NuMed

Your personal AI-based COVID-19 detection. Made using Mendix.

In this project, we propose a solution to handle the difficulties of the doctors and health practitioners during the pandemic by the development of a telemedicine application. This application can be used remotely by doctors and also patients or general users who live far away from a health facility, more specifically a COVID-19 referral hospital.

Together with the advancement of telemedicine, Artificial Intelligence (AI) technology can work as a complement of the telemedicine system to contribute to the automation of the disease modeling. We specifically want to focus on the explainable power of the AI. Extra care need to be consid- ered when working with the medical field as it corresponds to human life. The doctors expertise cannot be neglected. The eXplainable AI (XAI) is the way to make the decision-making process transparent and easily being validated by the experts. Two well-known methods in AI, namely Deep Learning and tree based machine learning can be used to build XAI and better explain the data collected by using telemedicine application. There are many methods to diagnose COVID-19. The most widely used method is nucleic acid-based testing such as Reverse-Transcriptase Polymerase Chain Reaction (RT-PCR). This method is difficult to be done remotely, because it needs specific laboratory and operator to process the sample. According to the recent literature, there are other methods that can also be used effec- tively to diagnose the presence of the COVID-19 disease. The three of them are diagnosing from chest X-ray (CXR), chest Computerized Tomography (CT-scan) and blood sample. These data can be easily generated by local hospital with a laboratory and/or a radiology department. Our proposed telemedicine system will be able to read from these three forms of data and return the predicted result to the users.

The proposed solution is not only building the intelligent telemedicine application, but also allowing the expertise to validate the result by seeing the AI model visualization. This feature will become an advantage of our proposed solution when compared to the other existing telemedicine app in the market.

In this project, I built the frontend using Mendix and connected it to the backend services using REST API.