CC BY
35
Applying Machine Learning to Optimize Vaccine Distribution for COVID-19
Fredrick Ishengoma
Abstract—The widespread outbreak of COVID-19 has come with several challenges in terms of vaccine distribution, such as shortages in supply, logistical hurdles, and public uncertainty. However, the application of machine learning can potentially alleviate these challenges by offering valuable perspectives on the distribution of vaccines, forecasting demand, and recognizing areas with a higher transmission risk. This paper analyzes the utilization of advanced artificial intelligence techniques to optimize the allocation of vaccines for the COVID-19 virus. This paper delves into the machine-learning approaches employed or suggested for vaccine distribution, including decision tree models, neural networks, and simulation-based methodologies. In addition, the paper addresses the challenges and limitations of using machine learning for vaccine distribution, including the necessity for high-quality data and ethical considerations. In conclusion, this paper offers a comprehensive examination of the current state of research in the application of machine learning for optimizing vaccine distribution for COVID-19 and highlights areas for further study.
Keywords—Artificial Intelligence, COVID-19, Machine Learning, Vaccine Distributions.
I. Introduction
The COVID-19 pandemic has created a global crisis and posed significant difficulties in terms of vaccine distribution [1-2]. Among the challenges faces during the peak time of the COVID-19 includes logistical barriers, and public skepticism on getting vaccines to those who require them most [3, 4]. As the availability of vaccines increases, it becomes increasingly crucial to optimize their distribution to ensure they reach those who need them in the most efficient and effective manner.
One of the recent sophisticated solutions to enhance vaccine distribution is through the utilization of artificial intelligence techniques [5]. Machine learning is a formidable technology that can provide predictions on demand, pinpoint high-transmission risk areas, and offer an understanding into vaccine distribution patterns [6]. This can guarantee that vaccines are distributed to the areas where they are needed most and can also aid in alleviating logistical obstacles by identifying the most optimal routes for vaccine delivery.
Different machine learning methods have been employed or proposed for vaccine distribution, including decision tree models, neural networks, and simulation-based methodologies [6]. Decision tree models can predict vaccine demand based on factors such as population demographics and prior vaccination rates. Neural networks can detect high
transmission risk areas by analyzing data on the progression of the virus [7]. Simulation-based approaches can model vaccine distribution and identify potential bottlenecks in the distribution process.
One of the significant difficulties in utilizing machine learning for vaccine distribution is the requirement for data of high quality [8]. For artificial intelligence models to be effective, they need a substantial amount of precise and dependable data. However, collecting and refining data on vaccine distribution can prove challenging, especially during a pandemic. Furthermore, ethical considerations must be taken into account when using machine learning for vaccine distribution, such as privacy concerns and the possibility of bias in algorithms [9-10].
Despite these obstacles, the potential advantages of applying machine learning to vaccine distribution are substantial. Machine learning can offer understanding into vaccine distribution patterns, predict demand, and pinpoint areas of high transmission risk, ensuring that vaccines are distributed to the areas where they are needed most. It can also help in mitigating logistical barriers by identifying the most efficient routes for vaccine delivery [11]. For instance, decision tree models can predict vaccine demand based on factors such as population demographics and prior vaccination rates. Neural networks can identify high transmission risk areas by analyzing data on the spread of the virus. These insights can help to guide public health policy and build public confidence in the vaccine distribution process.
In spite of the difficulties that may arise, the ramifications of utilizing machine learning in vaccine distribution are substantial. By imparting a deeper understanding of the patterns that underlie vaccine distribution, forecasting demand, and pinpointing areas of elevated transmission risk, machine learning can aid in ensuring that vaccines are channeled to their most critical destinations and can also serve to alleviate logistical hurdles through the identification of the most expeditious delivery routes [11].
For example, decision tree models may be employed to predict vaccine demand based on such variables as population characteristics and prior vaccination rates, thereby enhancing the allocation of vaccines to their most vital locations. Neural networks can analyze data regarding the spread of the virus to determine areas of heightened transmission risk, thus enabling focused vaccination initiatives and streamlined delivery paths [6]. Furthermore, the insights derived from machine learning can prove instrumental in shaping public health policy and fostering public confidence in the vaccine distribution process.
