Table 3

Differentiating data-driven, machine learning approaches compared with statistical approaches

ModalitiesMedicineData-driven medicine
Analytical methodsStatistical methodsMachine learning or deep learning
Analytics strategyHypothesis drivenData driven
Data complexityLow dimensionalHigh dimensional
Data modelStable data modelEvolving, flat data model
Data sizeMegabytes or gigabytesTerabytes, petabytes, exabytes
Data sourceCentralisedDecentralised
Data sourceStructuredStructured, semistructured or unstructured
Data storageExcel or MySQLMySQL, NoSQL or Graph databases
Data typesTraditional data typesEvolving data types
Example methodStudent’s t-testRandom Forests
Example softwareExcel or SASR, Python
Sample sizeCohort size (~10 K)Large cohort size (>100 K)