Biomedical Data Pre-processing:

  Data can be consented if patients agreed to enter a specific study, or unconsented otherwise: use of unconsented data has additional limitations and privacy issues. Patient information: date of birth, sex, date of study entry/exit Routine medical data: height, weight, blood pressure, cholesterol levels, and medications used Specialized laboratory data: proteins, lipids, metabolites, glycans, imagine Genetic data: genotype or sequencing Gene expressions and epigenetic data (DNA methylation) Covariates of a model are also called exposures in epidemiology. They may represent an individual attribute, behaviour, an event, an environment exposure, etc.  Quantitative:  Discrete: typically counts (number of cigarettes smoked, number of visits to a GP)  Continuous: height, weight, cholesterol Categorical: nominal: sex, ethnic origin, blood group ordinal: disease stage, degree of pain Sometimes the distinction is a bit blurred: age is a continuous variable, but effectively it may be analysed as discrete. A quantitative variable may be summarised into categories which have use in clinical practice and policy guidelines. Renal function can be well assessed in terms of glomerular filtration rate (GFR). This is difficult to measure, so instead this is usually estimated from other easy to measure quantities (eGFR).

High Impact List of Articles

Relevant Topics in General Science