Fuzzy Expert System for Fluid Management in General Anaesthesia

**Ph.D. (Engineering and Technology) – Nagpur University, India; Master of Engineering (Civil with specialisation in Environmental Engineering), University of Roorkee, India.; Honor Gust professor in Department of Bioinformatics, University of Pune. E-mail: ashokwd@yahoo.com ***Ph.D. in Biotechnology, Institute of Biotechnology and Bioinformatics, University of Pune, India E-mail: ashosaini@yahoo.co.in

Abstract

Background: Fuzzy set and fuzzy logic founded by Prof. Lotfi Zadeh (1965) make it possible to define inexact medical entities as fuzzy sets and models the subjective information. Fuzzy logic is reasoning with fuzzy sets. In medicine, the contradictory natures are common facts. Anaesthetists use rules of thumb when managing patients. He adjusts the drug and fluids inflow, or possibly ventilation, to monitor physiological state of the patient. Real-world knowledge is characterised by incompleteness, inaccuracy and inconsistency. It is not possible to define precisely the terms such as high temperature, low mean arterial pressure (MAP), very high intravenous fluid rate (IFR), and alike. The field of surgery and anaesthesia is very wide as many factors contribute to it, such as diagnosis, image processing, and path physiological reasoning and anaesthesia control. Fuzzy logic seems suited to use in anaesthesia because of the way it so naturally represents the subjective human notions employed in much of medical decision making.
Patient and methods: We have selected 71 patient ASA I–II classes, aged between 15 and 50 years and weight between 40 and 85 kg. In this sequel, we have made an honest attempt to incorporate fuzzy techniques and developed a fuzzy expert system for fluid management in general anaesthesia. MAP and hourly urine output (HUO) are the fuzzy input to the fuzzy expert system as the antecedent parts of the rule and the output is the defuzzified value of IFR at the desired level.
Results: We have predicted nine different fuzzy rules by using Min–Max approach, and eventually we find out the action that must be taken by using centroid approach. Then out of nine fuzzy rules four rules will be fired for patients. Based on COA, the computed value of IFR for the above set parameters, which for one sample of patient data was 118 ml/hr. Similarly, we calculated the results of fired rule for all 71 patients and got results that were in the range of predefined limit by the experts.
Conclusion: It could be done with minimal capital outlay by having a human operator periodically enter MAP and HUO values into a personal computer. The objective of the study was to estimate IFR based on the linguistic description of MAP and HUO sum of these four actions. The rates of change of MAP and HUO could be fuzzified into sets such as DECREASING, STABLE, and INCREASING and would serve to indicate the trend in a patient’s fluid status. This would allow more precise control of fluid balance. Inputs from the domain experts and the judicious use of fuzzy techniques are important to achieve success. This modal is suitable for application only in otherwise healthy patients undergoing surgery involving minimal blood loss. For other patients undergoing surgeries involving moderate-to-severe blood loss, more complicated modals are needed utilising other parameters as well.

Fuzzy logic, fuzzy set, fuzzy expert system, general anaesthesia, mean arterial pressure (MAP), hourly urine output (HUO), intravenous fluid rate (IFR)

The complexity of medical practice makes traditional quantitative approaches of analysis inappropriate. Fuzzy set theory, developed by Professor Lotfi Zadeh (1965) (1), makes it possible to define these inexact medical entities as fuzzy sets. Professor Zadeh coined the term linguistic variable in 1973 and that has opened the doors for fuzzy logic based modeling in the variety of areas of science and technology including medical informatics. The concept of partial membership that occurred to Professor Zadeh, Chair Professor Electrical Engineering and Computer Science (EECS) at University of California Berkeley USA in 1964 while he was visiting parents in New York. In 1973, he coined a new term Linguistic Variable and that has given rise to the term Fuzzy logic that is being extensively used by many in the world. Fuzzy logic is used for a wide variety of devices (2),(3).