However, the application of machine learning to optimize vaccine distribution remains in its nascent stages and there is ample room for further exploration and growth. This paper provides a comprehensive survey of the current state of research in the utilization of machine learning to enhance vaccine distribution for COVID-19 and highlights avenues for future inquiry. Additionally, the study delves into the challenges and limitations that must be considered when applying machine learning to vaccine distribution.
demographics, prior vaccination rates, and related factors [12-13]. Through the deployment of historical data, these models can be trained to detect patterns in the data that are indicative of vaccine demand. For instance, a decision tree model could be deployed to prognosticate vaccine demand in various regions of a country based on factors such as population density and the prevalence of healthcare workers in the region (Figure 1).
II. Materials and Methods
The methodology employed in this study encompassed a synthesis of relevant literature and a comprehensive analysis of code. In order to attain a thorough comprehension of the utilization of Machine Learning (ML) in vaccine distribution during the ongoing COVID-19 pandemic, a systematic examination of pertinent literature was performed. This entailed an extensive search of academic databases, such as PubMed and Scopus, to identify relevant articles and studies that met specific eligibility criteria, including a focus on the use of ML for vaccine distribution during the COVID-19 pandemic. The eligible articles were then subject to synthesis to extract pertinent information and discern the current state of the field.
In addition to the synthesis of literature, a code analysis was also conducted. This involved a review of open-source code related to the utilization of ML for vaccine distribution during the COVID-19 pandemic. The code was analyzed to identify the algorithms utilized and the data sources employed, as well as to obtain insights into the implementation of ML models and the challenges encountered in their application. The results of the code analysis were then compared with the findings from the literature synthesis to attain a comprehensive understanding of the utilization of ML for vaccine distribution during the COVID-19 pandemic.
The conjunction of literature synthesis and code analysis provided a comprehensive comprehension of the utilization of ML for vaccine distribution during the COVID-19 pandemic. The literature synthesis offered an overview of the current state of the field, while the code analysis provided insights into the implementation of ML models. The combination of these two methodologies provided a well-rounded understanding of the utilization of ML for vaccine distribution during the COVID-19 pandemic and aided in the identification of areas for future research.
III. Machine Learning Technology and Methods for COVID-19 Vaccine Distribution
The application of machine learning as a tool for optimizing vaccine distribution during the ongoing COVID-19 pandemic is of paramount importance. A cornucopia of machine learning methodologies, including decision tree models, neural networks, and simulation-based approaches, have been put forward or utilized for this purpose.
Decision tree models, a subclass of supervised machine learning algorithms, are particularly efficacious in the prediction of vaccine demand based on population
:roT sklearn impon tree import panda? as pd
# Load the cat. a into a pancas Lata Frame
data = pd. read_cav( "vaccine_deinand_data.csv,r)
# Define the features and target variables
X = data [ [ rpopulatiDn_densityT, 1 number_Qf_healthcare_worliera 1 ] ] y = data [1 vac cine_demar.d']
# Create the decleicn tree model elf = tree.DecisionIreeRegressor(} elf = clf.=lt(X, v}
# Jse the model to make predictions predictions = elf.predict{Kj
Figure 1: An example of how to create a decision tree model in Python using the scikit-learn library for COVID-19 vaccine distribution.1
Neural networks, another category of machine learning algorithms, are also instrumental in the optimization of vaccine distribution [14-16]. These models can be leveraged to identify areas of elevated transmission risk through the analysis of data related to the progression of the virus. Neural networks can be taught to recognize patterns in the data that are indicative of transmission risk, such as the incidence of new cases in a given area. By identifying areas of heightened transmission risk, neural networks can aid in the targeting of vaccines to the areas where they are most urgently needed and can help alleviate logistical constraints through the identification of the most expeditious delivery routes.
Additionally, Neural Networks can help alleviate logistical constraints in vaccine distribution by identifying the most efficient delivery routes, allowing for vaccines to reach their intended recipients in a timely and effective manner. The utilization of Neural Networks in vaccine distribution can greatly enhance the overall effectiveness of the distribution process and assist in overcoming the
1 The provided code is a demonstration of the creation of a decision tree model utilizing the scikit-learn library in the Python programming language. It commences with the importation of the necessary libraries, including tree from sklearn and pandas. Subsequently, it loads a csv file named "vaccine_demand_data.csv," which serves as the data source for the model.
