ACE (Autism Care Ecosystem) Mobile

An advanced, AI-driven mobile application for Autism spectrum disorder management, screening, and therapy.

Developed by Aditya Mishra, Khushneet Singh, and Akarsh Solanky

---

📖 Overview

ACE Mobile is a comprehensive, intelligently designed Flutter application built to support the entire autism care triad: Children, Parents/Caregivers, and Medical Professionals.

By blending gamified therapy exercises, on-device machine learning for behavioral assessment, and large language models (LLMs) for personalized guidance, ACE bridges the gap between clinical visits and at-home developmental support.

📸 See the full visual walkthrough: APP_FLOW.md — screenshots of every screen in the app.

🏗️ System Architecture Diagram: docs/system_architecture.html — open in a browser for a high-resolution, presentation-ready architecture overview.

---

🎥 Video Explanation

VIDEO-2026-03-01-23-05-29.mp4

---

✨ Core Features & Capabilities

The ACE ecosystem is divided into distinct, role-based experiences powered by advanced technologies.

🧠 1. AI-Driven Screening & Diagnostics

• M-CHAT AI Interviewer: We've digitized the standard Modified Checklist for Autism in Toddlers (M-CHAT). Instead of static forms, parents interact with an AI interviewer that dynamically asks the 20 M-CHAT questions, scoring responses in real-time to generate an early-risk assessment report.
• Emotion Assessment (Google ML Kit): A behavioral screening tool where children watch emotionally evocative visual stimuli. Using the device's camera and google_mlkit_face_detection, the app measures 46 facial landmarks (smiling probability, eye openness) to detect atypical empathy responses or reduced emotional congruence.
• Eye Contact Gamification: The "Butterfly Exercise" encourages children to maintain eye contact with the screen. ML Kit tracks gaze vectors and face angles, rewarding sustained attention with visual feedback.
• Physical Imitation (TensorFlow Lite - MoveNet): A motor-skills diagnostic tool. The app uses movenet.tflite (a lightweight pose estimation model) to track 17 key body joints at 30+ FPS. Children mimic on-screen poses (e.g., raising arms, clapping), and the app calculates the cosine similarity between the child's pose and the target pose, assessing gross motor planning capabilities.

💬 2. Context-Aware AI Chat Assistant

• Role-Based Personas: An LLM-powered assistant (housed in features/AI_Chat_Assistant) that knows exactly who it's talking to.
  • For Parents: Acts as a supportive, empathetic pediatric advisor, using the child's name, age, and recent diagnosis data.
  • For Doctors: Acts as a clinical data retrieval assistant, providing quick summaries of patient metrics and literature.

🧘‍♀️ 3. Therapy & Grounding Tools

• Breathing Pacer: A fully animated 4-2-6 breathing cycle guide (4s inhale, 2s hold, 6s exhale) with expanding/contracting visual cues and phase-aware colors.
• 5-4-3-2-1 Sensory Grounding: Built-in interactive grounding cards to bring a child back to the present moment during high anxiety.
• Meltdown Prediction Engine: The TherapyScreen features a "Live Monitor" dashboard designed to ingest physiological data (simulated/wearable integration like heart rate and skin conductance) and establish daily patterns. It predicts incoming emotional dysregulation before a full meltdown occurs.

👨‍⚕️ 4. Doctor Dashboard & Patient Management

• Patient Roster: Doctors have a dedicated dashboard summarizing their assigned patients, sorted by risk level (high → medium → low → pending).
• Summary Statistics: Live-computed cards showing total patients, high-risk flags this week, sessions completed today, and average assessment scores.
• Interactive Progress Graphs: Patient detail screens feature fl_chart LineCharts with color-coded trend lines per session type, date-based axes, inline legends, and touch-interactive tooltips.
• Formatted Session Metrics: A type-aware SessionMetricsCard widget renders human-readable metrics for each session type (M-CHAT flags, pose match %, gaze scores, dominant emotions) instead of raw JSON.
• Prescriptive Therapy Plans: Doctors can assign therapeutic tasks (e.g., "Complete Eye Contact for 30s") with titles, descriptions, and due dates. Actions sync to the parent's home screen in real-time via Supabase Realtime channels.
• Clinical Notes: A doctor-to-parent messaging system allowing broadcast or targeted clinical notes.
• Join Code Linking: Parents share a unique join_code with their doctor to securely establish a clinical relationship.

