Team:Patras/System

System

The AI model, as implemented, depends on a variety of pre-installed libraries and a Python compiler. It would be of no practical use to apply such an advanced AI system if anyone needs to set up his personal computer to run this model properly. A PHP based platform has been developed for this reason. A user interface offers a simple and understandable point of view to the health professional using it. The procedure is divided into three (3) distinct steps:

1. Read the guidelines,
2. Upload the image
3. Get the results

Step-1: The user follows the guidelines through a carousel slideshow (Figures 1 & 2). The user can press the "Next" button anytime in case the guidelines have been read before.

Figure 1.

Figure 2.

Step-2: The user selects an image to load by clicking the corresponding button (Figure 3) that opens the file browser window. Then, the image is previewed in the application before uploaded to the system (Figure 4).

By clicking the "Get Results" button, the user uploads the image to the system, and he/she is redirected to the /results.php page.

Figure 3.

Figure 4.

Step-3: The results are presented to the user (Figure 5). The "dosing recommendation" button opens a model (Figure 6) showing the patient's genotype, and the dosage should be administrated by the physician. The "Upload another image" button redirects the user to the index page.

Figure 5.

Figure 6.

Gathering the images uploaded from every user assists in improving the AI model's accuracy. Consequently, every time a user clicks the upload button, the image is used to derive the desired result, uploaded and saved in a database in a longblob64 format (Figure 7) through an SQL query command.

Figure 7.

The image is practically saved to the database, and afterward, it is retrieved to be used by the Python code. A PHP command executes via shell the given input and returns the complete output as a string. We utilize this command to emulate the Python running in terminal with the image's name passed as an input. The output of the shell is the result presented to the user.

For the time being, the application is hosted in a private server with access to the public. Visit Haippocrates[dot]com to find our web-based application supported by an AI system.

However, the transition from private to a public server is vital for real-life utilization and can be easily achieved by installing the required packages needed for the server.