The code then proceeds to specify the features and target variables for the model through the utilization of a pandas DataFrame, where "population_density" and "number_of_healthcare_workers" are designated as the features and "vaccine_demand" is designated as the target variable. The decision tree model is then constructed through the application of the DecisionTreeRegressor() function and is fit to the designated features and target variables. Finally, the model is utilized to generate predictions based on the input features.
International Journal of Open Information Technologies ISSN: 2307-8162 vol. 11, no.5, 2023 challenges posed by the COVID-19 pandemic (Figure 2).
import rm:py is np import panda? as pd
fro t. sklear:i. preprocessing import MinMaxScaler frozr. tensorflovi.keras .modeIs import 5eijuen~ial from tensorflow.keras.layers import Dense
# load the ¿a~ a into a pandas DataFraroe data = pd.read_cav["ccvid_caTa.csv")
# Define the features and target varlatlea X = data[[r populat i-n_dens _ty ,
' nimLber_of_healtlicare_workers1 , ' age_groupa ' r 'location' , ' travel_historyT]]
y = data ['higl^ris ^transmission" ]
# Soale the Input data scaler = MinMaxScaler [)
X = scaler .fit-_transfor^.(X)
# Create the neural network: Tf.odel model = Sequential[)
model. add (Der.se {1 jj iijnt_Qi:f.-j, actlvatior_= 1 relu 1 ) ) model .add(Der.se {3, actlvatlon=' relur) ] model .add(Der.se <1, actlvatlon= Sigmoid1 ) )
# Compile the motel
model .compile (loss= rbinary_crosaentropy1 , opti:r.izer='adamr, :c.etrics=[1 accuracy'])
# Fit the model to the data
mcdel .fit. ¡X, y, epocha=100, tatch_siie=32)
# T?se the model to make predictions predictions = model.predict(X)
Figure 2: An example of how a neural network can be implemented in Python to identify areas of high-risk transmission of COVID-19 and help target vaccines to the areas where they are most needed2
Simulation-based methodologies are a formidable tool that can be harnessed to elevate the dissemination of vaccines. This is attained through the creation of computergenerated simulations that model the vaccine distribution process and identify any potential obstacles. The implementation of these methodologies necessitates the construction of models of the distribution system, incorporating elements such as transportation networks, storage capacities, and vaccination centers. Demographic information, including population density and the concentration of healthcare workers, can also be integrated into the models to facilitate the determination of areas with the most pressing need for vaccines [17].
The simulation of various scenarios and the identification of the most efficacious distribution strategies can optimize vaccine dissemination and guarantee that vaccines reach those who require them the most (Figure 3).
2 The code in question represents a demonstration of a neural network implemented in the Python programming language, aimed at determining areas of elevated transmission risk of COVID-19 and aiding in the targeting of vaccines to their most critical destinations. It begins by loading a csv file named "covid_data.csv," which serves as the data source for the model.
The code then proceeds to define the features and target variable, scale the input data, and construct the neural network model comprised of three layers. The model is then compiled using a binary_crossentropy loss function, Adam optimizer, and accuracy metric. Finally, the model is fit to the data and employed to generate predictions.
import sinipy
def vaccine_diatrifcution(env, population, vaccirie5_per_day) : while T^ue:
yi el-i env. time oat i 1} vaccinated = nin(population, va cc ine)
population -= vaccinated
env = a irnçjy - Envi:: onaient ■! )
population = lû 03
vaccines_per_day = 103
env.proce a 3{vaccine_di stribution(env,
population, vaccir.ea_per_day j )
env.run(until=3€S:
print("After 1 year, there are atill {} people to vaccinate.".format tpopulation: )
Figure 3: An example of how simulation-based approaches can be implemented in Python using a library such as SimPy for vaccine distribution3
Furthermore, these methodologies can assist those in decision-making positions in anticipating and remedying any potential problems prior to their occurrence, such as vaccine shortages or logistical challenges. Additionally, they can be utilized to assess the effectiveness of different distribution strategies and detect opportunities for improvement.
