COVID Symptoms Can Be Uncovered By New Machine-Learning Tool

Unlike the old days, man has changed a lot in terms of technology and lifestyle. He relies on various machines for his day-to-day livelihood and this even includes transportation and communication. These machines help a lot in solving most of the issues facing these days and putting more effort into building other inventions to make his daily life burden free. 

With the arrival of artificial intelligence software, people nowadays won’t even have to think twice before doing anything as they could just ask the screen in front of them regarding anything they need to know even if it is about anyone anywhere in the world. And won’t this be of any help if we use this intelligence in health care?

Lately, there has been news spreading that artificial intelligence software has been utilized in order to extract insights from health records. Similarly, they were also utilized to shed light on chronic and ever-lasting symptoms of the plague that has been sticking to the planet for the past two years. 

Machine Learning Tool And Long COVID Symptoms

Numerous people who suffered from this deadly viral infection are still suffering from a condition called long COVID. It’s a health condition in which individuals who were once diagnosed with the virus will experience some kind of chronic symptom from their initial days. This condition has made the situation more like a pandemic within a pandemic. And in order to prevent such a deadly situation, scientists are now using machine learning to find out the real reason why some group suffers or develops such debilitating long-lasting symptoms. 

Lately, a group of researchers from the United States has invented some machine learning tools which will analyze electronic health records or EHRs. This will work together to identify common symptoms and define subtypes of long COVID. Earlier, another research team published a study in biomedicine that states that there are strong correlations between long covid subtypes and pre-existing conditions such as variations in blood glucose levels and stress and anxiety issues. 

Justin Reese, a Computer Research Scientist at Berkeley Lab’s Biosciences Area, explained that with this research, they could understand the current condition of long COVID. And thus help implement proper treatment by supporting clinicians to develop the proper therapies that suit every variety accordingly. Some of the symptoms of this health disorder include nausea, abdominal pain, persistent cough, and other lung symptoms, and treatment for all these differ from one another. As a result, with such an invention, it’ll be helpful for physicians to provide treatment by identifying the proper disease. 

To check whether the software works properly, the researchers tested the software using the EHR information collected from 6,469 long COVID patients who were diagnosed with COVID-19 infections.

As per the analysis, Reese remarked that they confirmed the features of long-covid in every patient included in the EHR data. And with this, they compared the similarities using semantic similarity that allows matching between two health conditions. For example, cough is different from shortness of breath. But both these conditions come under the category of lung disorders. Likewise, they compared all these symptoms and paired up the patients in order to check the similarities between these two long COVID patients. And doing so they score these patients and execute unsupervised machine learning in order to discover the subtypes of long COVID.    

The patients who have undergone machine learning have been categorized into clusters. Later they are distinguished by examining the connection between their symptoms and the pre-existing disease. Other than this, patients are even categorized based on demographic features like their age, gender, and race.    

Reese and his colleagues are of the opinion that this software could help researchers as it is suitable for various EHR systems. Also, it allows them to collect data from different medical establishments. As of the sources, this current work was funded by the National COVID Cohort Collaborative.