Artificial Intelligence in ANESTHESIA
Artificial Intelligence in ANESTHESIA
in Anaesthesia
Current Trends & Future Prospects
Robot
Anesthetist
Outlines
• Audience Poll
• Introduction and Definition of AI
• History and Evolution of AI
• Stages and Types of AI
• How AI works and Uses of AI
• History of AI in Healthcare
• History of AI in Anesthesiology
• Current status of AI in Anesthesiology
• Future of AI in Anesthesiology
• Will AI replace Anesthesiologist
• Journal Articles of AI in Anesthesia
• Conclusion
• Take home message
• My Verdict
• Cartoons & Thanks
Tusha
• Artificial Intelligence
• Quantum Computing
Artificial intelligence is the capability of computers to
respond in a manner resembling human intelligence
Human Computer
Dictionary Definition
Tusha
53%
40%
AI is like Electricity
Without electricity we can’t think this world Today
Like this
Without AI we won’t think this world in Future
Tusha
Fathers
of
AI
To studies of how
An approach to make
human brain thinks,
computer, robot, or
learns, decides and
product to think how
works, when it tries to
smart human think
solve problems
To improve computer
functions which are The ultimate aim is
related to human technological
knowledge, for singularity the point at
example, reasoning, which technology
learning, and problem- overtakes the human
solving
Artificial Intelligence (AI), (John McCarthy)Tusha
Is the capability of computers to respond in a
manner resembling human intelligence (Machine to
thinks without any human intervention) examples are
Drones, Self driving cars, Search engines like Google
and Medical diagnosis with procedures e.g. Robots in
medicine
AI
1956
DS Machine Learning (ML), (Arthur Samuel)
Is a subset of artificial intelligence (AI) that
Data Science provides systems the ability to automatically
is a multi-disciplinary field learn and improve from experience without
that uses scientific methods,
processes, algorithms and
ML 1959
being explicitly programmed (Provides
statistical tools to explore data, so here
systems to extract knowledge machine learns automatically from past data
(It applies all mathematical without any confusion of any programme)
rules)
DL
1965
Deep Learning (DL), (Alexey Ivakhnenko )
Is a subset of machine learning where artificial neural
networks, algorithms inspired by the human
brain, learn from large amounts of data (Human
brain thinking, also called as Deep Neural Network or
Deep Neural Learning)
Tusha
Artificial Intelligence
AI Machine Learning
Subset of AI
ML
Deep Learning
DS DL Subset of ML
Data Science
All three
AI + ML + DL
DATA
SCIENCE
2020’s 2030’s
Tush AI in the 1970s AI in the 1980s
ar gave way to accelerated The rapid growth of artificial
advancements, particularly intelligence continued
focusing on robots and through the 1980s. Mercedes-
automatons. WABOT-1, the Benz built and released a
first anthropomorphic robot, driverless van equipped with
was built in Japan at Waseda cameras and sensors
University. Its features
History of AI included moveable limbs,
ability to see, and ability to AI in the 1990s
converse
The end of the millennium has
helped artificial intelligence in its
AI in the 1960s continued stages of advance
growth. Deep Blue, a chess-playing
Innovation in the field of computer developed by IBM became
artificial intelligence grew the first system to win a chess game
AI from 380 BC to 1900 and match against a reigning world
rapidly through the 1960s. champion Sony introduced AIBO
Various mathematicians, Unimate, the first industrial
theologians, philosophers, (Artificial Intelligence RoBOt), a
robot was developed to work robotic pet dog
professors, and authors on a General
mused about mechanical Motors assembly
techniques, calculating
machines, and numeral AI from 2000-2010
systems AI continued its trending
AI in the 1950s upward. Professor Cynthia
Breazeal developed Kismet, a
Advances in the field of robot that could recognize
AI from 1900-1950 artificial intelligence came. and simulate emotions with
Claude Shannon, “the father of its face. Honda releases
People took the “robot” idea information theory,” published
and implemented it into their ASIMO, an artificially
“Programming a Computer for
research, art, and discoveries Playing Chess. In 1956 John intelligent humanoid robot
and in 1929 the first robot McCarthy gave the official birth
was built in Japan of the word AI
AI 2010 to Present Day Tusha
From 2010 onward, artificial
The last decade was In Smartphone having voice assistants
intelligence has become
immensely important for AI and computers that have
embedded in our day-to-day
innovation “intelligence” functions are everything
existence
through advance AI
2015-2017 2016
2010
Google released Google
Microsoft launched Kinect for Google DeepMind’s Alpha Go,
Home, a smart speaker that
Xbox 360, the first gaming a computer program that
uses AI to act as a “personal
device that tracked human plays the board game Go,
assistant” to help users
body movement using a 3D defeated various (human)
remember tasks, create
camera and infrared detection champions
appointments, and search for
information by voice.