The utilization of reinforcement learning is yet another mode of harnessing the capabilities of machine learning for optimizing vaccine distribution [18-20]. Reinforcement learning constitutes a variant of machine learning that models the vaccine distribution process as a decision-making quandary and endeavors to optimize it through trial-and-error learning [21]. This approach can facilitate the identification of the most efficient distribution strategies by simulating various scenarios and learning from the outcomes.
In the context of vaccine dissemination, reinforcement learning can be leveraged to optimize the allocation of resources, such as vaccines and healthcare workers, to distinct regions or populations. This is accomplished through the training of a model on historical data, comprising population demographics, vaccine uptake, and logistics information to facilitate the learning of decision-making
3 The code fragment represents a demonstration of the application of simulation-based methodologies to optimize vaccine distribution, implemented in Python utilizing the SimPy library. It defines a function, "vaccine_distribution," which models the dispersal of vaccines to a population via a loop that continues until all individuals have received the vaccine. Within the loop, it employs the command 'yield env.timeout(1)' to pause for a period of one day, followed by a calculation of the number of individuals who can be vaccinated by taking the minimum of the population and the available vaccines per day. It then updates the population by subtracting the number of vaccinated individuals. Subsequently, it establishes a simulation environment, sets the values for population and vaccines per day, and runs the simulation for a duration of one year. The final line of code outputs the number of unvaccinated individuals at the conclusion of the simulation.
processes regarding resource allocation. Subsequently, the model can simulate various scenarios and determine the most effective distribution strategies for different populations. Optimizing scheduling vaccination clinics can also be achieved through reinforcement learning, taking into account factors like vaccine availability and demand, ensuring that vaccines reach those who require them most, and minimizing wastage while maximizing the utilization of resources.
Integrating decision tree models, neural networks, simulation-based methodologies, and clustering algorithms, such as K-Means and Gaussian Mixture Model, can provide a more comprehensive understanding of vaccine distribution patterns and facilitate the identification of the most effective strategies for reaching different populations [22-24]. Clustering algorithms can group similar regions with analogous characteristics together, predict the demand of vaccines in these regions, and allocate resources accordingly, prioritizing regions with high predicted demand for vaccines (Figure 4).
sklearn.cluster import EMeana import pandas as pd
f load The data ir.t-? a pandas Da ta Frame
da~ a = pd. re ad_cav ¡"vacc: ne_deirar.d_data .cav")
t Define the features to be used f-^r clustering
Yl = ¿at a [ ! 1 papula ti-?n_-densi ty ' r 1 numberf_h=a 1 thcare_worte rs', 'paat_vaccination_rate1r 1viru3_spread_rate1 I]
t Creat-e the K-Means nodel trtsans = KMeana(n_cluster3=5)
t Fit the jrjcdeL. to the data treans. fit (X)
t Use the irjcdeL. to predict the cluster fax
eazh da^a peint
clusters = kaneana .predict iii:
Figure 4: An example of how simulation-based approaches can be implemented in Python using a library such as SimPy for vaccine distribution4
4 The ccode exemplifies the utilization of the K-Means algorithm to categorize regions with similar characteristics, taking into account factors such as population demographics, previous vaccination rates, and the spread of the virus. The necessary libraries are imported first, followed by the loading of a data file named "vaccine_demand_data.csv" that serves as the foundation of the model. The features utilized for clustering, population density, concentration of healthcare workers, prior vaccination rate, and virus spread rate, are then defined. Subsequently, the K-Means model with
5 clusters is created and fitted to the data. The model is then employed to predict the cluster for each data point, resulting in the separation of the dataset into 5 distinct clusters, grouping regions that possess similar traits. These clusters can be leveraged to anticipate the demand for vaccines in each cluster and allocate resources accordingly.
Additionally, clustering algorithms can be employed in a synergistic manner with reinforcement learning techniques to optimize vaccine distribution. For instance, clustering can be utilized to divide the population into smaller, more manageable subgroups, and reinforcement learning can then be utilized to make informed decisions on which subgroup should be targeted next, based on the current state of the vaccination coverage. It is imperative to emphasize that these clustering algorithms ought to be utilized in conjunction with other machine learning methodologies, and not as a standalone solution. The quality and accessibility of data will play a crucial role in determining the effectiveness of these algorithms. Moreover, it is advisable to consistently test and validate any models on diverse datasets and to frequently update them in response to the dynamic spread of the virus.
