How iamvibhav/autisense works
Main Components Based on Project Type
• Since AutiSense is an ML-driven healthcare tool with a focus on privacy, its components include:
• Data Source: The dataset (i.e., the CSV in the “dataset” directory) that holds the Autism Screening data.
• Model Artifacts: The “models” directory contains the trained AdaBoost model, scalers, and label encoders.
• Core Processing Logic: The “neura.py” (and its notebook “neura.ipynb”) serve as the core machine learning module that handles data preprocessing, model inference, and possibly training routines.
• Web Interface: Although not present in the provided file tree, the README indicates an “app.py” that implements a Gradio-based UI for doctors, patients, and researchers.
• Security & Privacy Layer: As described in the README, the system leverages access control (RBAC), encryption, and perturbation techniques to protect patient data.
Analyzing the File Structure
• Top-Level Files/Folders:
• README.md – Contains full project description, architecture overview, and instructions.
• dataset/ – Holds the raw dataset (CSV) used for predictions.
• models/ – Includes the saved model artifacts (model pickle, scaler, label encoders).
• neura.py & neura.ipynb – Provide the ML logic and data handling routines.
• requirements.txt – Lists dependencies required by the project.
• These elements indicate a separation of concerns: data management (dataset), model management (models), and processing/interaction (neura scripts and the Gradio UI as mentioned in the README).
• Web UI Component (Gradio Interface): Handles input from doctors/patients and displays predictions.
• Backend Processing Module (ML Engine): Represented by the “neura.py” component that loads the model artifacts.
• Data Storage/Management: Represents the dataset files.
• Model Artifacts Repository: Contains trained model files and encoders.
• Security and Privacy Module: Although not in its own file, it’s an essential cross-cutting concern enforcing RBAC, encryption, perturbation, and secure access based on the README.
Architectural Patterns and Design Principles
• Modular Design: Highlight that changes or updates can be made independently in the ML engine or the UI.
• Security by Design: Include a dedicated side-panel or an overlay that shows how RBAC, encryption, and data perturbation mechanisms integrate across modules.