Back
Labeling

NLP: A Potential Savior of the Covid-19 Outbreak

Ananya Avasthi
October 6, 2021

The world was forever changed when a new disease started to spread all over the world in 2020. Coronavirus disease (COVID-19) is an infectious disease caused by the SARS-CoV-2 virus.The virus may metastasize from an infected person’s mouth or nose in minuscule liquid particles once they cough, sneeze, speak, sing or breathe. These particles vary from larger metabolic process droplets to smaller aerosols. The virus may infect someone who is merely breathing in the air. It can also spread by touching a contaminated surface and then one’s eyes, nose, or mouth. Covid-19 has a higher chance of spreading in crowded and indoor settings.

Panic spread worldwide when the news of the drastic spread of Covid-19 engulfed the world with fear. Countries initiated lockdowns, and there was a massive shortage of medical assistance compared to the Covid-19 cases. A significant part of the population was succumbing to Covid-19, and a few percent of those patients passed away due to either lack of medical assistance or late detection. Covid-19 is still affecting today’s world. The production of the vaccine helps to slow the spread of Covid-19. Like any virus, the Covid-19 strain steadily keeps on evolving, making this threat far from over. After such a pandemic, the medical sphere needs any help it can get. Natural Language Processing (NLP) has played a massive role in assisting our medical warriors in helping overcome the pandemic. NLP assists AI in understanding natural language. NLP understands the basic techniques of word definition, phrases, sentences, and texts, as well as syntactic( knowledge of word meanings and vocabulary) and semantic processing (understanding the combination of terms). It also develops machine translation (MT), question-answering (QA), data retrieval, discussion, document production, and recommendation programs.

Uses of NLP during the pandemic

Natural language processing (NLP) focuses on obtaining information by probing text and speech data. NLP tools like image recognition, speech processing, and speech understanding technologies have taken over consumer products like cell phones and smart speakers. Recent accomplishments of NLP include automatic speech recognition, information extraction, and image captioning. Organizations are deploying NLP models to develop clinical voice assistants to transcribe patient visit information into their electronic health records (EHR) (An electronic health record contains a patient’s entire medical history). They deploy this technology to shorten the time a doctor spends on documentation: This creates more time for a clinician to work with patients directly during the pandemic. NLP that preprocesses EHR and then finds and classifies disease-relevant keywords for early detection of various diseases is also becoming popular.

A joint study performed by he Miama Cardiac and Vascular Institue and Florida International Universitycompared three different NLP models and their relative usefulness to predict and forecast COVID-19 cases.

From January 2020 to October 2020, the joint applied three NLP algorithms to several chest CTs. Experts drew up a correlation matrix and concluded that out of the 3 NLP algorithms, the two that worked with image finding and researching from several datasets were the most successful. These tools helped in identifying Covid-19 in patients, which led to early treatment. Using NLP, tracking the progression of new COVID-19 cases in all 50 states was done using a real-time interactive U.S. map model. That map displayed the temporal relationship between the COVID-19 pandemic and the NLP models. 

The NLP models studied during the pandemic provided powerful insights to radiology departments, hospitals, and communities. The first NLP model was more of a basic algorithm for deducing viral infection that included CT findings indicative of viral infection. However, the viral infection was not specific to Covid-19, and people contracted the flu during the pre-pandemic time. The second NLP model focused solely on CT findings that indicated COVID-19. The third NLP model targeted the data provided by the radiologist that suggested viral infection or was highly suggestive of COVID-19. The third model performed the best out of the 3 NLP models. This NLP study demonstrates the vast usefulness that NLP can have in predicting COVID-19 outbreaks: This helps in early detection and saving lives and gives a substantial helping hand to our medical team.

The COVID-19 pandemic is pressing society and the economy in an unprecedented way. Overall, the U.S. response has not been as successful as other countries: This may have been caused by the lack of amenities like testing kits and essential awareness when the pandemic started. If the United States focuses on early warning and detection systems to keep track of respiratory diseases, it can prevent the massive loss of life and the instability that the pandemic causes. Medical organizations can use NLP to stay ahead of Covid-19through early detection. As NLP stores more data, it may come up with even better detection and prevention programs.