The utilization of Natural Language Processing (NLP) techniques in the analysis of unstructured data sources, such as social media content, news articles, and other forms of textual information, can offer valuable insights into public opinion and sentiment pertaining to the vaccine distribution process [25-27]. This can aid in the formulation of informed public health policies and outreach strategies by providing a deeper understanding of the reasons behind vaccine hesitancy.
One approach of NLP that can be employed is Sentiment Analysis, which classifies text data into different categories of sentiment, such as positive, negative, or neutral, through the utilization of machine learning algorithms. This methodology can be employed to identify trends of public opinion towards the vaccine distribution process and inform targeted outreach efforts.
Another NLP approach is Topic Modeling, which identifies patterns of public opinion and sentiment pertaining to specific topics. This can be employed to identify issues related to vaccine distribution, such as safety concerns, and inform targeted outreach efforts by addressing these specific topics. Additionally, NLP can be leveraged to identify transmission risk patterns in data by analyzing the number of new cases in specific areas, thus guiding the targeted distribution of vaccines to areas of most need, and mitigating logistical constraints by determining the most efficient delivery routes.
Moreover, Bayesian models can be used to estimate the distribution of vaccine coverage and identify regions where coverage is low, and where more vaccines should be allocated. Machine learning holds immense potential in optimizing vaccine distribution during the COVID-19 pandemic, and various methods such as Decision Tree models, Neural Networks, Simulation-based approaches, Reinforcement Learning, Clustering, NLP, and Bayesian models can be utilized in combination to provide a more comprehensive understanding of vaccine distribution patterns and determine the most effective strategies for reaching different populations
Table 1: Summary of policy implications
Area Description Implications
Utilizing AI/ML in Vaccine Allocation Use ML to enhance vaccine allocation by understanding patterns, predicting demand, and identifying high-risk areas. Direct vaccines to areas in need, predict demand, address public hesitancy, and improve distribution efficiency.
Anticipating Vaccine Demand Use ML to predict demand based on demographic factors and vaccination rates. Ensure vaccines reach those in need, regardless of demographics, and increase public confidence.
Addressing Public Hesitancy Use ML to identify factors impacting vaccine uptake. Provide specific messaging and education to address these factors. Increase vaccine uptake and improve public confidence.
IV. IMPLICATIONS FOR PRACTICE
The utilization of artificial intelligence and machine learning techniques to enhance the allocation of COVID-19 vaccines has substantial ramifications for its practical applications. By means of rendering comprehension of the patterns of vaccine dissemination, anticipating the necessity, and pinpointing areas with a high incidence of contagion, machine learning can contribute to ensuring the optimal distribution of vaccines to where they are most indispensable and alleviate the logistical hindrances.
One of the paramount implications for practice is the capability to direct vaccines to locations that necessitate them the most. By recognizing areas with a heightened transmission risk, machine learning can aid in directing vaccines to these areas, thereby alleviating logistical obstacles and guaranteeing that the vaccines reach the individuals who require them the most, particularly in remote or under-served regions where conventional distribution methods may be ineffective.
Another significance for practice is the potential to foretell vaccine demand. By employing machine learning to predict vaccine demand based on elements such as demographic characteristics of the population and previous vaccination rates, public health officials can guarantee that vaccines are distributed to where they are most required. This can assure that vaccines reach the individuals who need them the most, regardless of their demographic traits, such as age, gender, race, or socio-economic status. Machine learning can also help to tackle public reluctance by identifying the factors that impact an individual's willingness to receive vaccinations and by providing specific messaging and education to address these factors. By affording insights into the reasons behind an individual's hesitation to receive vaccines, machine learning can foster public confidence in the vaccine distribution process and augment vaccine uptake.
The utilization of machine learning in vaccine distribution has the potential to enhance the efficiency of the dissemination process as well. Simulation-based methodologies can be employed to simulate the distribution of vaccines and identify potential bottlenecks in the process. This can assist in identifying areas that require additional resources and optimize the distribution of vaccines to ensure that they reach the individuals who need them the most.
The application of machine learning to optimize the distribution of COVID-19 vaccines has substantial implications for practice. By providing insights into vaccine distribution patterns, predicting demand, and identifying areas of high-risk transmission, machine learning can ensure that vaccines are distributed where they are most needed and alleviate logistical constraints. Furthermore, machine learning can also tackle public hesitancy and improve the efficiency of the vaccine distribution process. Table 1 provides a summary of policy implications for employing machine learning techniques to foster COVID-19 vaccine distribution.
