed repetition patterns · Parent-directed coaching prompts · Not a medical device

Back to Portfolio
14 Day learning plan
2 Echolalia types detected
0 PHI stored after session
UN SDGs 3, 4 & 10 aligned
Project Context

The concept was designed around adults who support children with echolalia.

Parents & Guardians Speech-Language Pathologists Special Education Teachers Healthcare Providers Researchers
Historical Portfolio Work

AI4Echolalia is presented here as prior product and hackathon work, demonstrating DLYog’s approach to human-centered prototyping, safety, and privacy.

Prototype Scope

The project explored a home-based practice concept guided by AI and designed around the parent–child relationship.

Echolalia Detection

Classifies immediate vs delayed echolalia patterns in real-time voice sessions, giving parents actionable context about their child's communication.

AI Coaching Prompts

Generates parent-directed prompts based on session context. No anthropomorphic AI persona — the parent is always the coach.

14-Day Learning Plan

Structured daily practice schedule adapted to your child's patterns. Progress tracked across sessions with visual dashboards.

Observe Mode

Record and analyze longer natural conversations. AI identifies echolalia patterns and suggests follow-up strategies with SLP-informed context.

Privacy-First

Audio is never stored after transcription. Zero PHI retention. FERPA-style data controls. Built for families who trust us with sensitive moments.

Native PWA

Installable as a native app on iOS and Android. Works as a home screen app with offline capability — no app store required.

Prototype Flow

The concept organized practice around three simple steps.

1
Start a Session

Open the app, choose Practice or Observe mode. The AI listens only during the active session — nothing is stored beforehand.

2
AI Analyzes Patterns

Speech is transcribed and classified. The AI identifies whether echolalia is immediate or delayed and provides coaching context for the parent.

3
Review & Progress

View session insights, track progress on your 14-day plan, and follow up with your SLP using session summaries. No PHI is retained after the session.

Common Questions
Is this a medical device?
No. AI4Echolalia™ is positioned as an educational coaching tool only. It does not diagnose, treat, or prevent any medical condition. We are actively reviewing FDA/FTC regulatory frameworks and will comply with any applicable requirements as determinations are made.
Who is behind this?
AI4Echolalia™ was built by DLYog Lab Research Services LLC and received second place at the TED AI San Francisco 2024 Hackathon. It is now presented as historical portfolio work.
What privacy principles shaped the project?
The prototype explored real-time transcription without retaining audio, data minimization, and restricted handling of sensitive information.
Is the project currently offered as a flagship product?
No. This page preserves the project history and technical concepts as part of DLYog Lab’s portfolio.
What form factor did the prototype explore?
AI4Echolalia™ was developed as a Progressive Web App (PWA), allowing the team to explore an installable mobile experience without an app-store dependency.

Have a difficult workflow for AI?

DLYog now applies the lessons from projects like this to custom AI engineering, agentic automation, and private AI systems.

Discuss a Project Explore Services

Historical portfolio project · Educational concept · Not a medical device