Computational methods for studying relationship between nutritional status and respiratory viral diseases: a systematic review

This study sought to identify the computational methods used in studying the relationship between nutritional status and human respiratory viral infections. Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline, we searched the PubMed database for studies that used computational approaches to investigate the nutritional determinants of respiratory viral infections. Sixty seven (67) studies were selected after screening an initial 1713 search hits against the study eligibility criteria. Our findings revealed that machine learning, neural network (and deep learning), mathematical models, and statistical methods were used by 83.58%, 26.87%, 13.43% and 92.54% respectively while 19.40% of studies use other unconventional methods. Furthermore, 16.42% and 71.64% studies use formerly created datasets and raw data respectively while 11.94% studies were performed without prior datasets. The outcomes of the selected studies showed that task related to prediction, identification, investigation, association and determination were performed by 20.90%, 8.96%, 40.30%, 10.93% and 14.93% of studies respectively. 1.49% each of the studies performed recommendation, risk reduction, management and screening. The findings of this review answered three research questions and provided guidance for some future contributions in the domain of studies related to nutritional status and respiratory viral disease.