|
Research Tools
Introduction
Currently, roughly thirty percent of coronary artery bypass graft (CABG) patients
develop atrial fibrillation (AF) in the five days following surgery, increasing the risk of
stroke, prolonging hospital stay three to four days, and increasing the overall cost of the
procedure. According to some sources, over $1 billion is spent annually on this
problem in the US alone. Current pharmacologic and nonpharmacologic means of AF
prevention are suboptimal, and their side effects, expense, and inconvenience limit their
widespread use in all patients.
The main objective of this research is to develop a Bayesian network (BN)
classifier which can model/predict/assign risk of the occurrence of atrial fibrillation in
coronary artery bypass graft patients through the incorporation of different types of
patient data. Expert knowledge coming from doctors in the field will be combined using
Bayesian statistics with patient data and electrocardiogram (ECG) analysis, improving on
the Frequentist methods currently used. We intend to investigate profit or loss due to the
inclusion of the following data types:
- Collected Data- Risk factors and other medical indicators recorded in the hospital
after CABG
- ECG Features- Time, frequency, wavelet, and nonlinear domain features derived
from the ECG signal showing AF prediction potential
- Expert Knowledge- Cardiologist modified probability distribution and frequency
beliefs of input data based on past experience
Electrocardiogram (ECG)
The electrical activity in the heart can be recorded, resulting in the
electrocardiogram (ECG). This electrical signal is usually monitored with an electrode on
the surface of the chest, which records the voltage at that point in relation to time. Unlike many collected biosignals, ECGs contain strong reference points in the
signal, from which much information can be determined.
The R waves’ large amplitude
on the chest lead allows for identification of the ventricle contraction times, while the
atrial contraction can be determined from the intra-atrial lead.
When something is wrong in ny part of the heart's conduction system, the ECGs shape changes,
making it a useful tool for identifying the problem. For this reason, we derive mathmatical features from the signal in order to identify patients which have a disease.
Bayesian Networks
 A Bayesian network (BN) is a relatively new tool that uses probabilistic
correlations among multiple variables to make predictions or assessments of class
membership based on past data. The probabilistic relationships are represented in a network and can be used along with Bayes therom. The use of probabilities derived from past data is
similar to how a doctor currently makes decisions; A doctor assesses the past
occurrences of these symptoms and test results to determine at a likely diagnosis for a
current case. When a Bayesian network is used for risk stratification, classification results
and probabilistic context can be output together, allowing the doctor to observe why the
network made a decision, instead of the black box method where the doctor does not
understand the inner-workings and therefore will not trust it in a clinical setting.
Bayesian Statistics
Bayesian statistics is a structured method for the combination of objective data
(experimentally collected) and subjective opinion to predict future outcomes. For this
research, it is uniquely suited to incorporate expert knowledge from cardiologists into
probabilities from given data sets in order to assign risk. The data probabilities are easy to
calculate, but the quantification of an expert opinion is a little more difficult. This must
be made into a distribution with which it is easy to perform calculations. For this reason,
the data is converted to binary and a binomial distribution is used for representation of
the data's probabilities. |
|
|
|
|