Post on 09-Jan-2016
description
transcript
Mestrado Integrado em MedicinaIntrodução à Medicina II
18-05-2010
Professor
Doutor Altamiro Pereira
ARTIFICIAL INTELLIGENCE FOR
CRITICAL CARE MONITORING AND
DECISION SUPPORT
Final Presentation
IMPACT ON PATIENT OUTCOMES
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Introduction
Previously…
[1] Hanson CW 3rd, Marshall BE; Artificial intelligence applications in the intensive care unit; Critical caremedicine; 2001 Feb; 29 (2); 427-35
Artificial intelligence applications in the
intensive care unit [1]
Systematic review
from 2001
We’ve concluded that this study was more a state of the art than an Systematic
Review and knowing that we’ve decided to search articles with no date restriction
We found important to review once more the progresses in this
subject as it’s rapidly evolving.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Intensive Care Unit: Definition
An intensive care unit (ICU) is a specialized department in hospitals that provides life support
or organ support systems in patients who are critically ill and who usually require constant
monitoring.
Critical care is the permanent and thorough care provided to the critical patients in intensive care
units (ICUs).
Introduction
Critical Care: Definition
AI: Definition
Artificial intelligence can be defined as a field of science and engineering concerned with the
computational understanding of what is commonly called intelligent behaviour, and with the
creation of artefacts that exhibit such behaviour [2].
[2] Ramesh, A.N.; Kambhamti, C.; Monson, J.R.T.; Drew, P.J.; Artificial intelligence in medicine; Ann R Coll Surg Engl, 2004; 86: 334–338.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Introduction
AI and ICU
• Intensive care medicine frequently involves making rapid decisions on the basis of a
large and disparate array of information [3].
• Since the technology of monitoring astronauts’ vital signs in space was transferred to
the bedside in the 1960s, patient monitoring systems have become an indispensable
part of critical care [4].
[3] Jason H. T. Bates and Michael P. Young; Applying Fuzzy Logic to Medical Decision Making in theIntensive Care Unit; 2003 Apr[4] Ying Zhang, MEng Real-Time Development of Patient-Specific Alarm Algorithms; Proceedings of the 29th Annual International; Conference of the IEEE EMBS; Cité Internationale, Lyon, FranceAugust 23-26, 2007
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Introduction
Intensive care unit (ICU) is a hospital unit staffed and equipped to
provide intensive monitoring closely for a critical medical condition 5.
The bedside data must be extracted and organized to become useful
information for clinical decisions .
In ICU the amount of data makes its integration and interpretation time-consuming and inefficient [5].
Artificial intelligence (AI) systems are pointed as a direction to follow for the development of patient care.
[5]-HELDT, Thomas; LONG, Bill; VERGHESE, George C. ; SZOLOVITS, Peter; MARK, Roger G.; Integrating Data, Models, and Reasoning in Critical Care; Conf Proc IEEE Eng Med Biol Soc. 2006;1:350-3
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Introduction
AI in medicine is designed to support healthcare workers in
their every day duties, assisting with tasks that rely on the
manipulation of data and knowledge [6].
Development of artificial intelligence (AI) in medicine
Development of AI programs intended to:
- help the clinician in the formulation of a diagnosis;
- help the clinician in making of therapeutic decisions;
- helping prediction outcome.
[6] RAMNARAYAN, Padmanabhan; KAPOOR, Ritika R.; COREN, Michael; NANDURI, Vasantha; TOMLINSON, Amanda L.; TAYLOR, Paul M.; WYATT, Jeremy C.; RITTO, Joseph F.; Measuring the Impact of Diagnostic Decision Support on the Quality of Clinical Decision Making: Development of a Reliable and Valid Composite Score; Journal of the American Medical Informatics Association, 2003; Volume 10 Number 6.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Introduction
Modern intensive care units (ICUs) employ an impressive array of
technologically sophisticated instrumentation to provide detailed
measurements of the pathophysiological state of each patient [5], [7].
Such systems include:
- Artificial neural networks (ANNs),
-Fuzzy expert systems,
-Evolutionary computation and hybrid
intelligent systems.
