Maxime Cannesson, MD, PhD
Department of Anesthesiology & Perioperative Medicine
University of California Los Angeles
Closed Loop Anesthesia
Conflicts of Interest
Consultant for:
Edwards Lifesciences
Masimo Corp.
Shareholder:
Sironis
Perceptive Medical
Research Support:
Masimo Corp
Edwards Lifesciences
Outline
• Introduction
• Feedback Control / Closed Loop / Automation
• Examples of Closed Loop Systems in Anesthesia
• The Challenges Ahead
Outline
• Introduction
• Feedback Control / Closed Loop / Automation
• Examples of Closed Loop Systems in Anesthesia
• The Challenges Ahead
A&A 2022
‘Finally, no monitoring tool, no
matter how accurate, by itself
has improved patient outcome’
Outline
• Introduction
• Feedback Control / Closed Loop / Automation
• Examples of Closed Loop Systems in Anesthesia
• The Challenges Ahead
Biology / Physiology
Feedback Control
• “Closed Loop” is a generic term with no specific meaning
• “Physiology” and “Life” are based on Closed Loop controls
• The specificity is in the Sensor and in the Controller
Closed Loop Systems are Everywhere
• AC
• Cruise Control (speed regulators in modern cars)
• Electric Oven
• Elevators
• ….
The main unit pumps cool air throughout the house, the temperature is
monitored by sensors in the home, and this data is then compared to the
settings on the thermostat and used to increase or decrease the cooling so
that the temperature hovers near the set point. While it would be possible to
manually adjust the flow rate, the odds are that you would commonly
overshoot or undershoot your desired temperature and the system would
require many adjustments every hour to even modestly approximate the
performance the automatic controller easily achieves.
Air conditioning
Hemodynamic Monitor
Closed Loop Hemodynamic
BIS
ANI
Closed Loop
Anesthesia & Analgesia
Anesthesia Analgesia 2016
PCLC: Physiological Closed Loop Control
• Step 1: Creating Database (Memory)
Expert manual tracing of > 10,000 LV 4 ch and 2 ch views
• Step 2: Identifying LV cavity (Pattern recognition)
• Step 3: Tracing the endocardial border (Sensor)
• Step 4: Tracking the LV border (Processor)
• Step 5: Calculating LV volumes and EF (Processor)
Cannesson et al. J Am Coll Cardiol 2007
Cannesson et al. J Am Coll Cardiol 2007
Cannesson et al. J Am Coll Cardiol 2007
R
1
Cannesson et al. J Am Coll Cardiol 2007
R
2
Outline
• Introduction
• Feedback Control / Closed Loop / Automation
• Basics of Automation
• Examples of Automated Systems in Anesthesia
• The Challenges Ahead
Control CL
Depth of Anesthesia
0
20
40
60
80
100
120
140
00:00
00:09
00:17
00:26
00:34
00:43
00:52
01:02
01:11
01:20
01:29
01:39
01:49
01:58
02:08
02:17
02:26
02:35
02:44
02:53
03:02
03:12
03:21
03:30
03:40
03:50
03:59
04:08
04:18
04:27
04:37
04:47
04:56
BIS
Male 19 years, Colectomy/coelioscopy+laparotomy
Courtesy Dr Ngai Liu – Hoch Hospital, Paris France
0
20
40
60
80
100
120
140
00:00
00:08
00:16
00:24
00:32
00:40
00:48
00:56
01:05
01:14
01:22
01:31
01:40
01:49
01:57
02:06
02:15
02:24
02:31
02:40
02:48
02:57
03:05
03:14
03:23
03:31
03:40
03:49
03:57
04:06
04:15
04:24
04:33
04:42
04:50
04:59
BIS
Intubation
Incision
Laparotomy