Fuzzy logic has been used in applications that are amenable to conventional control algorithms on the basis of mathematical models of the system being controlled, such as the high-frequency mechanical ventilator of Noshiro and coworkers (4). It has a particular advantage in areas where precise mathematical description of the process is impossible and is thus especially suited to support medical decision-making (5).Fuzzy logic is reasoning with fuzzy sets. In medicine, the contradictory natures are common facts.

The sources of uncertainty can be classified as follows (6).
(1) Information about the patient.
(2) Medical history of the patient, which is usually, supplied by the patient and/or his/her family. This is usually highly subjective and imprecise.
(3) Physical examination. The physician usually obtains objective data, but in some cases the boundary between normal and pathological status is not sharp.
(4) Results of laboratory and other diagnostic tests, but they are also subject to some mistakes and even to improper behavior of the patient prior to the examination.
(5) The patient may include simulated, exaggerated, and understated symptoms, or may even fail to mention some of them.
(6) We stress the paradox of the growing number of mental disorders versus the absence of a natural classification (7).

The classification in critical (i.e. borderline) cases is difficult, particularly when a categorical system of diagnosis is considered.

Fuzzy logic plays an important role in medicine (6),(8),(9),(10)examples showing that fuzzy logic crosses many disease groups are the following.
(1)To predict the response to treatment with citalopram in alcohol dependence (11).
(2)To analyze diabetic neuropathy (12) and to detect early diabetic retinopathy (13).
(3)To determine appropriate lithium dosage (14),(5).
(4)To calculate volumes of brain tissue from magnetic resonance imaging (MRI) (16), and to analyze functional MRI data (17).
(5)To characterize stroke subtypes and coexisting causes of ischemic stroke (18),(19),(20),(21).
(6)To improve decision-making in radiation therapy (22). (7) To control hypertension during anesthesia (23).
(8)To determine flexor-tendon repair techniques (24).
(9)To detect breast cancer [25, 26], lung cancer (27), or prostate cancer (28).
(10)To assist the diagnosis of centr

Material and Methods

We selected 71 patient ASA I-II classes in age between 15 and 50 year and weight between 40 and 85 kg, undergoing various surgical procedures. The success of fuzzy rule-based system (fuzzy expert system) depends upon the opinion of the domain experts on various issues related to the study.

Experts’ opinion on MAP, HUO and IFR
The most important parameters for deciding the IFR are MAP and HUO. The opinions of the experts are detailed below.

Map is to be kept within normal Physiological limits. In low MAP, there will be dehydration, blood loss, and any types of shock. Normal MAP is due to normal homodynamic condition. Also high MAP is due to light plane of anaesthesia, hypertensive patient and cardiac diseases (IHD, VHD). In other expert observation, they have got that the low MAP is due to deep plane of anaesthesia and hypotension in optimised patient. The other expert claimed that MAP is very important for vital organs’ blood supply and below 70 mmHg the organs like liver and brain are likely to suffer from ischaemia; in case of hypotensive anaesthesia, the systolic BP can be decreased to the tune of 60 mmHg, where MAP is much below the acceptable lower limit for short period, where the value can be considered as normal. In individuals who are hypertensive, the range is to be maintained on higher side. The higher MAP is undesirable. Other experts claimed that MAP of at least 80 mmHg should be there for adequate vital organ perfusion and peripheral tissues. MAP below 80 mmHg may provide adequate blood to peripheral tissue, significantly producing lactic acidosis, and produce anaerobic metabolism. MAP of greater than 100 mmHg is unnecessary and may actually increase intro-operative blood loss and may result in congested operative field. In other expert idea, the reason was that every vital organ in the body has a range of MAP for its optimal functioning. Below this, the mechanism of auto-regulation fails and the function of that organ will suffer, therefore taking into consideration this range (<60 mmHg) for organs like brain, kidney, liver, and heart. The experts defined their reasons as following: 0.5 ml/kg of urine output is necessary to maintain the kidney function. Low MAP can happen due to dehydration, CRF (chronic renal failure), acute renal failure (ARF) and can also be normal due to normal homodynamic management. High MAP happened due to over hydration, diabetics, non-diabetics and ureoacidosis.