2011 2015
Apple released Siri, a virtual Elon Musk, Stephen Hawking,
2017
assistant on Apple iOS operating and Steve Wozniak among The Face book Artificial
systems. Siri uses a natural-
3,000 others started the Intelligence Research lab
language user interface to infer,
observe, answer, and
development and use of introduced chatbots to
recommend things to its human autonomous weapons for communicate with one
user wars and autonomous cars another
2014 2018
2014
Samsung introduced Bixby, a
Microsoft released Amazon created Amazon
virtual assistant. Bixby’s
Alexa, a home assistant that
Cortana, their version of a developed into smart speakers
functions include Voice, where
virtual assistant similar to the user can speak to and ask
that function as personal
Siri on iOS questions, recommendations,
assistants
and suggestions
Tusha
Landmark Years in
Landmark Year of
2011 AI in Anesthesiology
Tusha
Natural
language
processing
(NLP)
Machine
Chatbot + What to Learning and
expect for AI
Virtual in 2020
Automated
assistants and beyond ?
Machine
Learning
Autonomous
vehicles
Tusha
of AI
1) Artificial Narrow Intelligence (ANI)
Weak AI
- Machines has a narrow defined role of task
- e.g. SiRi, Alexa, Sophia, Self Driven Car
2) Artificial General Intelligence (AGI)
Strong AI
- Machine starts thinking just like humans
- under creation and developing all over world
3) Artificial Super Intelligence (ASI)
Super Strong AI
- Computers or Machines will surpass human beings
- Not exist and only seen in science fiction movies
Tusha
Reactive Machines AI Self-Awareness AI
Involves machines that have human-level
A basic type of AI system, they consciousness. This form of AI is not
can only react to currently currently in existence, but would be
existing situations. e.g. Deep considered the most advanced form of
Blue, a chess-playing artificial intelligence known to man and
supercomputer created by IBM.
(So it works on preset data) (Pray to GOD it will not exist)
Four Types
of
Artificial
Intelligence
Limited Memory AI Theory of Mind AI
Comprises of machine learning Is the decision-making ability equal
models that derive knowledge to the extent of a human mind, but
from previously-learned by machines. Two notable examples
information, stored data, or are the robots Kismet and Sophia,
events. e.g. Autonomous created in 2000 and 2016.
vehicles, or self-driving cars. (So it works on emotional
(So it works on past data) intelligence)
Tusha
How AI Works
AI
Tusha
Uses of AI
Aeronautics
Construction Metal Industries
Automobiles
Robotics
Architect
Factories
Air-Condition
Windmill
Medicine
Solar Energy
Computers
Smartphone Electricity
History of AI in Healthcare
First problem-solving program, or expert system, known as Dendral and
1960-70 considered one of the most significant early uses of artificial intelligence
in medicine
Uses of AI in Healthcare
Tusha
Uses in Healthcare
Artificial Intelligence
Tusha
Artificial Intelligence
In
Anesthesiology
In historical development, anesthesia was the
earliest subject to implement artificial
intelligence
Current Status
of
AI in Anaesthesiology
Currently AI
Removal of the ET tube
involved in all
at the conclusion of Anesthetizing Laboratory evaluation
most surgeries
patient, who
requires
Observation of all vital Mask ventilation of
monitors during surgery an unconscious patient
The diagnosis
and treatment of
any complication
during or
following the
anesthetic
Tusha
From onwards
Anesthesia Practice
with AI will be
Tusha
• e.g. English Speech ---> English Text ---> Local text --->
Local Speech
Contd.
• Our future Anaesthesia work station will be
operated by AI with just voice command & data
entry
• Monitors will be controlled and self checked by AI
• Switch On and Switch Off Anaesthesia will become
reality through AI in near future
AI in Tele Anaesthesia
• With AI you will be virtually present with patient and
colleague anaesthesiologists to guide them in their Peri
operative period and in all procedures
• Even you can control all OT gadgets with your Smartphone
AI Apps
• Tele anaesthesia is not a new concept but combined with AI
it is very useful for teaching, in CMEs and Conferences (e. g.
my webinar talk)
• Combined with Augmented Reality(AR) and AI will change
the concept of Tele Anaesthesia, and you will be giving
virtual anaesthesia to distant OT patients (and it’s a reality
e.g. Tele anaesthesia from Canada to Italy by Dr Hammerling
and colleagues)
Tusha
In Future
from 2025 Onwards
We
Tusha
2025 Onwards
Tusha
Augmented Reality
Tusha
AI Anaesthesiology
AI will work in Anaesthesiology as a Specialty
• Anaesthesiologists need a good mix of cognitive and
dexterity based labour
• AI will primarily result in the automation of cognitive work,
it may be that our hands prevent full automation of the
specialty
• The general dexterity that humans possess allows for a wide
range of functional interaction with our environment
• As AI-based automation systems gain further capability, they
may be able to perform semiautonomous anesthesia
maintenance, where the AI-enabled machine takes over
specified domains of anesthesia maintenance
Tusha
Contd.