Such systems can:
- Reduce the problem of information
overload,
- Providing alarms more specific than
today’s single-variable limit alarms,
- Decrease healthcare cost in ICU,
(…)[5]-HELDT, Thomas; LONG, Bill; VERGHESE, George C. ; SZOLOVITS, Peter; MARK, Roger G.; Integrating Data, Models, and Reasoning in Critical Care; Conf Proc IEEE Eng Med Biol Soc. 2006;1:350-3[7]-MAHFOUF, M.; ABBOD, M.F.; LINKENS, D.A., A survey of fuzzy logic monitoring and control utilisation in medicine; Artificial Intelligence in Medicine, 2001; 21: 27-42.
18-05-2010
Turma 66fmup0910@gmail.com
Does the usage of monitoring or clinical decision support systems
that include AI technology improve the quality of patient care?
Professor
Doutor Altamiro Pereira
Research question
18-05-2010
Turma 66fmup0910@gmail.com
To review the usefulness of AI monitoring and Clinical Decision
Support (CDS) systems when applied to patients in the ICU by the
interpretation of the patient outcomes.
Professor
Doutor Altamiro Pereira
Aim
18-05-2010
Turma 66fmup0910@gmail.com
• To study the benefits and drawbacks of artificial intelligence
monitoring and CDS systems for critical care when compared to non AI-
methods.
• To find out the impact of the usage of AI systems in the different
patient outcomes in ICU's.
Professor
Doutor Altamiro Pereira
Specific Objectives
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
Study design:
Systematic review
1 - An exhaustive search, in electronic databases, and inclusion of primary
studies.
2 - Quality assessment of included studies and data extraction (review, by
two persons, of the title and the abstract or the article. Same process for the
full article. A third opinion may be requested).
3 - Synthesis of study results (SPSS and Review Manager).
4 - Interpretation of results and report writing.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
Population:
Articles which report AI applications for monitoring (including warning (alert)),
decision support or prescription support in the
intensive care unit.
Articles which report AI applications for monitoring (including
warning (alert)), decision support or prescription support in the
intensive care unit.
18-05-2010
Turma 66fmup0910@gmail.com
Data collection methods:
→ Search strategy
ProfessorDoutor Altamiro Pereira
Methods
Articles included by reviewerArticles were searched in:
PubMed;
ISI Web of Knowledge;
SCOPUS.
with no date restriction
18-05-2010
Turma 66fmup0910@gmail.com
Data collection methods - Query:
ProfessorDoutor Altamiro Pereira
Methods
Articles included by reviewer
18-05-2010
• Our query was based in the following keywords with the represented basic
relationship between them:
“Artificial Intelligence” AND (“Critical Care” or “Intensive Care Unit”) AND
“Trial”.
• Using this structure, we used synonyms and acronyms based on MeSH terms
related to the keywords mentioned above, keeping the described relationship, in
order to increase our pool of articles.
Turma 66fmup0910@gmail.com
Data collection methods - Query:
ProfessorDoutor Altamiro Pereira
Methods
Articles included by reviewer
18-05-2010
• We were very benevolent with the terms included, in order to reduce the chance
of excluding important articles through an insufficient search.
• Due to restrictions in size and structure imposed by the different search engines,
the query used in each engine had to be molded accordingly.
• Since PubMed was the one with the less restrictions and with the biggest article
pool, we decided to design our query through there and then make the necessary
adjusments to the other search engines.
Turma 66fmup0910@gmail.com
Query terms:
Artificial intelligence:
Computer reasoning, machine intelligence, machine learning, computer
vision system, knowledge acquisition, fuzzy logic, expert systems, knowledge bases, neural
networks (computer), neural network model, perception, direct support system, robotic,
telerobotic;
Intensive care unit:
Critical care (unit), surgical intensive care (unit), neonatal intensive care
(unit), infant newborn intensive care (unit), pediatric intensive care (unit), ICU, PICU, NICU, CC,
burn(s) unit, respiratory care unit, coronary care unit.
