Page 12 - AIH-1-2

Artificial Intelligence in Health AI in the battle against COVID-19

The utilization of AI and big data in managing the 6.2. Machine learning prediction models
COVID-19 pandemic has been unprecedented. The A multitude of research studies have investigated the use of
analysis of vast datasets has provided insights that were machine learning techniques in predicting and detecting
previously unattainable, demonstrating the evolution of COVID-19 based on symptomatology. One notable study
42
AI and data analytics in the context of pandemics. The in this domain is presented by Ahamad et al., who
50
COVID-19 pandemic has also been a catalyst for the developed a machine-learning model targeting early-stage
rapid development and adoption of AI in various aspects symptoms of SARS-CoV-2 infection. Utilizing supervised
of healthcare and public health. This includes areas such machine learning methods, they focused on patient
as disease detection and diagnosis, vaccine development, characteristics and clinical details such as fever, cough,
treatment strategies, and epidemiological modeling. and lung infection to predict COVID-19 status with over
Significant applications of AI have been identified in the 85% accuracy. Zoabi et al. introduced a machine-learning
51
COVID-19 pandemic, building on the results from prior approach using data obtained from tested individuals in
6
research and the lessons learned from past health crises. The Israel. They trained their model on information such as
pandemic has highlighted the potential of AI to contribute sex, age, exposure to the infected individual, and clinical
significantly to managing public health emergencies and is symptoms recognized at the time of testing. Their model
likely to set a precedent for future use in similar scenarios. achieved high accuracies in COVID-19 detection and

6. AI in COVID-19 detection and diagnosis identified key symptoms such as fever and cough as leading
indicators for positive diagnosis.
The COVID-19 pandemic has spurred an unprecedented In the paper published by Menni et al., data were
52
reliance on AI technologies in disease detection and obtained from a COVID-19 symptom tracker smartphone
diagnosis. This section elucidates the multifaceted role app with 2.6 million users in the United States and the
of AI in confronting the diagnostic challenges posed by
COVID-19, highlighting innovative methodologies and United Kingdom. The study found a strong association
their implications in medical diagnostics. between the loss of smell and taste and COVID-19-positive
cases. Logistic regressions were employed, and a symptom
6.1. Imaging techniques prediction model was developed, showing high sensitivity
and specificity in predicting COVID-19.
The integration of AI into imaging techniques has played an
important role in the detection and diagnosis of COVID-19. 6.3. NLP in symptom assessment
43
Deep learning models, particularly convolutional neural NLP has been instrumental in the development of
networks, have been employed to discern patterns in chest AI-based chatbots and virtual health assistants during the
X-ray images and computed tomography scans indicative of COVID-19 pandemic. 53,54 These platforms are capable of
viral infection. 44,45 Various large datasets of medical images conducting preliminary symptom assessments through
from COVID-19 patients were independently collected patient interactions, streamlining the assessment and
for training and validating deep learning models used in triage process, and facilitating early detection of potential
detecting COVID-19 in patients. 46,47
COVID-19 cases. Interactive digital health assistants, such
These deep learning models not only detect COVID- as Symptoma, have shown to be more accurate than online
19 but also predict and assess the severity of the disease, questionnaires in identifying COVID-19 cases because
which is vital for accurate diagnosis and effective patient users can input more detailed information regarding
management. These AI-driven systems can quantify their symptoms through a natural language conversation
the degree of lung damage, detect signs of pneumonia, with the system. By offering accessible and immediate
55
and identify other complications associated with severe assistance to the public, these tools alleviate the stress and
COVID-19 infections. Advanced imaging techniques overwhelming volume faced by telephone hotlines and
48
have enabled health-care professionals to gauge the extent medical institutions.
of lung involvement and other critical factors that classify Furthermore, AI chatbots with advanced NLP
the severity of the infection. This capability is crucial for capabilities have extended their services to include
49
triaging patients, determining appropriate levels of care, mental health support. The pandemic has led to increased
and making timely decisions regarding treatment strategies. levels of stress, anxiety, and other mental health issues
These AI-driven tools that analyze medical images have among the population. Chatbots have provided a first
demonstrated remarkable efficacy in enhancing the speed line of psychological support, offering coping strategies,
and accuracy of COVID-19 diagnosis, thereby alleviating mindfulness exercises, and, in some cases, referral to
the burden on healthcare systems. mental health professionals. 56

Volume 1 Issue 2 (2024) 6 doi: 10.36922/aih.2401