As the number of COVID-19 deaths worldwide reaches 408 000, the coronavirus continues its rapid global spread. Healthcare resources are under pressure. When it comes to testing patients for COVID-19, some countries, such as France, are so short on test supplies that they test the patients with substantial breathing difficulties solely.
As a consequence, the identification and prioritization of individuals most at-risk of COVID-19 death have become a critical challenge. In some places, such as Lombardy, it is a tightrope walk to handle the influx of severe patients among the people seeking medical advice. On the one hand, medical resources are scarce. They include doctors, nurses, tests as well as hospital beds. On the other hand, it is mandatory to examine patients to issue a diagnostic. Yet considering everybody might mean not being reactive enough for those who need it the most.
Some contaminated patients develop a severe form of the illness, and sometimes COVID-19 death, while others show no symptoms. The contamination is considered a binary phenomenon: individuals are either exposed or unexposed, infected or uninfected, symptomatic patients or asymptomatic carriers. However, there might be grey areas to explore, in which identifying potential COVID-19 deaths, treatment opportunities, or even new approaches to containment might exist. This approach is a qualitative one versus the common quantitative one used so far. This quantitative approach pertains, for instance, to the viral load of a patient. It also encompasses not only whether a patient is overweight but also on his/her actual body mass index (BMI). Epidemiologists, doctors, and researchers have gathered such data with regards to COVID-19 development among infected and noninfected patients. TADA, the Small Data Machine Learning tool from MyDataModels, is ideally suited to process it. These data usually belong to the category of Small Data because they are often a few hundred, or at best, a few thousand records. TADA can learn from these few records and predict which new patients are the most at risk of developing a severe form of the disease, including hospitalization, intubation, and lethality.
For instance, smoking might be considered a risk-reducing factor according to a French study, yet it does not affect patients in a Mexican one. In this recent research, diabetes and obesity do. In a few clicks, with no specific data science training and actually without any training, epidemiologists can use TADA to isolate prevalent criteria among these hundreds or thousands of data.
Blood type is said to have an impact on contracting the virus; the next question is whether it impacts those people becoming severely ill, or even those who suffer COVID-19 death. Again, in a few clicks, a reliable prediction can be obtained using TADA. American epidemiologists suspect race to be a discriminating factor in contracting the most severe forms of the disease. Heart disease might increase the risk that a person with COVID-19 will experience a severe infection. In a few minutes, TADA can provide a fair idea of the impact of race and heart disease on lethality.
MyDataModels gives a self-service solution to healthcare professionals who need to identify the patients at risk of COVID-19 death.
“TADA can anticipate which patients are most likely to develop a dangerous form of COVID-19.”
Understanding where the health risks lie might help protect, for instance, healthcare workers on the front lines of the COVID-19 pandemic, moreover, if the viral load exposure becomes a decisive factor. Several young and robust healthcare workers have suffered COVID-19 death. We might keep track of their total exposure, and put in place viral-dosimetry controls so that one individual can avoid repeated interactions with some set of highly contagious patients. It might help save hospital beds for people at higher risks. In a nutshell, it might help save lives.