We also used the term “trial” in order to narrow down and specify our research.Professor
Doutor Altamiro PereiraTurma 6
6fmup0910@gmail.com
Methods
18-05-2010
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Query used in the PubMed Database
("trial"[all fields]) AND ((("critical care"[MeSH Terms] OR ("critical"[All Fields] AND "care"[All Fields]) OR "critical care"[All Fields]) OR (critical[All Fields] AND
cares[All Fields]) OR ("intensive care"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All Fields]) OR "intensive care"[All Fields]) OR (intensive[All Fields]
AND cares[All Fields]) OR ("intensive care"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All Fields]) OR "intensive care"[All Fields] OR ("surgical"[All
Fields] AND "intensive"[All Fields] AND "care"[All Fields]) OR "surgical intensive care"[All Fields]) OR (("surgical procedures, operative"[MeSH Terms] OR
("surgical"[All Fields] AND "procedures"[All Fields] AND "operative"[All Fields]) OR "operative surgical procedures"[All Fields] OR "surgical"[All Fields]) AND
Intensive[All Fields] AND Cares[All Fields]) OR ("intensive care, neonatal"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All Fields] AND "neonatal"[All
Fields]) OR "neonatal intensive care"[All Fields] OR ("neonatal"[All Fields] AND "intensive"[All Fields] AND "care"[All Fields])) OR (("infant, newborn"[MeSH
Terms] OR ("infant"[All Fields] AND "newborn"[All Fields]) OR "newborn infant"[All Fields] OR "neonatal"[All Fields]) AND Intensive[All Fields] AND Cares[All
Fields]) OR (("infant, newborn"[MeSH Terms] OR ("infant"[All Fields] AND "newborn"[All Fields]) OR "newborn infant"[All Fields] OR ("infant"[All Fields] AND
"newborn"[All Fields]) OR "infant newborn"[All Fields]) AND ("intensive care"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All Fields]) OR "intensive
care"[All Fields])) OR (("infant, newborn"[MeSH Terms] OR ("infant"[All Fields] AND "newborn"[All Fields]) OR "newborn infant"[All Fields] OR ("infant"[All Fields]
AND "newborn"[All Fields]) OR "infant newborn"[All Fields]) AND Intensive[All Fields] AND Cares[All Fields])) OR
(("intensive care"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All Fields]) OR "intensive care"[All Fields]) OR ICU[All Fields] OR icus[All Fields] OR
picu[All Fields] OR picus[All Fields] OR nicu[All Fields] OR nicus[All Fields] OR ("intensive care units"[MeSH Terms] OR ("intensive"[All Fields] AND "care"[All
Fields] AND "units"[All Fields]) OR "intensive care units"[All Fields] OR ("critical"[All Fields] AND "care"[All Fields] AND "unit"[All Fields]) OR "critical care unit"[All
Fields]) OR ("critical care"[MeSH Terms] OR ("critical"[All Fields] AND "care"[All Fields]) OR "critical care"[All Fields]) OR ("intensive care"[MeSH Terms] OR
("intensive"[All Fields] AND "care"[All Fields]) OR "intensive care"[All Fields]) OR ("burn units"[MeSH Terms] OR
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Query used in the PubMed Database
("burn"[All Fields] AND "units"[All Fields]) OR "burn units"[All Fields]) OR ("burn units"[MeSH Terms] OR ("burn"[All Fields] AND "units"[All Fields]) OR "burn
units"[All Fields] OR ("burns"[All Fields] AND "unit"[All Fields]) OR "burns unit"[All Fields]) OR ("respiratory care units"[MeSH Terms] OR ("respiratory"[All Fields]
AND "care"[All Fields] AND "units"[All Fields]) OR "respiratory care units"[All Fields]) OR ("respiratory care units"[MeSH Terms] OR ("respiratory"[All Fields] AND
"care"[All Fields] AND "units"[All Fields]) OR "respiratory care units"[All Fields] OR ("respiratory"[All Fields] AND "care"[All Fields] AND "unit"[All Fields]) OR
"respiratory care unit"[All Fields]) OR ("coronary care units"[MeSH Terms] OR ("coronary"[All Fields] AND "care"[All Fields] AND "units"[All Fields]) OR "coronary
care units"[All Fields]) OR ("coronary care units"[MeSH Terms] OR ("coronary"[All Fields] AND "care"[All Fields] AND "units"[All Fields]) OR "coronary care
units"[All Fields] OR ("coronary"[All Fields] AND "care"[All Fields] AND "unit"[All Fields]) OR "coronary care unit"[All Fields]))) AND (("artificial intelligence"[MeSH
Terms] OR ("artificial"[All Fields] AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields]) OR ("artificial intelligence"[MeSH Terms] OR ("artificial"[All
