types of AI Flashcards
Deep learning
Uses artificial neural networks to analyze large datasets
Neural network
Uses interconnected notes to analyze data and identify patterns
Which type of AI gives more complex and accurate predictions?
Deep learning
Which type of AI mimics human decision-making processes?
Rule-based and RPA systems
What have rule-based systems been able to do?
Analyze electrocardiogram data to diagnose heart disease
What are the two main advantages of rule-based systems?
Transparency/trust, ease
What are the two main limitations of rule-based systems?
Need accurate rules, cannot address ambiguity
RPA
Robotic process automation; automates repetitive tasks by mimicking humans
What does RPA focus on?
Administrative tasks
Which type of AI works with existing interfaces of IT infrastructure and can work without interruptions?
RPA
RPA is bad for tasks with ___ and ___
Complex decision making; unstructured data
Machine learning
Ability for computer to learn after given instructions
What is the main function of machine learning?
Make predictions
What type of AI are computer-aided diagnosis systems used to interpret medical images?
Machine learning
Supervised learning
Uses a labeled dataset where the correct answer is given
Unsupervised learning
Does not use a labeled data set but identifies patterns from the dataset
NLP
Natural language processing; analyzes human language
2 examples of NLP?
Analyze EHRs and patient notes, reduces fraud in billing
What type of AI is used in surgical procedures, rehab, and patient monitoring?
Robotics
Which type of AI works with surgeons to provide real-time feedback and improve precision?
Robotics
Expert systems replicate ___ abilities of human experts
Decision-making
CDSS
Clinical decision support system
Example of expert system use?
Makes real-time recommendations for treatments and alerts with patient data
Computer vision
Processes visuals
Federated learning
Machine is trained through multiple datasets
What type of AI encompasses: synthesizing medical images, generating new chemical structures, running simulations to test treatment strategies on AI
Generative AI
STSD system
Socio-technical systems design; considers human, social, and organizational factors
Support vector machines
Machine learning used for classification/regression analysis
How do SVMs classify tasks?
Identifies hyperplanes in high-dimensional space
CNNs
Convolutional neural networks; type of deep learning that identifies patterns
Are CNNs good for image analysis?
Yes
RNNs
Recurrent neural networks; type of deep learning that uses sequential data analysis
What type of AI are image reconstruction, ECG analysis
RNN
Decision trees
Type of machine learning that classifies tasks
3 types of CDSS?
Knowledge-based, ML-based, NLP-based
acronym KMN
How do knowledge-based systems work?
Uses a set of rules to analyze data
Use of ML-based systems?
Make predictions based on patterns
Use of NLP-based systems
Read medical records and patient notes to make decisions
acronym MP
The Turing Report stated that of the 647 AI tools used to support the COVID-19 pandemic, ___ were fit for clinical use, only ___ warranted further potential
0; 2
3 general categories of applications of AI in healthcare
Non-clinical, increasing productivity, enhancing delivery of care
How can AI be used in the back office?
Finance management, revenue cycle management, detecting inaccuracies in billing
acronym FRB
How can AI enhance the delivery of care?
Helps patients optimise treatment options and medical dosages
McKinsey Global Institute suggested that only ___% of the time in healthcare is fit for automation
35
3 fields in order with the most amount of time that can be automated
Medical equipment preparers, medical assistants, OSHA technicians
acronym EAO
3 fields in order with the LEAST amount of time that can be automated
Chiropractors, orthotist, dental hygienist
acronym COD
How can AI help in medical imaging?
Identifying subtle patterns in x-rays, CT scans, MRIs and create personalized treatments
6 steps for drug discovery with AI
Target identification
lead discovery
lead optimization
preclinical testing
clinical testing
regulatory approval
1 pro of AI drug discovery?
Cheaper because of fewer failed clinical trials
2 cons of AI drug discovery?
Hard to find rare diseases to obtain data; large pharmaceutical companies already have established workflows
How can NLP and RPA help in administrative tasks?
Reduce errors and improve accuracy in data collection and analysis
AI can identify errors and issues that lead to claim denials and thus ____
Improve claims management
NLP can help in administrative tasks by doing what?
Streamlining patient registration
How can predictive analytics help in healthcare?
Identify patients who are at risk of not paying their bills
What is PACS?
Picture archiving and communication systems in radiology
What is the most common example of a complementary technology?
RPA