V. Conclusion
The COVID-19 pandemic has posed formidable difficulties in terms of vaccine distribution. Machine learning represents a formidable tool that can be harnessed to optimize vaccine allocation by providing a deeper understanding of the patterns of vaccine dissemination, foretelling demand, and locating regions with a high transmission risk. A plethora of machine learning approaches have been proposed, encompassing decision tree models, neural networks, simulation-based methodologies, clustering algorithms, and natural language processing techniques. The utilization of machine learning to optimize the distribution of COVID-19 vaccines has substantial ramifications for practice, including the assurance of optimal distribution to areas of utmost necessity and the alleviation of logistical hindrances. Furthermore, machine learning can play a crucial role in addressing public reluctance and enhancing the efficiency of the distribution process.
It is imperative to acknowledge that the implementation of these techniques demands a substantial amount of technical aptitude and resources. Additionally, the quality and availability of data plays a crucial role in determining the performance of these models. Furthermore, it is necessary to continuously test and validate models on various datasets and update them frequently to keep pace with the ever-evolving spread of the virus. Machine learning can prove to be a potent weapon in the battle against COVID-19 by optimizing vaccine distribution. By leveraging machine learning methodologies to analyze data pertaining to the spread of the virus, public sentiment, and vaccine demand,
public health officials can target vaccines to areas that necessitate them the most and ensure that vaccines reach the individuals who require them the most.
The ongoing COVID-19 situation presents a wealth of possibilities for future studies aimed at enhancing the use of machine learning in vaccine administration. One important area of study is the advancement of machine learning algorithms that can accurately predict vaccine demand and pinpoint regions of high viral transmission. The potential of deep learning models, which have proven to be effective in other aspects of public health, such as disease diagnosis, could be explored. To increase the precision of these algorithms, alternative data sources or innovative techniques for integrating information from multiple sources could be incorporated.
Another area of interest is the examination of the utilization of machine learning to streamline the logistics of vaccine distribution. This could involve the development of models that can determine the most efficient delivery routes or predict the effects of alternative distribution methods on overall vaccination coverage. The use of simulation-based approaches, such as agent-based models, to optimize vaccine distribution could also be analyzed.
Using machine learning to address public vaccine hesitancy is another area of exploration. This could involve the analysis of unstructured data sources, such as social media and news articles, to identify patterns of public perception and sentiment. Reinforcement learning and decision-making techniques aimed at optimizing vaccine distribution based on public sentiment could also be studied.
Lastly, there is a need for further investigation into the ethical and legal implications of using machine learning in vaccine administration. This could encompass inquiries into data privacy, data governance, and data bias. Research could also be conducted to ensure equitable vaccine distribution across all populations through the use of machine learning. In conclusion, there are several avenues for future research that could be pursued to optimize the use of machine learning in vaccine administration during the COVID-19 pandemic, including the advancement of algorithms, optimizing logistics, addressing public hesitancy, and investigating ethical and legal implications. Table 2 summarizes future research works.
Table 2: Summary of future areas for research
Area of Research Description Methods
Adv. of ML Algorithms Dev. algorithms to predict vaccine demand and identify transmission risk areas. Explore deep learning. Model exploration
Logistics Optimization Use ML to improve vaccine distribution logistics, including delivery routes and distribution Model dev./analysis
impacts.
Public Vaccine Hesitancy Use ML to analyze public sentiment from unstructured data sources. Study reinforcement learning and decision-making. Data analysis, study
Ethical/Legal Implications Investigate ethical and legal issues related to using ML for vaccine distribution, including privacy, governance, and bias. Ensure equitable distribution. Investigation, research
References
[1]
F. Mollarasouli, N. Zare-Shehneh, and M. A. Ghaedi, "A review on corona virus disease 2019 (COVID-19): current progress, clinical features and bioanalytical diagnostic methods," Microchim Acta, vol. 189, pp. 103, 2022. doi: 10.1007/s00604-022-05167-y.