Fields] AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields] OR ("computer"[All Fields] AND "reasoning"[All Fields]) OR "computer reasoning"[All
Fields]) OR ("artificial intelligence"[MeSH Terms] OR ("artificial"[All Fields] AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields] OR ("machine"[All
Fields] AND "intelligence"[All Fields]) OR "machine intelligence"[All Fields]) OR AIs[All Fields] OR ("artificial intelligence"[MeSH Terms] OR ("artificial"[All Fields]
AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields] OR ("machine"[All Fields] AND "learning"[All Fields]) OR "machine learning"[All Fields]) OR
(("knowledge"[MeSH Terms] OR "knowledge"[All Fields]) AND Representation[All Fields]) OR (("knowledge"[MeSH Terms] OR "knowledge"[All Fields]) AND
("Representations (Berkeley)"[Journal] OR "representations"[All Fields])) OR ("artificial intelligence"[MeSH Terms] OR ("artificial"[All Fields] AND "intelligence"[All
Fields]) OR "artificial intelligence"[All Fields] OR ("computer"[All Fields] AND "vision"[All Fields] AND "systems"[All Fields]) OR "computer vision systems"[All
Fields]) OR ("artificial intelligence"[MeSH Terms] OR ("artificial"[All Fields] AND "intelligence"[All Fields]) OR "artificial intelligence"[All Fields] OR ("computer"[All
Fields] AND "vision"[All Fields] AND "system"[All Fields]) OR
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Query used in the PubMed Database
"computer vision system"[All Fields]) OR (("knowledge"[MeSH Terms] OR "knowledge"[All Fields]) AND Acquisitions[All Fields]) OR ("fuzzy logic"[MeSH Terms]
OR ("fuzzy"[All Fields] AND "logic"[All Fields]) OR "fuzzy logic"[All Fields]) OR ("expert systems"[MeSH Terms] OR ("expert"[All Fields] AND "systems"[All
Fields]) OR "expert systems"[All Fields] OR ("expert"[All Fields] AND "system"[All Fields]) OR "expert system"[All Fields]) OR ("expert systems"[MeSH Terms]
OR ("expert"[All Fields] AND "systems"[All Fields]) OR "expert systems"[All Fields]) OR ("knowledge bases"[MeSH Terms] OR ("knowledge"[All Fields] AND
"bases"[All Fields]) OR "knowledge bases"[All Fields] OR ("knowledge"[All Fields] AND "base"[All Fields]) OR "knowledge base"[All Fields]) OR ("knowledge
bases"[MeSH Terms] OR ("knowledge"[All Fields] AND "bases"[All Fields]) OR "knowledge bases"[All Fields]) OR ("knowledge bases"[MeSH Terms] OR
("knowledge"[All Fields] AND "bases"[All Fields]) OR "knowledge bases"[All Fields] OR "knowledgebases"[All Fields]) OR ("knowledge bases"[MeSH Terms] OR
("knowledge"[All Fields] AND "bases"[All Fields]) OR "knowledge bases"[All Fields] OR "knowledgebase"[All Fields]) OR ("neural networks (computer)"[MeSH
Terms] OR ("neural"[All Fields] AND "networks"[All Fields] AND "(computer)"[All Fields]) OR "neural networks (computer)"[All Fields] OR ("neural"[All Fields]
AND "network"[All Fields]) OR "neural network"[All Fields]) OR ("Neural Netw"[Journal] OR "IEEE Trans Neural Netw"[Journal] OR ("neural"[All Fields] AND
"networks"[All Fields]) OR "neural networks"[All Fields]) OR ("neural networks (computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields]
AND "(computer)"[All Fields]) OR "neural networks (computer)"[All Fields] OR ("neural"[All Fields] AND "network"[All Fields] AND "model"[All Fields]) OR "neural
network model"[All Fields]) OR ("neural networks (computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields] AND "(computer)"[All Fields]) OR
"neural networks (computer)"[All Fields] OR ("neural"[All Fields] AND "network"[All Fields] AND "models"[All Fields]) OR "neural network models"[All Fields]) OR
("neural networks (computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields] AND "(computer)"[All Fields]) OR "neural networks
(computer)"[All Fields] OR "perceptron"[All Fields]) OR ("neural networks (computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields]
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Query used in the PubMed Database
AND "(computer)"[All Fields]) OR "neural networks (computer)"[All Fields] OR "perceptrons"[All Fields]) OR ("neural networks
(computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields] AND "(computer)"[All Fields]) OR "neural