• Anaesthesia might never be fully automated because it
involves dexterity-based labour
• Current robotic devices don't have exact dexterity
required for tasks
• However, AI could be used in anesthesia to develop
more advanced clinical decision support tools based on
machine learning
• Ultimately, AI and machine learning could enable
anesthesia to become a true perioperative medicine
specialty rather than just an intraoperative specialty
• AI could assume some of anaesthesiologists' cognitive
workload and support "a renewed emphasis on the
doctor-patient relationship,"
Tusha
Will AI reduce
Errors in Anesthesiology ?
• Depends on Safety Vs. Complexity (Complication)
• Human dexterity (skill hands) works will not be
replaced by AI Anaesthesia or Robotic Anaesthesia
• AI can control the gadgets, monitors but
anaesthesiologist will require all the time to do
dexterity works like to fill the syringes, to put chest
electrodes or to take clinical decision in odd situations
Tusha
that
But
Tusha
Will AI replace Anesthesiologist ?
Like when the patient‘s well being deteriorates in the OT, almost all
surgeons want an anesthesiologist being to discuss what it will take
to get the patient stable again.
Talking to a robot or machine (no matter how smart it is) just isn’t
the same as addressing the guy who’s been doing anesthesia for
your patients for long time
Tusha
>> On the flip side, we should start actively using and testing AI that is available in order
to experience it and learn to corroborate it in our anaesthesia practice
>> The AI of today will help anaesthesiologists to skip the boring stuff and hopefully
make them more focused and smarter. If you have a talented team and the resources to
play with intelligent monitors and anaesthesia machines, you will get the best results
>> However, in the future we might see particular roles being replaced. The first people
to be let go will likely be those roles that are very task-based and lack creativity
>> What is predictable is the intention for AI development. Five years from now, we
might see amazing game-changing inventions as well as continuous steady progress
towards more intelligent, self-thinking monitors, anaesthesia machines and TCI pumps
Tusha
Intelligent control in Anaesthesia
(super super speciality)
• Intelligent control is a new and interdisciplinary subject
developed on multiple directions, such as artificial
intelligence and automatic control
• In 1967, Leonde and Mendel first used the concept of
"intelligent control" to apply techniques of memory and
goal decomposition to improve their ability to deal with
uncertainty
• The intelligent control summarized as the combination of
automatic control and artificial intelligence
• Closed loop TCI with anaesthesia depth monitor is
intelligent control system ( Auto Anaesthesia )
• It is widely used in clinical applications, including cardiac
surgery, pheochromocytoma surgery, gastrointestinal
surgery and has had good clinical results
Tusha
Articles
from
Tusha
CONCLUSIONS
Artificial intelligence has the potential to impact the practice of anesthesiology in aspects
ranging from perioperative support to critical care delivery to outpatient pain management
https://www.ncbi.nlm.nih.gov/pubmed/31939856
Recent advances in the technology of anesthesia
F1000Res v.9; 2020 PMC7236591, 2020 May
Christian Seger, Maxime Cannesson,
Conclusions
Carlson JN, Das S, De la Torre F, Frisch A, Guyette FX, Hodgins JK, Yealy DM
CONCLUSIONS
https://www.ncbi.nlm.nih.gov/pubmed/26986814
Tusha
CONCLUSIONS
https://www.ncbi.nlm.nih.gov/pubmed/28883203
Tusha
Machine Learning Approach for Preoperative Anaesthetic Risk Prediction
Int. J. of Recent Trends in Engineering and Technology, Vol. 1, No. 2, Nov 2009
CONCLUSIONS
In the research work, three supervised machine learning schemes were applied on the
preoperative assessment data to predict the anaesthetic risk of the patients and the
performance of the learning methods were evaluated based on their predictive
accuracy and ease of learning
Tusha
Prediction of Bispectral Index during Target-controlled Infusion of
Propofol and Remifentanil: (A Deep Learning Approach)
Hyung-Chul Lee, M.D.; Ho-Geol Ryu, M.D., Ph.D.; Eun-Jin Chung, M.D.; Chul-Woo Jung, M.D., Ph.D.