The urine output is low because of dehydration, blood loss, inadequate fluid replacement, major abdominal surgeries and laparoscopy. The urine output was on higher side due to over infusion, lasix intro-operative and high plane of anaesthesia. Optimum urine output is 0.5–1 ml/kg/hr. If higher amount of IV fluid is given, the output will be high. We should label the output high only when it goes above the input. When the urine output is less than 0.5 ml/kg/hr the kidneys suffer; the high level of urine output is undesirable. The urinary output denotes adequate renal perfusion, functionally as well patient hydration status. Urinary output of 0.5–1 ml/kg/hr is sufficient in normally kidney to ensure adequate perfusion. Urine output of more than 1 ml/kg/hr may produce electrolyte imbalance, especially hypokalaemia.

We have made humble attempt to implement the concept of fuzzy rule-based systems that incorporated fuzzy techniques in decision making on the application of IFR. Fuzzy logic algorithm uses the information on only two parameters, in order to arrive to desired level of IFR. These include MAP and HUO. The algorithm considers both the values of these parameters at the time of decision to be made and their rates of change. The values of the parameter are used to arrive at a characterisation of the patient’s current condition, and the rates of the change are used to decide on the trend in this condition. Both current condition and trend are then used to decide if IFR should be altered and by how much. Of c

Results

Construction of fuzzy sets
The first step in the development of the fuzzy logic-based expert system is to construct fuzzy sets for the parameters MAP, HUO and IVF for the various linguistic variables such as low, medium and high in case of MAP, HUO as LOW, MAINTAIN, MODERATE, HIGH and VERY HIGH. These fuzzy sets are designed based on the knowledge base of the domain experts. To put it other way, each parameter has a so-called range of discourse, which is partitioned into a number of overlapping fuzzy sets. The complexity of the fuzzy algorithm increases dramatically with the number of fuzzy sets. Each fuzzy set has amplitude associated with every point in its range that varies between 0 and 1, depending on how strongly a particular point in the range is considered to belong to that set.
The defined procedure was implemented for MAP and HUO as follows:

Considering MAP first, we note that this quantity may be too high, acceptable or too low, so we will divide its range of possible values into three corresponding fuzzy sets. Starting with the set corresponding to acceptable values for MAP, we first ask what range of values for MAP would be designed unquestionably normal. Let this be 70–100 mmHg (not everyone might agree with this, so this choice merely captures the experience of one particular ‘expert’). We thus create a fuzzy set labelled NORMALMAP and assign values of MAP between 70 and 100 mmHg to a membership level of 1.0 in this set ((Table/Fig 1)). Now we address the more vague issue of what range of values for MAP could possibly be normal but might also be abnormal. Let this be 100–120 mmHg at the upper end and 50–70 mmHg at the lower end. In other words, if MAP is above 120 mmHg it is unquestionably too high, whereas between 100 and 120 mmHg it could go either way. Similarly, if MAP is below 55 mmHg it is without doubt too low, whereas between 50 and 70 mmHg there is some doubt about whether it is normal or too low. These uncertainties are represented by membership levels in NORMALMAP that decrease linearly from 1.0 at the inner boundaries of the uncertain regions down to 0 at the outer boundaries ((Table/Fig 1)). We can construct LOWMAP and HIGHMAP fuzzy sets in a similar manner. These begin at the inner boundaries of the uncertain regions with membership levels of 0 and proceed linearly up to membership levels of 1.0 at the outer boundaries, precisely the converse of the situation for NORMALMAP. Above 120 mmHg, we have already established that MAP is too high, so values greater than 120 mmHg have a membership level of 1.0 in HIGHMAP as well as for values of MAP below 50 mmHg in LOWMAP. There is no absolute rule that says the uncertain parts of the fuzzy sets must ascend or descend linearly. However, it is important that the various set memberships always add to unity for every value of the fuzzy variable because membership values essentially represent probabilities of set membership. Straight lines are the most straightforward way of achieving this condition.
We were expecting nine different fuzzy rules, so we asked experts about different levels of IFR and finally from those different opinions we got five different levels for IFR as LOW, MAINTAIN, MODERATE, HIGH and VERY HIGH. Then we asked the experts about the values and definite values of IFR in the same fashion as MAP and HUO. Finally, we have predicted nine different fuzzy rules by using Min–Max approach, and eventually we find out the action that must be taken by using centroid approach.