networks (computer)"[All Fields] OR ("connectionist"[All Fields] AND "model"[All Fields]) OR "connectionist model"[All Fields])
OR ("neural networks (computer)"[MeSH Terms] OR ("neural"[All Fields] AND "networks"[All Fields] AND "(computer)"[All
Fields]) OR "neural networks (computer)"[All Fields] OR ("connectionist"[All Fields] AND "models"[All Fields]) OR
"connectionist models"[All Fields]) OR ("robotics"[MeSH Terms] OR "robotics"[All Fields]) OR ("robotics"[MeSH Terms] OR
"robotics"[All Fields] OR "telerobotic"[All Fields]) OR (Remote[All Fields] AND ("surgery"[Subheading] OR "surgery"[All Fields]
OR "operations"[All Fields] OR "surgical procedures, operative"[MeSH Terms] OR ("surgical"[All Fields] AND "procedures"[All
Fields] AND "operative"[All Fields]) OR "operative surgical procedures"[All Fields])) OR (Remote[All Fields] AND ("surgical
procedures, operative"[MeSH Terms] OR ("surgical"[All Fields] AND "procedures"[All Fields] AND "operative"[All Fields]) OR
"operative surgical procedures"[All Fields] OR "operation"[All Fields])) OR Telerobotic[All Fields] OR robotic[All Fields] OR
(direct[All Fields] AND support[All Fields] AND system[All Fields]) OR (direct[All Fields] AND support[All Fields] AND
systems[All Fields]))
Turma 66fmup0910@gmail.com
Query used in the ISI Web of Knowledge Database
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Ts=(artificial intelligence OR Computer Reasoning OR Machine Intelligence OR AI OR AIs OR Machine Learning
OR Knowledge Representation OR Computer Vision System OR Knowledge Acquisition OR fuzzy logic OR
expert system OR Knowledge Base OR Knowledgebase OR Neural Network OR Connectionist Model OR robotic
OR Telerobotic OR Remote Operation OR direct support system) AND Ts=(intensive care OR ICU OR icus OR
picu OR picus OR nicu OR nicus OR critical care unit OR critical care OR intensive care OR burn units OR burns
unit OR respiratory care units OR respiratory care unit OR coronary care units OR coronary care unit OR Critical
care OR critical cares OR intensive care OR intensive cares OR Surgical Intensive Care OR Surgical Intensive
Cares OR Neonatal Intensive Care OR Neonatal Intensive Cares OR Infant Newborn Intensive Care OR Infant
Newborn Intensive Cares) AND TS=(trial)
Turma 66fmup0910@gmail.com
Query used in the SCOPUS Database
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
TITLE-ABS-KEY(("intensive care" OR icu OR icus OR picu OR picus OR nicu OR nicus OR "critical care unit" OR "critical
care" OR "intensive care" OR "burn units" OR "burns unit" OR "respiratory care units" OR "respiratory care unit" OR
"coronary care units" OR "coronary care unit" OR "Critical care" OR "critical cares" OR "intensive care" OR "intensive
cares" OR "Surgical Intensive Care" OR "Surgical Intensive Cares" OR "Neonatal Intensive Care" OR "Neonatal Intensive
Cares" OR "Infant Newborn Intensive Care" OR "Infant Newborn Intensive Cares") AND ("artificial intelligence" OR
"computer reasoning" OR "machine intelligence" OR ai OR ais OR "machine learning" OR "knowledge representation" OR
"knowledge representations" OR "computer vision systems" OR "computer vision system" OR "knowledge acquisition" OR
"knowledge acquisitions" OR "fuzzy logic" OR "expert system" OR "expert systems" OR "knowledge base" OR "knowledge
bases" OR "knowledge base" OR "neural network" OR "neural networks" OR "neural network model" OR "neural network
models" OR perceptron OR perceptrons OR "connectionist model" OR "connectionist models" OR "robotics" OR
"telerobotics" OR "remote operations" OR "remote operation" OR "direct support system" OR "direct support systems" OR
"robotic" OR "telerobotic") AND "trial") AND SUBJAREA(mult OR agri OR bioc OR immu OR neur OR phar OR mult OR
medi OR nurs OR dent OR heal OR mult OR ceng OR CHEM OR comp OR eart OR ener OR engi OR envi OR mate OR
math OR phys)
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Methods
Inclusion/exclusion criteria:Inclusion criteria
1.Study design (clinical trials);
2.Study participants of included articles are patients in the intensive care unit;
3.Studies that describe AI systems’ intervention on monitoring, warning (alert), decision
support or prescription support;
4. Study outcomes include mortality, morbidity, quality of life, length of stay
or other patient outcomes.
