 Open Access
Statistical independence for the evaluation of classifierbased diagnosis
 Emanuele Olivetti^{1, 2},
 Susanne Greiner^{1, 2} and
 Paolo Avesani^{1, 2}Email author
 Received: 31 May 2014
 Accepted: 19 November 2014
 Published: 11 December 2014
Abstract
Machine learning techniques are increasingly adopted in computeraided diagnosis. Evaluation methods for classification results that are based on the study of one or more metrics can be unable to distinguish between cases in which the classifier is discriminating the classes from cases in which it is not. In the binary setting, such circumstances can be encountered when data are unbalanced with respect to the diagnostic groups. Having more healthy controls than pathological subjects, datasets meant for diagnosis frequently show a certain degree of unbalancedness. In this work, we propose to recast the evaluation of classification results as a test of statistical independence between the predicted and the actual diagnostic groups. We address the problem within the Bayesian hypothesis testing framework. Different from the standard metrics, the proposed method is able to handle unbalanced data and takes into account the size of the available data. We show experimental evidence of the efficacy of the approach both on simulated data and on real data about the diagnosis of the Attention Deficit Hyperactivity Disorder (ADHD).
Keywords
 Attention Deficit Hyperactivity Disorder
 Prediction Accuracy
 Attention Deficit Hyperactivity Disorder
 Confusion Matrix
 Confusion Matrice
1 Introduction
Classificationbased machine learning techniques are increasingly adopted in computeraided diagnosis because they have limited need for a pathophysiological model of the disease under investigation. The efficacy of such modelfree approaches depends on many factors, like the size of the available training sample. A bigger sample size allows for the training of a more robust classifier and might improve the prediction accuracy (PA) on the test set. However, even an enormous amount of data does not guarantee the correct diagnosis of a disease via classifier.
Once a suitable classification algorithm has been trained, its efficacy has to be assessed by predicting the diagnostic groups of subjects in a test set and comparing them against the true values. It is common practice to calculate one or more metrics such as the PA, F1Score, Matthews Correlation Coefficient, \(\kappa \)statistic [6, 11, 12] or AUC/ROC [3, 13] to decide whether the classifier is able to discriminate between healthy controls and one or more stages/types of the pathology of interest. Each of those metrics has different strengths and drawbacks. For example, PA is not able to properly handle datasets where the number of available examples per class is not equal, a setting referred to as unbalanced. More importantly, common metrics for evaluating classifiers do not depend on the actual test set size, i.e. they do not measure the amount of evidence the results of prediction provide.
The confusion matrix is a convenient way to represent results of a classifier because all the metrics used to evaluate classifiers can be computed from it. In the same way, the method that we propose in this work is based on the analysis of the confusion matrix. Specifically, we propose to quantify the evidence between two alternative hypotheses about the underlying generation mechanism of the observed confusion matrix. The first hypothesis is that the predicted class labels are statistically independent from the true class labels. This is the case were the classifier is not able to discriminate the classes. The second hypothesis is that the predicted class labels are statistically dependent on the true class labels. In this case, the classifier predicted according to the true class labels. The degree of evidence in favour of one hypothesis or of the other is the measure that we propose for evaluating the classifier.
In order to implement the proposed method, we draw from the statistics literature and adopt a recent Bayesian test of independence for contingency tables [2], which was proposed in a context different from that of classification. The proposed method is able to handle imbalancedness, takes the sample size of the test set into account and provides the correct answer in cases in which standard metrics are misleading. Furthermore, this kind of approach can be extended to the multiclass setting, while traditional evaluation methods are often tailored and limited to the binary setting. We defer the presentation of the multiclass case to future work.
In the following, we describe the standard classification task for diagnosis in medical image analysis and define where the problem concerning result evaluation lies. Subsequently, we introduce the Bayesian test of independence and show its efficacy on a simulated toy example and on real data concerning the computeraided diagnosis of the attention deficit hyperactivity disorder (ADHD).
2 Methods
The first part of this section formally defines the notation and framework of classificationbased diagnosis. The second part introduces the Bayesian hypothesis testing framework and the proposed solution to the problem of evaluating the classification result.
2.1 Classificationbased diagnosis
Let \({\mathcal{X}}={\mathbb{R}}^d\) be the multidimensional feature space under investigation, e.g. medical image data, and let \({\mathcal{Y}}=\{1,\ldots ,c\}\) be the set of classes that represents the possible values of the variable of interest.
Each training example is then a vector \({\mathbf{X}} \in {\mathcal{X}}\), e.g. the data from one subject of the study, with class label \(Y \in {\mathcal{Y}}\), e.g. the subject’s pathology. Let \(P_{XY}\) be the unknown joint distribution over \({\mathcal{X}} \times {\mathcal{Y}}\). We are given a previously trained classifier \(f:{\mathcal{X}} \rightarrow {\mathcal{Y}}\) that predicts the variable of interest given the data about which the performance is to be determined. We call \(\epsilon = E_{XY}[f({\mathbf{X}}) \ne Y]\) the generalization error of \(f\).
In practical cases, the test set is of finite size \(m\), therefore we do not know the actual joint distribution \(P_{XY}\) and \(\epsilon \) can only be estimated. Let \(S=\{(x_1,y_1),\ldots ,(x_m,y_m)\}\) be the test set which is assumed to be an i.i.d. set of observations, i.e. examples, drawn from \(P_{XY}\). The standard estimator of \(\epsilon \) is \(\hat{\epsilon } = \frac{e}{m},\) where \(e\) is the total number of misclassified examples.
The set of true class labels and predicted class labels can be summarized by the confusion matrix \(\varvec{Z}\), which is a contingency table (see Fig. 1) that reports the number of occurrences \(z_{ij}\) of each possible pair of outcomes. The sum \(\sum {z_{ij}} = m\) equals the test set size and the diagonal contains all correctly classified examples \(\sum _{i}{z_{ii}} = me\). The estimated PA is defined as \(PA = \frac{me}{m} = 1\hat{\epsilon }\).
2.2 Evaluation by the Bayesian test of independence
As noted in [8], when data are unbalanced with respect to the classlabel distribution, the PA (or the misclassification error rate) of a classifier can be a misleading statistic to assess whether the classifier actually discriminated the classes or not. An alternative solution to the issue of evaluating classifiers through the error rate/accuracy is testing the full confusion matrix.

