Table 7.

Applications of using AI techniques in supporting COVID-19 patients.

Ref.	Application	Type of Data	AI Technique	Challenge
[203,204,205]	Chatbots to support COVID-19 patients and their relatives	Guidelines and information from a medical expert	NLP (i.e., information extraction, text summarization, and classification), speech recognition, and automated question answerers tools.	- Require a large amount of data to handle questions related to an unsaved query. - The challenge related to using various language expression (i.e., language slang)
[35,209,210]	Mining text to understand the community’s response towards governmental and health strategies (i.e., social distance, lockdown)	Text gathering from news, social media posts, healthcare, and governmental reports	NLP (i.e., information extraction, text summarization and classification)	- Privacy issues in different countries - Insufficient data may lead to skewed results. - Imprecise results leading to anxiety among the population.
[32,95,207]	Monitoring patients with temperature to maintain safety precautions) i.e., mask-wearing, social distancing, etc.)	Images extracted from infrared cameras in streets and public enterprises.	CNN models and pre-trained models (i.e., DesNet, AlexNet, etc.) and other computer vision tools and libraries	- Capturing the in-body temperature through remote sensors may lead to imprecise results. - Issues related to the invasion of privacy
[87,96,100,101,102]	Predict the spread of infection (number of expected patients, spread rate, disease peak, etc.)	Demographic data, population density, and compartmental tests,	Statistics tets and DL techniques (i.e., RNN and LSTM)	- Models such as compartmental models may be complex. - Insufficient data
[28,36,43,63,211,212,213,214,215,216,217,218,219]	COVID-19 medical diagnosis using medical images	Medical images (i.e., X-ray, CT scan, and ultrasound)	ML and DL CNN models, and AI computer vision tools	- Insufficient medical images lead to an imbalanced dataset.
[220,221,222,223,224]	Diagnosis and triage patient according to health status. Prescribe treatment, medical plan and make risk evaluation	Patient medical history (Electronic health record (EHR)), Patient symptoms, laboratory test result.	ML techniques (i.e., SVM, KNN, MLP, etc.), Fuzzy logic systems, and DL techniques (i.e., LSTM, RNN)	- Unavailability of patient’s data (therapeutic outcomes and physiological data). - Privacy issues - Incomplete data may lead to biased or accurate result in the prediction
[225,226,227]	Analyses of viral RNA and track genetic changes. Predict the viral structure of the second and third waves.	Protein sequence and viral RNA	DL and Deep reinforcement learning tools	- Analyzing a large dataset for RNA or protein sequence may take a long time, result in unexplainable models
[161,163,184,185,228,229,230,231]	Analyze chemical compounds and interaction for vaccine development	Viral structure, protein sequence, drug–drug interaction, drug–protein interaction, and protein–protein interaction.	DL models, computer vision tools, reinforcement learning, and optimization techniques	- Results need large bed experiments to be verified, which may take a long time. - Possibility of long-term risk.
[206,207,208]	Develop robots to support both patient and medical staff, cleaning, vital signs monitoring, deliver food and treatment	Training autonomous agent using environment simulation	DL models, computer vision tools, reinforcement learning, and optimization techniques	- Training autonomous agents and implementing them in machines may take great effort and time. - Maintaining a high level of safety must be guaranteed
[232]	Develop a reponse tracker (OXGRT) to capture the government policies and the degree of response	Aggregating huge dataset that is continuously updated	Use AI techniques to explore the empirical effect of government policies on the spread of COVID-19 cases	-