Big Data, Artificial, Intelligence, and Brain Health Flashcards
(12 cards)
5 features of Big Data
Volume: number of data points
Variety: data may cross different types (structured/unstructured)
Velocity: pace of data generation
Veracity: data quality and accuracy
Value: potential to create benefits and insights
Types of health data collected in BC
Patient data
Opeerations data
Population data
Social determinants of health data
Artificial intelligence definition
Artificial systems that appear to think like humans
Machine learning definition and 2 types
Systems that can learn from experience or data without human programming
Supervised learning: models are trained on known, labeled data. Requires huge volume of data and human labour
Unsupervised learning: models learn from unlabeled data. Requires huge processing power
AI Prognoses
Predicting alzheimers using brain scans.
Training the ML model using MRI data to predict AD.
Identifies the most predictive brain regions. Found that hippocampus and amygdala to be most predictive.
Clinical decision making
Surgically remove epileptogenic brain region to treat seizures.
Proposed ML model uses unlabeled features of the raw iEEG output to identify seizure origin
Neurotech
Controlling limb prosthesis with neural activity
Training ML on the map between activity and limb movement
Ethical issues
Accountability, moral accountability, bias/discrimination, privacy
Accountability
AI introduces harm that no one person can predict or prevent
Moral Accountability
Duty to explain one’s reasons and actions to others…
AI processes may be unexplainable to their users - “black box”
Bias and discrimination
Groups that are under-represented in AI models may receive lower quality care
Too many white participants: over-representation
Groups that are overrepresented/under will skew ML or AI models
Privacy
AI outputs can be sensitive… there are questions as to how this data should be controlled, especially when predictions include uncertainty
