COVID-19 is the novel coronavirus that caused pandemic over the world. It has a high infection rate and it is urgent to diagnose many patients as we can to prevent the spread of the virus at the early stage. Rapid diagnostic tools development is urgent to save lives.
What it does
MantisCOVID is a cloud-based pre-diagnosis tool to be accessed from the internet. This tool delivers a rapid screening test by analyzing the X-ray Chest Radiography scans via Artificial Intelligence (AI) and it also evaluates the mortality risk of patients with the synthesis of the patient’s history by machine learning methods.
For rapid pre-screening objective, mantisCOVID is the offered solution via this proposal. The evaluation platform has two outputs after screening the group of patients as the prediction about the risk in COVID-19 via x-ray scans and the risk of the mortality rate. Two parameters can be used as a decision support system to direct the patient for the other highly sensitive tests. In the case where mantisCOVID cannot catch COVID-19 patient via AI elimination from x-ray scans, the physician can get an opposite decision via evaluating the risk of the mortality rate. If the patient has no symptoms and negative test results but have a higher risk of the mortality rate, it is strongly advised that the patient has to be followed with the other tests. Although forecasting of risk of the mortality rate is not easy due to dependence over many conditions, we have tried to estimate the mathematical model as linear regression in terms of 7 parameters via using the open datasets over the patients. The system works over the cloud as a web page and each physician can access to the system for the available computations. The system is designed to be scattered over millions of people to provide scalability via cloud.
The system has two types of forecasting such as the risk of the mortality rate and AI patient elimination. The methods used to implement the processing flow of forecasting the risk of the mortality rate is given in the following details.
Accomplishments that I'm proud of
15 seconds for the X-ray scan evaluation
What I learned
For certain diagnosis, this platform is not usable. A detailed test as PCR or CT is required for the diagnosis of Covid19. This test will be an eliminator such as labelling a 10000 patients over a million X-ray scans and the risk of the mortality rate can be computed to define the risks over the patient.
What's next for MantisCOVID : Rapid X-ray check platform
MantisCOVID is a cloud-based pre-screening platform that would be useful for the elimination of non-risky patients in the coronavirus pandemic. Since the algorithms are developed and evaluated using the open source data and platforms, this system needs to be evaluated via the clinical perspective by the legal authorities. This platform is an internet-based pre-diagnosis tool to be verified via the Food and Drug Agency. However, a physician would use this tool as an assistant to scan and diagnose the patients due to the rapid diagnosis requirement to prevent spreading of the coronavirus pandemic. In fact, the gold standards used to diagnose the covid19.
patients also have high false alarm rates due to the epidemiology of the disease is not fully known. As a future work, the clinical validation of the internet-based platform will start with the known legal authorities and other state of the art AI architectures will be used to increase the sensitivity and specificity of the platform. After the coronavirus pandemic, this platform would also be used for scanning the society in terms of the risk of the mortality rate and X-ray scans in any other illnesses and symptoms.