Skip to content

Flutter-based cross-platforms mobile application to streamline Lateral Flow Tests (LFTs) processing and management by forcing as minimum user interaction as possible and utilizing lightweight TFLite model.

Notifications You must be signed in to change notification settings

adityojulian/LFTrack

 
 

Repository files navigation

LFTrack: Instant, Accurate, and Secure LFT Results at Your Fingertips

Also visit LFTrack microsite for more info!

Project Description

This Flutter-based cross-platforms mobile application is designed to streamline the way Lateral Flow Tests (LFTs) are processed and managed by forcing as minimum user interaction as possible while maintaining user-friendly interface and intuitiveness. The app, through integrated light model (TFLite) and data management technologies, enables users to scan LFTs using their smartphone cameras, automatically interpret the results, and seamlessly store these results in a secure database.

From the moment the LFT is positioned within the camera's frame, the app initiates a countdown, performs the scan, interprets the result, and stores it—all without unnecessary steps or inputs from the user.

Key-Features

🌟 Automatic Detection and Scanning: Utilizes the smartphone camera to detect LFTs and automatically starts the scanning process.

⚡ Instant Result Interpretation: Employs integrated light-weight TFLite model to interpret test results accurately within seconds.

🔥 Direct Database Integration: Automatically uploads results to a secure Firebase database, ensuring data integrity and privacy.

📱 User-friendly Interface: Designed for ease of use, requiring minimal interaction from the user to complete the scanning process.

Screens

Instant scan with minimum interaction

Instant scan with minimum interaction

Notification dialog for scan errors

Notification dialog for scan errors

Search scanned LFT based on label and ID

Search scanned LFT based on label and ID

Filter based on date or date range

Instant scan with minimum interaction

Convert LFT results to CSV

Notification dialog for scan errors

Switch to Darkmode for better readability

Search scanned LFT based on label and ID

User Persona

This user persona is designed using Xtensio.

User Persona

Low Fidelity Mockup

Low-fidelity prototype image as a sequence of screens for the mobile application, detailing the user journey through various steps of the app's functionality.

Low Fidelity Mockup

Here's a description of each screen:

  1. Splash Screen: A simple screen with a logo in the center, serving as the initial loading screen for the app.
  2. On-boarding #1: Instructions for the user to place a lateral flow test (LFT) on a flat surface, with a "Next" button.
  3. On-boarding #2: Further instructions to position the LFT correctly, with "Previous" and "Next" navigation buttons.
  4. On-boarding #3: A prompt for the user to wait for a countdown to scan the LFT, with navigation buttons.
  5. No LFT: The main screen (landing) indicating that no test is present, with "Home" and "History" options.
  6. Out of Frame #1: A screen showing an incorrectly placed LFT that is out of frame.
  7. Out of Frame #2: Another example of an out-of-frame LFT, this time angled and partially out of view.
  8. LFT in Frame: The correct placement of an LFT within the frame for scanning.
  9. Countdown: A sequence of screens showing a countdown from 3 to 1, before the app scans the LFT.
  10. Successful Scan: A confirmation screen stating "New record added" after a successful scan.
  11. Failed Scan (Scanned): An error screen indicating the record is already in the database.
  12. Failed (Model Failed): An error screen stating "Failed to predict result," suggesting the app couldn't interpret the LFT result.
  13. History (Main): A screen showing the history of scans, with an "Export" feature and a search bar, along with a detailed list of previous scan results.
  14. History (Date): This screen allow the user to select a date and view the history of scans for that particular day.
  15. History (All Scans): Displays all scan results with the option to "Export" the data. The scans are labeled with outcomes like "Positive," "Negative," or "Invalid."
  16. Export Options: A screen dedicated to exporting data, with filters for date range and result type, along with a toggle for including invalid results.

Core Technologies

  • Flutter: The application is developed using Flutter, which enables a cross-platform development approach from a single codebase.
  • TensorFlow Lite: TensorFlow Lite is used for running the machine learning model responsible for LFT predictions.
  • Google ML Kit: Barcode scanning features are powered by Google ML Kit, facilitating accurate and efficient barcode recognition. Material 3: The app's user interface is designed following Material 3 guidelines to ensure a modern and cohesive look-and-feel.
  • Firebase Authentication: User authentication processes are handled using Firebase Authentication, ensuring secure access control.
  • Firestore: Firestore is the chosen database solution for LFTrack, allowing for real-time data synchronization and storage.
  • GetX: The GetX library is employed for state management, providing a robust solution for managing the app's state reactively.

Development Environment

$ flutter --version
Flutter 3.19.6 • channel stable • https://github.com/flutter/flutter.git
Framework • revision 54e66469a9 (5 days ago) • 2024-04-17 13:08:03 -0700
Engine • revision c4cd48e186
Tools • Dart 3.3.4 • DevTools 2.31.1

Android Emulator

✅ Google Pixel 6a

✅ API Version 34

✅ Harware Acceleration: On

Installation

Pre-requirements:

  • Install Flutter and Dart
  • Install simulation device or
  • Have a phone plugged to the main machine
$ git clone https://github.com/adityojulian/LFTrack.git
$ cd LFTrack
$ flutter pub get
$ flutter run

Contact Details

I'm happy to answer your questions and please feel free if you want to contribute to this project.

About

Flutter-based cross-platforms mobile application to streamline Lateral Flow Tests (LFTs) processing and management by forcing as minimum user interaction as possible and utilizing lightweight TFLite model.

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Dart 74.9%
  • C++ 12.5%
  • CMake 9.9%
  • HTML 1.0%
  • Swift 0.9%
  • C 0.7%
  • Other 0.1%