G. Farnoosh, G. Alishiri, SRH. Zijoud, et al., "Understanding the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease (COVID-19) based on available evidence - a narrative review," J. Mil. Med., vol. 22, pp. 1-11, 2020. doi: 10.30491/JMM.22.1.1.
M. Shiri and F. Ahmadizar, "An equitable and accessible vaccine supply chain network in the epidemic outbreak of COVID-19 under uncertainty," J. Ambient Intell. Human Comput., 2022. doi: 10.1007/s12652-022-03865-2.
E. A. Andoh and H. Yu, "A two-stage decision-support approach for improving sustainable last-mile cold chain logistics operations of COVID-19 vaccines," Ann. Oper. Res., 2022. doi: 10.1007/s 10479-022-04906-x.
Y. Peng, E. Liu, S. Peng, et al., "Using artificial intelligence technology to fight COVID-19: a review," Artif. Intell. Rev., vol. 55, pp. 4941-4977, 2022. doi: 10.1007/s10462-021-10106-z. D. O. Oyewola, E. G. Dada, and S. Misra, "Machine learning for optimizing daily COVID-19 vaccine dissemination to combat the pandemic," Health Technol., vol. 12, pp. 1277-1293, 2022. doi: 10.1007/s12553-022-00712-4.
D. Qiu, Y. Yu, and L. Chen, "Emotion Analysis of COVID-19 Vaccines Based on a Fuzzy Convolutional Neural Network," Cogn. Comput., 2022. doi: 10.1007/s12559-022-10068-6.
B. Jahn, S. Friedrich, J. Behnke, et al., "On the role of data, statistics and decisions in a pandemic," AStA Adv. Stat. Anal., vol. 106, pp. 349-382, 2022. doi: 10.1007/s10182-022-00439-7.
C. Lawrence, D. J. Vick, T. Maryon, et al., "Ethical allocation of COVID-19 vaccine in the United States: an evaluation of competing frameworks for the current pandemic and future events," J. Public Health Pol., vol. 43, pp. 234-250, 2022. doi: 10.1057/s41271-022-00338-w.
[10] A. A. Nichol and K. M. Mermin-Bunnell, "The ethics of COVID-19 vaccine distribution," J. Public Health Pol., vol. 42, pp. 514-517, 2021. doi: 10.10
[11] Heidari, A., Jafari Navimipour, N., Unal, M., et al. "Machine learning applications for COVID-19 outbreak management." Neural Comput & Applic., vol. 34, pp. 15313-15348, 2022. https://doi.org/10.1007/s00521-022-07424-w
[12] Yaesoubi, R., You, S., Xi, Q., et al. "Generating simple classification rules to predict local surges in COVID-19 hospitalizations." Health Care Manag Sci, 2023. https://doi.org/10.1007/s10729-023-09629-4
[13] Aljedaani, W., Abuhaimed, I., Rustam, F., et al. "Automatically detecting and understanding the perception of COVID-19 vaccination: a middle east case study." Soc. Netw. Anal. Min., vol. 12, pp. 128, 2022. https://doi.org/10.1007/s13278-022-00946-0
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[14] Liu, H., & Lang, B. "Machine Learning and Deep Learning Methods for Intrusion Detection Systems: A Survey." Applied Sciences, vol. 9, no. 20, pp. 4396, 2019. MDPI AG. https://dx.doi.org/10.3390/app9204396
[15] Barajas, M., Bhatkande, S., Baskaran, P., Gohel, H., & Pandey, B. "Advancing Deep Learning for Supply Chain Optimization of COVID-19 Vaccination in Rural Communities." In Proceedings of the 10th IEEE International Conference on Communication Systems and Network Technologies (CSNT), Bhopal, India, 2021, pp. 690695. https://doi .org/10.1109/CSNT51715.2021.9509710
[16] Torku, T. K., Khaliq, A. Q. M., & Furati, K. M. "Deep-Data-Driven Neural Networks for COVID-19 Vaccine Efficacy." Epidemiologia, vol. 2, no. 4, pp. 564-586, 2021. MDPI AG. https://dx.doi.org/10.3390/epidemiologia2040039
[17] AlShurman, B. A., Khan, A. F., Mac, C., Majeed, M., & Butt, Z. A. "What Demographic, Social, and Contextual Factors Influence the Intention to Use COVID-19 Vaccines: A Scoping Review." International Journal of Environmental Research and Public Health, vol. 18, no. 17, pp. 9342, 2021. MDPI AG. https://dx.doi.org/10.3390/ijerph18179342
[18] Hanie, R. L., & van Rensburg, J. J. "Using Reinforcement Learning Algorithms to Explore COVID-19 Spread in South Africa." In Proceedings of the International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems (icABCD), Durban, South Africa, 2021, pp. 1-9. https://doi .org/10.1109/icABCD51485.2021.9519325
[19] S. Khalilpourazari and H. H. Doulabi, "Using Reinforcement Learning to Forecast the Spread of COVID-19 in France," in Proceedings of the 2021 IEEE International Conference on Autonomous Systems (ICAS), Montreal, QC, Canada, pp. 1-8, 2021. doi: 10.1109/ICAS49788.2021.9551174.
