This document provides an introduction to artificial intelligence (AI). It discusses the history of AI, including its origins in the 1950s and important figures like Alan Turing. It also covers the history of AI in medicine, such as early work in the 1960s and IBM's Deep Blue defeating Gary Kasparov in 1997. The document defines different types of AI systems and provides examples. It discusses the current applications of AI in cardiology, such as digital stethoscopes, ECG and echo image analysis, and AI use in intensive care. In conclusion, it states that AI is already here and will significantly impact fields involving imaging and data, while creating innovations that need to be adopted early in clinical practice.
2. What is AI?
artificial intelligence is the science of making
machines do things that would require
intelligence if done by humans
…….Marvin Minsky
3. history
Alan Turing, British mathematician and
computer scientist - progenitor of AI
Decoding the secret messages sent via german
enigma machine during WW-II using machine
intelligence
5. history
AI - coined by stanford scientist John McCarthy
at the dartmouth conference in 1956.
Frank Rosenblatt (1958) - the perceptron
a 3-layer structure - early precursor of the
artificial neural network and present day deep
learning
6. history of ai in medicine
earliest work - 1960s
edward shortliffe - innovative heuristic
programming project MYCIN
7. history of ai in medicine
Artificial intelligence failure:
lack of favourable work flow logistics
lack of integration to accommodate the
clinicians, and, also
due to expectations that were unrealistically
high but never met.
8. history of ai in medicine
renewed focus on data mining and machine learning - 1990’s
IBM supercomputer DEEP BLUE defeated Gary Kasparov in
chess (1997)
Three main factors heralding the new AI era:
increasingly large volumes of available data that requires new
computational methodologies (Big Data),
the escalating capability of computational power and cloud
computing
the emergence of machine and deep learning
10. A, B, C, D of AI
Algorithms: sets of steps to accomplish certain
tasks
calculation
data processing
automated reasoning
e.g. NASA’s solar panel operation on ISS
11. A, B, C, D of AI
Big Data: escalated in such exponential way to make
conventional data processing applications inadequate
4 V’s:
Volume
Variety
Velocity
Veracity
12. A, B, C, D of AI
Cognitive computing:
machine learning
supervised learning
unsupervised learning
reinforcement learning
13. A, B, C, D of AI
deep learning
convoluted neural network
recurrent neural network
natural language processing
Internet of things (IoT)
14. Types of AI
type definition
level of human
involvement
examples
assisted
system providing &
automating repetitive
tasks
little or none
UR ROBOTS for
blood work
augmented
humans & machines
together make
decisions
some or high
IBM WATSON for
oncology
autonomous
decisions made by
adaptive intelligent
systems
autonomously
little or none
IDx-DR
diagnosing retinal
images
15. current status of ai in cardiology
clinical domain: digital stethoscope
imaging domain:
ecg analysis
echo imaging
ct/ mr imaging
intensive care domain
25. AI in intensive care
big data
machine learning
pattern recognition
deep learning or rnn to
analyse & make
decisions
26. To conclude
ai is here and we cannot escape it
imaging and data rich fields will be significantly
impacted
ai will take away jobs…..unlikely
innovations need to be accepted early in clinical
practice…..before they make us primitive (e.g.
nokia, kodak)