Fuzzy expert system
A fuzzy expert system is a form of artificial intelligence (computer hardware and software packages that try to emulate human intelligence, using reasoning and learning to solve problems) that uses a collection of membership functions (fuzzy logic) and rules (instead of Boolean

Discussion

Fuzzy logic is utilised for improved monitoring in pre-term infants (36). A self-organising anomaly detection system for an electrocardiogram (ECG) using a fuzzy logic reasoning method was also developed (37). In anaesthesia, many applications have been reported in the use of fuzzy logic to control drug infusion, for maintaining adequate levels of anaesthesia, muscle relaxation, and patient monitoring and alarm. In the field of orthopaedics, there has been no reported application of fuzzy control. The field of anaesthesia is where most of the applications of fuzzy control have been reported. It involves monitoring the patient’s vital parameters and controlling the drug infusion to maintain the anaesthetic level constant. It includes depth of anaesthesia (38), muscle relaxation (39),(40) and hypertension during anaesthesia (41), arterial pressure control (42) and mechanical ventilation during anaesthesia (43), and postoperative control of blood pressure (44). Different methods have been used, which utilise fuzzy logic, the first being a real-time expert system for advice and control (RESAC) based on fuzzy logic reasoning (45). Later examples involve a basic fuzzy logic controller (46), self-organising fuzzy logic controller (47) and hierarchical systems (48). Recent work in anaesthesia monitoring and control concentrated on a multi-sensor fusion system using cardiovascular indicators, such as systolic arterial pressure (SAP), heart rate (HR) and audio-evoked response signals (AER) (49). It is interesting to consider how a fuzzy logic algorithm for controlling fluid balance might be implemented in practice. One of the ways is given below.

It could be done with minimal capital outlay by having a human operator periodically enter MAP and HUO values into a personal computer. Intravenous fluid flow could then be manually adjusted according to the resulting fuzzy logic calculation. However, the best and more efficient approach is to design fuzzy logic-based pump for the management of fluid during anaesthesia. This would greatly increase both reliability and savings in labour. Automation would require the following series of steps: (1) MAP and HUO would be measured at regular intervals by suitable transducers (such as a urine container placed on an electronic scale), (2) the values of MAP and HUO would be acquired by a computer, (3) the fuzzy calculations would be made and (4) the computer would control the fluid delivery rate from a motorised dispenser. Realising these various steps is an engineering problem and is readily soluble, given sufficient resources. It is easy to see how the algorithm could be extended to include additional fuzzy variables such as HR or central venous pressure. The rates of change of MAP and HUO could also be obtained by taking differences between successive hourly measurements. These rates of change could be fuzzified into sets such as DECREASING, STABLE, and INCREASING and would serve to indicate the trend in a patient’s fluid status. This would allow more precise control of fluid balance. Of course, for each additional variable, there is a substantial increase in algorithm complexity because the rule table gains an additional dimension, which means considering many more scenarios. Expert knowledge is the key to success, as all the fuzzy logic-based expert systems needs adequate infuses from the domain experts. A balance, therefore, needs to be struck between the number of independent fuzzy variables used and the number of fuzzy sets for each variable versus the precision of control achieved. This modal is suitable fo

Acknowledgement

This work was supported by Director of Bioinformatics Centre of University of Pune, India; Professor Indira Ghosh; Dr. S. Bahagwat, the HOD of Anaesthesia in Ruby Hall Clinic, and her colleagues; Dr. Kalpana Kerkal, the HOD of Anaesthesia in B. J. Medical Colleges, and her colleagues; Dr. Kane, the HOD of Anaesthesia in Jahangir Hospital, and her colleagues; Dr. F. Ravanshadi, Anaesthetist in Arya Hospital, and his colleagues from Iran; and Mr. Sayed Naser Mossavi, Anaesthetist Assistant in Arya Hospital, Iran.