COVID and perioSir, in February 2021, Marouf et al. discovered a strong association between periodontitis and severity of COVID-19 infection with 568 patients. 1 Shamsoddin used the same data from the Hamad Medical Corporation (HMC) in the state of Qatar for confirmation of the Marouf et al. study and supported its conclusions. 2 However, Shamsoddin suggested that a more rigorous methodology, an unbiased approach, and studies on a broader scale (external validity, generalisability) would be needed to clarify the relationship between periodontitis and COVID-19. 2 Meta-analyses of epidemiological studies have shown that patients with periodontitis are more likely to have a more severe course of COVID-19. 3 Current evidence suggests that this association can be explained by the direct role of periodontitis in exacerbating pulmonary infections and the indirect effect of periodontitis in triggering systemic inflammation and priming the immune system to increase the response to severe COVID-19 infection. 4 Mishra et al. showed that the probability of developing severe COVID-19 was 2.81 times higher in patients with periodontitis. There is an association between periodontitis and severe COVID-19. 5 The existing studies showed that periodontitis is strongly associated with increased risk of intensive care unit (ICU) admission and increased blood levels of biomarkers associated with poor prognosis. The result suggested that while it is wise not to get COVID-19, it is wiser to reduce periodontitis with regular oral care.

Tree-structured LSTM (Long Short-Term Memory) is a promising concept to consider long-distance interaction over hierarchies with syntactic information. Besides, compared with chain-structured one, tree-structured LSTM has better modularity of learning process. However, there still remains the challenge concerning hyperparameter tuning in treestructured LSTM. Mainly, hyperparameter of mini-batch SGD (Stochastic Gradient Descent) is one of the most important factors which decides the quality of the prediction of LSTM. For more sophisticated hyperparameter tuning of mini-batch SGD, we propose a constrained recursion algorithm of tree-structured LSTM. Our algorithm enables the program to generate an LSTM tree for each batch. By doing this, we can evaluate the tuning of hyperparameter of mini-batch size more correctly compared with chain-structured one. Besides, our constrained recursion algorithm can traverse the LSTM and update the weights over several LSTM tree with a breadth-first search. In the experiment, we have measured the validation loss and elapsed time in changing the size of mini-batch. We have succeeded in measuring the learning process's stability with small batch size and the instability of overfitting with large batch size more precisely than chain-structured LSTM.

This paper investigated impact of COVID-19 on mortality under 5 years old. The mortality effects were calculated with two CDC datasets such as the dataset from 2015 to 2020 and the provisional dataset from 2020 to 2022. The result shows that there is no effect of COVID-19 on mortality under 5 years old.