📈 5. Progress Tracking & Community

• Progress Dashboard: Parents can view a historical log of all completed assessments and therapy sessions with milestone tracking.
• Secure Cloud Sync: Game scores, behavioral metrics, and LLM-generated summaries are synchronized in real-time to Supabase (sessions table) with strict matching via unique parent-child constraints.
• Community Forum: A social space where parents and caregivers share experiences, victories, and peer support.
• Relational Architecture: View our Database Schema for details on how profiles, children, sessions, therapy_plans, therapy_actions, and clinical_notes interact.

---

🗺️ User Flows & Navigation

ACE Mobile uses a robust routing architecture initialized in main.dart, handling three distinct user journeys:

Onboarding Flow

1. Splash Screen (/splash): A visually rich, animated entry point featuring drifting glow orbs, a particle field, and a segmented loading bar. Behind the scenes, it pre-loads ProfileProvider data from SharedPreferences.
2. Authentication (/login): Email/Password and Google Sign-In powered by Supabase Auth (AuthService). Includes robust error handling with user-friendly messages for all Supabase exceptions.
3. Role Selection (/role_selection): First-time users declare their role (Parent/Caregiver vs. Doctor/Therapist). Gracefully handles and displays errors during profile initialization.

Parent / Caregiver Flow

1. Home Dashboard: The central hub. Displays a "Good Morning, [Name]" header, today's therapy action goals (synced from doctor in real-time), and quick-launch buttons for the child's daily exercises.
2. Therapy Tab: Access to the breathing pacer, grounding exercises, and the ACE AI Chat assistant.
3. Assessments: Launch the M-CHAT AI, Eye Contact game, Pose Imitation, or Emotion Assessment. Results are pushed to Supabase for the doctor to review.
4. Progress: Historical view of all completed sessions and milestones.
5. Community: A forum view for parents to connect, share victories, and seek peer support.

Doctor / Professional Flow

1. Doctor Dashboard: A macroscopic view of all connected families with summary statistics (total patients, risk flags, daily sessions, average scores).
2. Patient Roster: Searchable list of assigned patients, sorted by risk level with recent activity indicators.
3. Patient Details: Clicking a patient reveals session history, interactive progress graphs (fl_chart), formatted session metrics, and risk flags.
4. Therapy Plan Management: Assign daily action checklists, adjust therapy levels, and toggle action completion — all reflected on the parent's home screen within 3 seconds via Supabase Realtime.
5. Clinical Notes: Send broadcast or targeted messages to parents.
6. Doctor Profile: View and manage account information and settings.

---

🏗️ Architecture & Directory Structure

The application utilizes the Provider pattern for reactive state management, adhering to a Feature-First folder structure with a dedicated service abstraction layer.

ace_mobile/
├── android/ & ios/         # Native platform configurations (Permissions, Info.plist)
├── assets/
│   ├── images/             # appLogo.png, UI assets
│   └── models/             # movenet.tflite (On-device pose estimation model)
│
└── lib/
    ├── core/               # App-wide constants
    │   └── constants.dart  # Colors (appColors), typography, theme definitions
    │
    ├── backend/            # Centralized cloud services
    │   ├── supabase_client.dart  # .env based Supabase initialization
    │   ├── supabase_service.dart # Legacy business logic (Profiles, Children, Sessions)
    │   ├── backend.dart          # Barrel exports
    │   └── services/             # Dedicated service layer
    │       ├── auth_service.dart           # Supabase Auth (Email/Password, Google, session management)
    │       ├── profile_service.dart        # User profile CRUD
    │       ├── child_service.dart          # Child records & join code management
    │       ├── session_service.dart        # Assessment session persistence & retrieval
    │       ├── therapy_service.dart        # Therapy plans & action items (Realtime sync)
    │       └── clinical_notes_service.dart # Doctor-to-parent clinical messaging
    │
    ├── shared/             # Reusable UI components (buttons, custom app bars)
    │
    ├── features/           # Distinct domain modules
    │   ├── splash/         # Animated entry screen
    │   ├── auth/           # Login UI, AuthWrapper, Role Selection
    │   ├── profile/        # ProfileProvider: User data, role, current child data
    │   ├── onboarding/     # Multi-page app tutorial walkthrough
    │   │
    │   ├── dashboard/      # Parent navigation shell
    │   ├── HomeScreen.dart  # Parent home dashboard (goals, quick actions)
    │   │
    │   ├── assessment/     # M-CHAT AI logic and AssessmentProvider
    │   ├── emotion_assessment/ # Emotion detection (google_mlkit_face_detection)
    │   ├── eye_contact/    # Camera overlay and gaze tracking (EyeContactProvider)
    │   ├── imitation/      # TFLite MoveNet vision logic (ImitationProvider)
    │   │
    │   ├── Therapy/        # Breathing pacer, grounding, live monitor
    │   ├── AI_Chat_Assistant/ # Chat UI (chat_bubble.dart), OpenRouter LLM integration
    │   ├── progress/       # Progress tracking dashboard
    │   ├── community/      # Social feeds and list views
    │   │
    │   └── doctor/         # Clinician-only module
    │       ├── doctor_dashboard_provider.dart  # Live stats, patient fetching, Realtime subscriptions
    │       ├── clinical_notes_provider.dart    # Clinical notes state management
    │       ├── doctor_bottom_navbar.dart       # Doctor navigation shell
    │       └── screens/
    │           ├── doctor_dashboard_screen.dart    # Summary stats, recent patients
    │           ├── doctor_patients_screen.dart     # Full patient roster (risk-sorted)
    │           ├── patient_detail_screen.dart      # Session history, progress graphs, metrics
    │           ├── doctor_therapy_plan_screen.dart  # Prescriptive action plan management
    │           ├── doctor_progress_screen.dart     # Aggregate progress analytics
    │           └── doctor_profile_screen.dart      # Doctor profile & settings
    │
    └── main.dart           # Entry point: Supabase init, Provider injection, Theme setup