Male 19 years, Colectomy/coelioscopy+laparotomy
Courtesy Dr Ngai Liu – Hoch Hospital, Paris France
0
20
40
60
80
100
120
140
00:00
00:08
00:16
00:24
00:32
00:40
00:48
00:56
01:05
01:14
01:22
01:31
01:40
01:49
01:57
02:06
02:15
02:24
02:31
02:40
02:48
02:57
03:05
03:14
03:23
03:31
03:40
03:49
03:57
04:06
04:15
04:24
04:33
04:42
04:50
04:59
0
2
4
6
8
10
12
14
16
BIS
Propofol
Intubation
Incision
Laparotomy
Male 19 years, Colectomy/coelioscopy+laparotomy
Courtesy Dr Ngai Liu – Hoch Hospital, Paris France
00:00
00:09
00:17
00:25
00:34
00:43
00:52
01:01
01:10
01:19
01:29
01:38
01:48
01:57
02:06
02:16
02:25
02:33
02:43
02:52
03:01
03:10
03:20
03:29
03:38
03:48
03:57
04:06
04:15
04:25
04:35
04:44
04:54
05:03
BIS
Propofol
Remifentanil
Intubation
Incision
Laparotomy
Male 19 years, Colectomy/coelioscopy+laparotomy
(µg.ml
-1
)
(ng.ml
-1
)
Courtesy Dr Ngai Liu – Hoch Hospital, Paris France
Closed-loop systems (CLS), when compared to human management:
1) More Stable Anesthesia / More frequent adjustements
2) Less Overshoot / Undershoot
2017
META-ANALYSIS
Fluid Management
Based on Urine output
Closed-Loop Fluid Administration Compared to
Anesthesiologist Management for Hemodynamic Optimization
and Resuscitation During Surgery: An In Vivo Study
Joseph Rinehart, Christine Lee, Cecila Canales, Allen Kong, Zeev Kain, Maxime Cannesson
Anesthesia Analgesia 2013
Closed-Loop Fluid Administration Compared to
Anesthesiologist Management for Hemodynamic Optimization
and Resuscitation During Surgery: An In Vivo Study
Joseph Rinehart, Christine Lee, Cecila Canales, Allen Kong, Zeev Kain, Maxime Cannesson
Anesthesia Analgesia 2013
Closed-Loop Fluid Administration Compared to
Anesthesiologist Management for Hemodynamic Optimization
and Resuscitation During Surgery: An In Vivo Study
Joseph Rinehart, Christine Lee, Cecila Canales, Allen Kong, Zeev Kain, Maxime Cannesson
Anesthesia Analgesia 2013
Assisted Fluid
Management (AFM)
Software
Vasopressors
Anesthesiology 2019
30
40
50
60
70
80
90
100
110
1
11
21
31
41
51
61
71
81
91
101
111
121
131
141
151
161
171
181
191
201
211
221
231
241
251
261
271
281
291
301
311
321
331
341
351
361
371
381
391
401
411
421
431
441
451
461
Closed-Loop vs. Unmanaged Vasodilation
Sham1 - P1 Sham2 - P7 Sham3 - P9 Sham4 - P11 Sham5 - P12
Sham6 - P13 Sham7 - P14
NTP low
dose
NTP high dose
Anesthesiology 2019
Anesthesiology 2019
NTP low
dose
NTP high dose
Closed Loop Vasopressor Use in ICU
Erasme Hospital, Brussels, Belgium
Integration
Outline
• Introduction
• Feedback Control / Closed Loop / Automation
• Examples of Closed Loop Systems in Anesthesia
• The Challenges Ahead
To Err is human…
The Dangers of AI in Medicine
Technologica
l Risks
Algorithmic bias and unrepresentative and / or
inacurate data
Cybersecurity
Ethical
Risks
Equity and Inclusion
Data Privacy and Security
Right to access
Legal Risks Liability and accountability
Professional
Risks
Automation and Unemployement
Ethics of labor unions
Medical Education and AI
Patients physicians relationship and Humanistic
medicine
Human /
Machine
Interface
Building trust in AI / Black Boxes
Risk Homeostasis
Automation bias and complacency
Error underreporting
Interpretable and explainable AI