Exclusion criteria
Articles that use data from the ICU as secondary data for the demonstration of AI systems
based only on system's performance outcomes.Articles not able to be filtered due to absence of abstract or full text were excluded. Those that didn’t meet the inclusion criteria were also excluded. Finally, those that met the inclusion and the exclusion criteria were excluded.
18-05-2010
Turma 66fmup0910@gmail.com
Study variables:
• Characteristics of the articles (year, author and country of publishing, etc…)
• Type of study (number of participants, duration, etc…)
• Domain of application (neurological, respiratory, cardiovascular, etc…)
• Area of application (monitoring, clinical decision support.)
• The patients‘ outcomes described in each article (mortality rate, length of stay,
quality of stay, morbidity, quality of life, cost of stay or other patient outcomes).
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Turma 66fmup0910@gmail.com
Statistical analysis:
• Storage of the data using SPSS;
• Analysis of the study variables using the appropriate frequency measures;
• Agreement statistic and Kappa statistic for the agreement analysis
Professor
Doutor Altamiro Pereira
Methods
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Search Results
Search was conducted on March 7th, 2010
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Reviewers' Agreement
In order to evaluate if the analysis
of the articles during the first
review was well executed, we’ve
analysed the agreement between
reviewers.
. Global agreement was 86%;
. Negative agreement was 94%;
. Positive agreement was 56%.
The fact that we had a high value in the negative agreement means that
we’ve correctly excluded the articles in the first review.
Abstract's Review
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Reviewers' Agreement
In order to evaluate if the analysis
of the articles during the first
review was well executed, we’ve
analysed the agreement between
reviewers.
The agreement levels remain relatively high.
Still, the relatively low level of the Kappa statistic
may be due to the fact that we had a low
number of articles in the second stage of the
review.
Full Text review
. Global agreement was 73%;
. Negative agreement was 46%;
. Positive agreement was 82%.
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Publication
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Study Type
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Outcomes
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Outcomes
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Outcomes
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
• It’s a decision/prescription support system articles. In this study, the objective was to
test if infusion rates of norepinephrine controled by a closed-loop fuzzy logic system would
reduce the duration of septic shock
• Duration of shock proved to be significantly shorter (P < 0.001) in the fuzzy group than in
the control group. The total amount of norepinephrine infused was significantly lower (P <
0.01) in the fuzzy group than in the control group. This resulted in a lower total dose being
administered and is related with a shorter duration of septic shock, although no statistical
corroboration was provided in the article.
Merouani2008 [8]
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
• Moreover, it was shown that the fuzzy logic group showed advantages
regarding the mean arterial blood presure (MAP). The patient's MAP slowly
oscillates around the target value set by the intensivist in the fuzzy group, while in
the control patients it tended to drift, with more marked amplitudes.
Merouani2008
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Dazzy2001 [9]
● Fuzzy logic is the base of a monitoring system. Fuzzy logic principles and neural network
techniques were applied to control intravenous insulin administration rates during glucose
infusion in critical diabetic patient.
● The results showed that the neuro-fuzzy nomogram allowed a faster decrease of blood
glucose (BG) levels below 10 mmol/l (A:7.8 vs. B: 13.2 h; P < .02).
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Dazzy2001
● It also allowed to reach and maintain a strict control condition around a specific point (6.7
mM/l) as safely as the old one did around a larger glycemic target (between 6.7 and 10 mM/l);
P < .0001.