---

⚙️ Technical Specs & Toolkit

• Framework: Flutter (Dart, SDK ^3.11.0)
• State Management: provider: ^6.1.2
• Backend Services: Supabase Auth & Database (supabase_flutter) with Realtime channels, Google Sign-In
• Machine Learning (Edge computing):
  • google_mlkit_face_detection: ^0.13.2 — 46 facial landmark tracking, eye open probabilities, smile probabilities, gaze vector analysis.
  • tflite_flutter: ^0.11.0 — Running the MoveNet single-pose lightning model natively via Android NNAPI / iOS CoreML delegates for 30+ fps pose tracking.
  • camera: ^0.11.4 — High-speed frame extraction for ML processing.
• Charts: fl_chart — Interactive line charts for progress visualization with per-session-type trend lines, tooltips, and legends.
• UI/UX:
  • google_fonts: ^8.0.2 (Poppins, Space Grotesk, DM Mono)
  • flutter_animate: ^4.5.2 (Micro-interactions)
  • persistent_bottom_nav_bar: ^6.2.1
• Database: PostgreSQL via Supabase — 6 relational tables: profiles, children, sessions, therapy_plans, therapy_actions, clinical_notes. Full schema in docs/DATABASE_SCHEMA.md.

---

🛠️ Performance & Size Optimizations

The app is heavily optimized for distribution and edge-ML performance:

• Minification: R8 code shrinking and resource shrinking (isMinifyEnabled) are configured in Android/app/build.gradle.kts.
• Targeted Dependencies: The bulky google_ml_kit umbrella package has been stripped in favor of the specialized google_mlkit_face_detection to keep native library (.so) payload minimal.
• ABI Splits: Separate binaries are built for arm64-v8a and armeabi-v7a drastically reducing the final APK/AppBundle size.
• Custom ProGuard: Keep rules injected for ML Kit and TFLite GPU delegates to ensure runtime stability post-minification.
• Resource Lifecycle Hardening: Every provider implements explicit dispose() overrides to cancel timers, stream subscriptions, and prevent resource leaks.
• Zero Static Analysis Warnings: Codebase passes flutter analyze with zero warnings — no unused imports, no unnecessary casts, all print() calls replaced with debugPrint().

---

💻 Local Development Setup

To run ACE Mobile locally, you must provide your own API keys for Supabase and OpenRouter via a .env file. These keys are deliberately excluded from source control.

1. Create a file named .env inside frontend_app/ace_mobile/.
2. Add the following keys (replace with your actual credentials):

GENAI_KEY=sk-or-v1-...
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_ANON_KEY=eyJhbGciOi...

The application will throw an initialization error if these keys are missing.

---

📸 Screenshots

For a comprehensive visual walkthrough of every screen in the app, see APP_FLOW.md.

---

🤝 Contributing

We welcome contributions to ACE Mobile! To contribute:

1. Fork the repository.
2. Create a new branch (git checkout -b feature/your-feature).
3. Commit your changes (git commit -m 'Add some feature').
4. Push to the branch (git push origin feature/your-feature).
5. Open a Pull Request.

Developed with ❤️ by Aditya Mishra, Khushneet Singh, and Akarsh Solanky.

---

📜 License

This project is licensed under the Apache License 2.0. See the LICENSE file for the full text.

---

Empowering families, connecting professionals, and supporting neurodiversity through technology.