Modern Machine-Learning Predictive Models for Diagnosing Infectious Diseases

Table 3

Dataset specifications for the reviewed articles.

Reference	Size	Data acquisition method	Infectious disease
[19]	Time-series of physiological data	Wearable sensors to collect multivariate physiological data	Tetanus and HFMD
[20]	60 individuals	Medical sensors, hub, and Android-based app to collect vital signs	Skin and soft tissue infection, urinary tract infection, and acute respiratory infection
[21]	49,721 users	App-based symptom tracker	SARS-CoV-2 (COVID-19)
[22]	88 individuals	Cellular phone voice recordings	SARS-CoV-2 (COVID-19)
[23]	37,599 tweets	Social media messages	Latent infectious diseases
[24]	1,317,018 classes, 7,731,914 axioms, and 1,269,340 inheritance relations	Multiple medical ontologies	507 infectious diseases
[25]	31268 reports	NLP tool (Topaz) to extract influenza-related findings	Influenza
[26]	52,306 patients	Routine blood tests	SARS-CoV-2 (COVID-19)
[27]	1391 patients	Routine laboratory results	SARS-CoV-2 (COVID-19)
[28]	295 patients	Clinical records	Candidemia
[29]	1118 patients	EHR	CDI
[30]	152 patients	Clinical records and chest CT images	SARS-CoV-2 (COVID-19)
[31]	2482 images	CT scan images	SARS-CoV-2 (COVID-19)
[32]	56,081 patients	Historical and real-time data	25 infectious diseases

SARS-CoV-2: severe acute respiratory syndrome coronavirus 2; EHR: electronic health record; HFMD: hand, foot, and mouth disease; CDI: Clostridium (Clostridioides) difficile infection; CT: computed tomography.