● Also, it was concluded that this nomogram can be efficiently and safely used to reach
faster and to maintain a near normal BG level in critically ill diabetic patients during
intravenous glucose and insulin infusion.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Ying1992 [10]
• It’s a fuzzy logic-monitoring study. It uses a fuzzy control system to provide closed-loop
control of mean arterial pressure (MAP) in post-surgical patients in a cardiac surgical
intensive care unit setting by regulating sodium nitroprusside (SNP) infusion.
• To evaluate the ability of the fuzzy control system handling the patients with different
sensitivity, computer simulation was conducted.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Ying1992
• It showed that the clinically fine-tuned fuzzy control SNP system could adapt to a wide
range of patient sensitivity, from the sensitive patients (K = -2.88) to the insensitive patients
(K = -0.18), a ratio of 16: 1.
• MAP is tightly controlled around the desired MAP level. The results of the clinical trials
on 12 patients revealed that the performance of the fuzzy control SNP delivery system was
clinically acceptable.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Huang2006 [11]
• Uses fuzzy logic control to provide continuous propofol sedation in order to reduce the
effect of agitation on intracranial pressure (ICP) in neurosurgical intensive care unit patients.
Fuzzy logic is used on monitoring patients.
• The results show that for mean and RMSD (root mean square deviation) of ICP errors,
the values of SOFLC group were significantly lower than those of RBC (rule-based
controller) and FLC (fuzzy logic controller) groups (P < 0.05), but the values of the RBC and
FLC groups showed no significant difference.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Huang2006
• It is also concluded that FLC can easily mimic the rule-base of human experts to achieve
stable sedation similar to the RBC group.
• Furthermore, the results also show that a SOFLC (self-organizing fuzzy logic controller)
can provide more stable sedation of ICP pattern because it can modify the fuzzy rule-base to
compensate for inter-patient variations.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Denai2009 [12]
• This study used a fuzzy logic control system on a clinical decision support system (CDSS)
in the cardiac intensive care unit (CICU).
• The study used a model that was able to reproduce conditions experienced by 7 post
operative cardiac surgery patients (hypertension, hypovolemia, vasodilation, systemic
inflammatory response system (SIRS)) and Simulated patient scenarios were developed in
collaboration with the expert/study anesthetist (3rd author) to reproduce a range of
pathophysiological conditions resembling those observed in post-cardiac surgery patients.
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Results
18-05-2010
Denai2009
• All in all, the interventions in this study resulted in a good control of the. Therefore, the
preliminary simulation studies showed good feasibility for the application of CDSS for controlling
the patients’ cardiovascular system following surgery in a real ICU.
• The results demonstrated steady-state performances of the fuzzy rule-based controller for
the range of hypertension considered. It was also demonstrated that the multi-drug fuzzy
controller produced good responses and target tracking with acceptable overshoots in the MAP
(Mean Arterial Pressure). Aditionnaly, as shown in the figure, the other hemodynamic
parameters reached reasonable final values.
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Discussion
● The review of the 5 articles found in our research showed us that the results
of the clinical trials we’ve found about this specific subject were considered to
be positive by the authors.
● Our results show us that this is an area of medical interest that can be
successfully used in the care of critical care patients from several areas of the
intensive care units. The articles found reported clinical trials in which the use of
the fuzzy logic systems reached better results than the non Artificial Intelligence
systems and in many cases with statistically significant results.
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Discussion
● The results of Denai et al and Merouani et al showed that the use of clinical decision
support systems in the ICU can be positive, and regarding that it should be seen as a way
of fighting the growing difficulty found by clinicians and other medical staff to make fast
decisions base on a great amount of information.
● The articles written by Ying et al, Huang et al and Dazzy et al, although with different
domains of application, show a very positive response in the use of AI-systems in
monitoring. These systems can represent a large benefit in the ICUs either in the decrease
of errors [15], the better adaptation to each particular patient or even in the maintaining of a
strict control of outcomes such as glucose levels [17].
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Discussion
● About the articles included we can also say that all of them used fuzzy logic and
that the majority of them were used in monitoring.
● The domain of applications of the studies found was also something that
concerned us from the beginning and in the articles included the domains where
artificial intelligence systems were used were cardiac intensive care [16][19],
insulin administration[17], neurosurgical intensive care[15] and also in the control
of septic shock[18].
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
Discussion
● In the areas and domains of application in which AI systems, in these cases
fuzzy-logic, were studied, evidence showed that it can be used efficiently and safely.