Conclusion
Our study demonstrated that the deep learning model is superior to traditional
PK–PD model in predicting BIS during Propofol and Remifentanil target-controlled
infusions in surgical patients. The major advantage of the deep learning approach
is its performance and extensibility. We expect that the accumulation of clinical
big data will make the deep learning model more powerful and extend its
application to a variety of clinical situations in the future.
https://anesthesiology.pubs.asahq.org/article.aspx?articleid=2656314
Tusha
Automated anesthesia carts reduce drug recording
errors in medication administrations - A single centre
study in the largest tertiary referral hospital in China
Wang Y, Du Y, Zhao , Ren Y, Zhang W.
J Clin Anesth. 2017 Aug;40:11-15. doi: 10.1016/j.jclinane.2017.03.051. Epub 2017 Apr 13.
The total error rate was 7.3% with the automated anesthesia carts (1 in 14
administrations) and 11.9% with conventional manual carts (1 in 8 administrations)
CONCLUSIONS
https://www.ncbi.nlm.nih.gov/pubmed/28625429
Tusha
First robotic tracheal intubations in humans
using the Kepler intubation system (KIS)
BJA: British Journal of Anaesthesia, Volume 108, Issue 6, June 2012, Pages 1011–
1016, https://doi.org/10.1093/bja/aes034
T. M. Hemmerling R. Taddei M. WehbeC. Zaouter S. Cyr J. Morse
Conclusions
We present the first human testing of a robotic intubation system for oral
tracheal intubation. The success rate was high at 91%. Future studies are
needed to assess the performance and safety of such a system
aes034
Tusha
Technical communication:
First robotic ultrasound-guided nerve blocks in humans using the Magellan system
Anesth Analg. 2013 Feb;116(2):491-4. doi: 10.1213/ANE.0b013e3182713b49. Epub 2013 Jan 9.
Conclusions
https://www.ncbi.nlm.nih.gov/pubmed/23302984
Tusha
"We have been working on closed-loop systems, where drugs are administered, their
effects continuously monitored, and the doses are adjusted accordingly, for the last five
years,” said Dr. Thomas M. Hemmerling of McGill’s Department of Aesthesia and the
Montreal General Hospital, who heads ITAG (Intelligent Technology in Aesthesia research
group), a team of anesthesiologists, biomedical scientists and engineers.
Father of
Robotic Anesthesia
Tusha
Pharmacologic Robots
Closed loop for hypnosis
Closed loop of Anaesthesia
Closed loop of Analgesia
Anesthesia robots to aid or replace manual gestures
Teleanesthesia
Regional anaesthesia procedure
Placing a perineural catheter for continuous nerve block
Robotic Intubation
Artificial intelligence
Automated Assessment of Difficult Airway With Facial Tusha
Recognition Techniques (PeScho)
https://clinicaltrials.gov/ct2/show/NCT02022397
Responsible Patrick Schoettker,MD PD, Associate Professor, University of Lausanne Hospitals
Party:
ClinicalTrials.gov NCT02022397
Identifier:
Other Study ID 183/09
Numbers: CTI ( Other Grant/Funding Number: Swiss Commission Technology and Innovation
12636.1 )
Advances in technology and monitoring can change the impetus for machine learning.
For example, a neural network developed to detect esophageal intubation from flow-
loop parameters and will be obviated by continuous capnography
Uses a very highly augmented data set in conjunction with logistic regression to produce
an algorithmic model that can, in post hoc analysis, detect the incipient onset of
hypotension up to 15 min before hypotension actually occurs
Neural network approach to predicting the Bispectral Index (BIS) based upon the
infusion history of propofol and remifentanil
The most plausible route to the introduction of artificial intelligence and machine
learning into anesthetic practice is that the routine intraoperative management of
patients will begin to be handed off to closed-loop control algorithms
Tusha
SEDASYS
Ethicon Endo-Surgery, a Johnson & Johnson
subsidiary, created SEDASYS, a computer-assisted
device that administers the prescription drug
propofol into the bloodstream via intravenous IV
infusion (approved in 2013)
In this pilot study, we tested a novel automatic anesthesia system for closed loop
administration of IV anesthesia drugs for cardiac surgical procedures with
cardiopulmonary bypass. This anesthesia drug delivery robot integrates all 3 components
of general anesthesia: hypnosis, analgesia, and muscle relaxation
CONCLUSIONS:
67
Tusha
Tusha
Today’s Robotic and Artificial Intelligence Theatre
Tusha
In 2030
AI will replace this type
of complex anaesthesia monitors and machine
into
Tusha
Tusha
With AI
Will go away
4)
AI will definitely reduce the errors in
anaesthesia practice
2025 2030
Tusha
BLOCKS
In Future
Anesthesiologist will be the assistant of AI Robots
Tusha
https://sites.google.com/site/tusharchokshisite
9825062245/9979319721
Tushar