[20] S. Chen, et al., "Reinforcement Learning Based Diagnosis and Prediction for COVID-19 by Optimizing a Mixed Cost Function from CT Images," IEEE Journal of Biomedical and Health Informatics, vol. 26, no. 11, pp. 5344-5354, Nov. 2022. doi: 10.1109/JBHI.2022.3197666.
[21] F. Trad and S. El Falou, "Towards Using Deep Reinforcement Learning for Better COVID-19 Vaccine Distribution Strategies," in Proceedings of the 2022 7th International Conference on Data Science and Machine Learning Applications (CDMA), Riyadh, Saudi Arabia, pp. 7-12, 2022. doi: 10.1109/CDMA54072.2022.00007.
[22] A. Singhal, P. Singh, B. Lall, and S. D. Joshi, "Modeling and Prediction of COVID-19 Pandemic using Gaussian Mixture Model," Chaos, Solitons & Fractals, vol. 138, pp. 110023, Sep. 2020. doi: 10.1016/j.chaos.2020.110023. PMID: 32565627; PMCID: PMC7296328.
[23] E. Kulah, Y. M. Çetinkaya, A. G. Ozer, and H. Alemdar, "COVID-19 Forecasting using Shifted Gaussian Mixture Model with Similarity-Based Estimation," Expert Systems with Applications, vol. 214, pp. 119034, Mar. 2023. doi: 10.1016/j.eswa.2022.119034. PMID: 36277990; PMCID: PMC9576929.
[24] M. Hamdi, I. Hilali-Jaghdam, B. E. Elnaim, and A. A. Elhag, "Forecasting and Classification of New Cases of COVID 19 Before Vaccination using Decision Trees and Gaussian Mixture Model," Alexandria Engineering Journal, vol. 62, pp. 327-333, Jan. 2023. doi: 10.1016/j.aej.2022.07.011. PMCID: PMC9263718.
[25] A. Marathe, A. Mandke, S. Sardeshmukh, and S. Sonawane, "Leveraging Natural Language Processing Algorithms to Understand the Impact of the COVID-19 Pandemic and Related Policies on Public Sentiment in India," in Proceedings of the 2021 International Conference on Communication Information and Computing Technology (ICCICT), Mumbai, India, pp. 1-5, 2021. doi: 10.1109/ICCICT50803.2021.9510070.
[26] R. Tang, L. Zhang, G. Zhang, and J. Wang, "Analysis of COVID-19 Rebound Based on Natural Language Processing," in Proceedings of the 2021 6th International Conference on Intelligent Computing and Signal Processing (ICSP), Xi'an, China, pp. 333-336, 2021P.
[27] Bose, S. Roy and P. Ghosh, "A Comparative NLP-Based Study on the Current Trends and Future Directions in COVID-19 Research," in IEEE Access, vol. 9, pp. 78341-78355, 2021, doi: 10.1109/ACCESS.2021.3082108.
Fredrick Ishengoma is a lecturer at the Department of Information Systems and Technology (IST), College of Informatics, the University of Dodoma, Tanzania. He holds a PhD of Information Systems from the University of Dodoma, Tanzania, a Master's degree in Computer and Information Engineering (CIE) from Daegu University, South Korea, and a Bachelor degree in Information and Communication Technology Management (ICTM) from Mzumbe University, Tanzania. His research interests include e -Government,
ICT, and ICT4D. email: ishengomaf@gmail.com
