vocab Flashcards
What term describes having multiple long-term conditions?
Multimorbidity
What was the Turing test?
Game where interrogator differentiates between a human and a computer based on questions
What did the Turing test predict?
Within 50 years, computers would win 70% of the time
What does MAAS stand for?
Multi-agency advice service
What 4 organizations does MAAS join together?
NICE, MHRA, HRA, CQC
What is it called when too many different variables affect predictions?
Overfitting
What is synthetic data and what’s its purpose?
Fake data to protect privacy
What are 2 examples of front-end?
Patient registration and insurance verification
What is included in the mid-cycle?
Coding of claims
What is included in the back-end?
Billing and collections
What is the reconciliation of different data sources into a single output called?
Data harmonization
What is palliative care?
Care for patients with serious illnesses
What does ICO stand for? what does it do?
Information commissioner’s office; makes guidelines on how data can be collected and used
What is frailty (as defined by British Geriatrics Society)?
State of health in aging where body systems lose their in-built reserves and cannot cope with illnesses
Is frailty higher in men or women?
Women
Are frailty and prefrailty higher in older adults in upper middle-income countries or high-income countries?
Upper middle
What is brittleness as it refers to algorithms?
When algorithms cannot generalize outside a narrow set of conditions
What is digital exclusion?
When people don’t have the skills to use digital devices or access to technology
What is semi supervised learning?
Using a small, labelled dataset to learn and apply those to a larger unlabelled dataset
What is drift?
AI could change due to the learning nature of the algorithms, so you need to constantly update it
What is generalisability?
applicability of a study to other contexts (refers to the extent to which a model made in one place can be run in another place)
What is interopability?
not being limited to a certain vendor’s predictive model (which may have been trained with a specific set of data)
What is cardioversion?
a procedure to restore the normal rhythm of the heart; nurse needs to be there before and after to take the measurement but if they miss it, they lose the patient from clinical studies
What is metadata?
references and information about the characteristics of data
What is transfer learning?
how was it used?
training on a large dataset and then retraining the model on a smaller specific dataset → improves generalization of an AI
Researchers in India used transfer learning to identify kidneys in ultrasound scans using only 45 training examples
what is good about dynamic procurement systems (DPS)?
align better with the fast-changing environment of AI digital solutions
DAIS
data, analysis, and intelligence service
OMOP data model
structures and harmonizes healthcare data sources (e.g. EHRs) into a database
