Fusion of risk assessment models with application to coronary artery disease patients

Abstract

Several risk score models are available in literature to predict death/myocardial infarction event for coronary artery disease (CAD) patients, within a short period of time. However, the choice of the most adequate model is not straightforward since there might not be a consensus about the best model to use in clinical practice Moreover, individually, these models present some weaknesses, such as the inability to deal with missing information. This work addresses these problems, proposing a Bayesian classifier strategy enabling the simultaneous use of several models (models' fusion). Thus, a higher number of risk factors can be used in the common model, while it can deal with missing information. The validation of the strategy is carried out through the combination of three current risk score models (GRACE, TIMI, PURSUIT). Results were obtained based on a dataset that comprises 460 consecutive patients admitted to the Cardiology Department of Santa Cruz Hospital, Lisbon, from 1999 to 2001. A comparison with the voting scheme, which considers exclusively the outputs of models to combine (models output combination) is also carried out. The proposed Bayesian approach had very satisfactory results, confirming the potential of its application to the clinical practice.

The CardioRisk project: Improvement of cardiovascular risk assessment

“…This algorithm focuses on the parameters P(X i |C); P(C) of the global model that was created through Eq. (3) [11].…”

“…where l is the number of individual models, b is the number of individual models that contain the attribute X i ∈ X, C j denotes each individual model that contains X i , w j is the weight of model j [11].…”

“…Recently, as a result of the HeartCycle project, a Bayesian strategy for the fusion of several risk tools has been developed [11]. Using this approach a higher number of risk factors can be simultaneously considered and missing information can be dealt with (a characteristic of Bayes' probability theory).…”

Improvement of CVD risk assessment tools' performance through innovative patients' grouping strategies

“…These tools are very useful although they present some important weaknesses: i) they ignore the information provided by other risk assessment tools that were previously developed, ii) each individual tool considers a reduced number of risk factors, iii) they have difficulty in coping with missing risk factors, iv) they do not allow the incorporation of additional clinical knowledge, v) some tools do not assure the clinical interpretability of the respective parameters. These problems have already been addressed in previous works of this research team [3][4] [5].…”

“…The reduction of dimensionality can be formalized as: given a P dimensional data vector 5 The intrinsic dimensionality of data is the minimum number of parameters needed to account for the observed properties of data [10].…”