Big Data are characterised by greater volumes of data from a greater variety of sources which are produced and processed at greater velocity. Huge digitised datasets from electronic medical records, registries, administrative datasets and genomic databanks can now be analysed by advanced computer programs to reveal patterns, trends and associations previously indiscernible using conventional analytic methods. These new insights may have important implications for clinical care. But Big Data can be limited by inaccuracies and bias inherent to observational datasets and which cannot be eliminated simply by using ever enlarging data sets or more sophisticated software. The hope and hype of Big Data cannot be allowed to override potential for harm, and the need for concurrent development of new research designs, better analytic methods and rigorous evaluation of predictive accuracy and effects on care and outcomes.
Keyphrases
- big data
- artificial intelligence
- machine learning
- healthcare
- palliative care
- rna seq
- deep learning
- quality improvement
- single cell
- public health
- pain management
- affordable care act
- adipose tissue
- drinking water
- chronic pain
- blood flow
- radiation therapy
- squamous cell carcinoma
- gene expression
- climate change
- data analysis
- electronic health record