● Although the results obtained by our research were not sufficient to state that the
use of Artificial Intelligence systems in the intensive care units is benefitial in every
single aspect, it seems to us that there is a need to go further on this subject as it
might represent a good breakthrough to enhance the quality and efficiency of the
treatment provided to Intensive Care Units’ patients.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Limitations
• Inability to access the full text of two articles due to the lack of response from their authors.
18-05-2010
Turma 66fmup0910@gmail.com
Professor
Doutor Altamiro Pereira
Authors
6th Class
Alves, J.; Conceição, M.; Dama, C.; Estevão-Costa, N.; Jesus, M.;
Lopes, S.; Neto, R.; Pereira, M.; Queirós, P.; Silva, V.; Videira, P.
Adviser: Pedro Pereira Rodrigues
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira
References
[1] Hanson CW 3rd, Marshall BE; Artificial intelligence applications in the intensive care unit; Critical caremedicine; 2001 Feb; 29 (2); 427-35
[2] Ramesh, A.N.; Kambhamti, C.; Monson, J.R.T.; Drew, P.J.; Artificial intelligence in medicine; Ann R Coll Surg Engl, 2004; 86: 334–338.
[3] Jason H. T. Bates and Michael P. Young; Applying Fuzzy Logic to Medical Decision Making in theIntensive Care Unit; 2003 Apr
[4] Ying Zhang, MEng Real-Time Development of Patient-Specific Alarm Algorithms; Proceedings of the 29th Annual International; Conference of the IEEE EMBS; Cité Internationale, Lyon, FranceAugust 23-26, 2007
[5]-HELDT, Thomas; LONG, Bill; VERGHESE, George C. ; SZOLOVITS, Peter; MARK, Roger G.; Integrating Data, Models, and Reasoning in Critical Care; Conf Proc IEEE Eng Med Biol Soc. 2006;1:350-3
[6] RAMNARAYAN, Padmanabhan; KAPOOR, Ritika R.; COREN, Michael; NANDURI, Vasantha; TOMLINSON, Amanda L.; TAYLOR, Paul M.; WYATT, Jeremy C.; RITTO, Joseph F.; Measuring the Impact of Diagnostic Decision Support on the Quality of Clinical Decision Making: Development of a Reliable and Valid Composite Score; Journal of the American Medical Informatics Association, 2003; Volume 10 Number 6.
18-05-2010
Turma 66fmup0910@gmail.com
ProfessorDoutor Altamiro Pereira Turma 6
6fmup0910@gmail.com
18-05-2010
References
[7] MAHFOUF, M.; ABBOD, M.F.; LINKENS, D.A., A survey of fuzzy logic monitoring and control utilisation in medicine; Artificial Intelligence in Medicine, 2001; 21: 27-42.
[8] Mehdi Merouani, Bruno Guignard, François Vincent, Stephen W Borron, Philippe Karoubi,Jean-Philippe Fosse, Yves Cohen, Christophe Clec'h, Eric Vicaut, Carole Marbeuf-Gueye, Frederic Lapostolle and Frederic Adnet; Norepinephrine weaning in septic shock patients by closed loop controled based fuzzy logic; 2008 Dec;
[9] Davide Dazzi, Francesco Taddei, Alessandra Gavarini, Enzo Uggerib, Roberto Negro, Antonio Pezzarossa; The control of blood glucose in the critical diabetic patient. A neuro-fuzzy method; 2000; 80-87
[10] Hao Ying, Member, IEEE, Michael McEachern, Donald W. Eddleman, Member IEEE, Louis C. Sheppard, Fellow, IEEE; Fuzzy Control of Mean Arterial Pressure in Postsurgical Patients with Sodium Nitroprusside Infusion; 1992 Oct;
[11] Sheng-Jean Huang, Jiann-Shing Shieh, Mu Fu, Ming-Chien Kao; Fuzzy logic control for intracranial pressure via continuous propofol sedation in a neurosurgical intensive care unit; 2006; 639-647;
[12] Mouloud A. Denaı, Mahdi Mahfouf, Jonathan J. Ross; A hybrid hierarchical decision support system for cardiac surgical intensive care patients. Part I: Physiological modelling and decision supportsystem design; 2009; 35-52;