\({\bf{H_0}}\): the predictions are statistically independent of the true class labels.

\({\bf{H_1}}\): the predictions are statistically dependent on the true class labels.
Guidelines for the interpretation of the logarithm of the Bayes factor \(log(B_{10})\) in terms of the strength of evidence in favour of \(H_1\) and against \(H_0\), from [5]
\(log(B_{10})\)  <0  0 to 1  1 to 3  3 to 5  >5 

Strength  Negative  Bare mention  Positive  Strong  Decisive 
In order to compute \(B_{10}\) for the hypotheses of interest of this work, it is necessary to define a sampling model for the confusion matrix \(\varvec{Z}\) under each hypothesis. Notice that while evaluating the classification results, the total number of examples per class in the test set can be assumed as known. This assumption is usually known as one margin fixed and it means that the row marginals of \(\varvec{Z}\) are known and then that the sampling model for each row of the confusion matrix is \({\mathrm{Bin}} (z_in_i,p_i)\), where \(z_i\) is one of the two values of the \(i\)th row (the other being \(n_i  z_i\)), \(n_i\) is the known \(i\)th row marginal and \(p_i\) the unknown probability of predicting that class when the true class is \(i\).
In [2], it is shown how to extend Eq. 8 to the multiclass case, which we do not present here.
3 Materials: the ADHD dataset
Our study refers to the ADHD200 Initiative and dataset which is dedicated to support the scientific community in studying and understanding the neural basis of ADHD. The aim of the initiative is also meant to support the clinical community with the advance of objective tools for computeraided diagnosis. Eight institutions collected neuroimaging datasets from almost one thousand young subjects (age 7–26) with and without ADHD. For each subject, multiple types of data were collected: phenotypic data, structural (T1) magnetic resonance imaging (MRI) data and functional MRI (fMRI) restingstate data. Accompanying phenotypic information included: age, gender, handedness and IQ measure. The ADHD200 dataset is publicly available and freely distributed with the support of the International Neuroimaging Datasharing Initiative.^{1}
Even though the ADHD200 dataset comprised three different levels of the ADHD disorder and the healthy controls, in this work, we restrict our analysis to the discrimination between two diagnostic categories, i.e. healthy controls and ADHD patients, by aggregating patients into one class. In the following, we refer to the whole dataset comprising the data of 1339 recordings from 923 subjects, where the diagnostic classes are distributed as follows: 62 % typically developing control and 38 % ADHD. For a few subjects, data were only partially available or corrupted. These subjects were excluded from our study.
In this work, we analyse the confusion matrices presented in [7]. We report a brief summary of the preprocessing and classification steps because a detailed presentation is beyond the scope of this paper and it can be found in [7]. The preprocessed data were retrieved from the NeuroBureau initiative^{2} and specifically from the Athena and Bruner pipelines managed by C. Craddock and C. Chu. Both structural (T1) volumes and statistical volume from fMRI restingstate recordings were transformed into vectors through the dissimilarity representation [9]. The classification algorithm adopted was the extremely randomized tree [4] with different crossvalidation schemes.
In Sect. 4, we use the confusion matrices obtained in [7] from phenotypic data (denoted as PHEN) and fMRI restingstate data preprocessed according to the spatial multiple regression proposed in [10] (denoted as SMR09).
4 Experiments
We compared the efficacy of the proposed test of independence against multiple standard metrics introduced in Sect. 2. Experiments were conducted on data from a simulated toy example and on real data concerning the computeraided diagnosis of ADHD brain disease. The code of the experiments is freely available from https://github.com/FBKNILab/brin2014.
4.1 Simulated toy example
In Table 2, we observed that perfect prediction, i.e. (c), produces the highest scores for all the metrics considered, i.e. 1.0. The score of the proposed method, i.e. \(\log (B_{10}) = 19.61\), means decisive evidence in favour of \(H_1\), according to the interpretation guidelines in Table 1. So it agrees with all other metrics. The case of perfectly random prediction irrespective of the prior distribution, i.e. (d), is again correctly detected by all methods by scoring \(0.0\), with the exception of the \(F1\) score. The score of the proposed method, i.e. \(\log (B_{10}) = 0.94\), is negative evidence for \(H_1\) ^{3} in agreement with most of the standard metrics.
The comparison of the cases (a) and (b) of Table 2 shows that prediction accuracy (\(PA\)) and \(F1\) score are not reliable for unbalanced datasets. The related confusion matrices represent opposite situations but those scores do not significantly change. For the confusion matrices in (a), the Matthews Correlation Coefficient (MCC) and Youden’s J score are undefined and only the \(\kappa \)statistic correctly detects the difference between (a) and (b). In agreement with the \(\kappa \)statistic, the proposed method reports negative evidence for \(H_1\) for case (a) and decisive evidence for \(H_0\) for case (b).^{4}
In Table 3 the confusion matrices represent the same situations of those in Table 2 but with a reduced number of examples. This means that their interpretation in terms of scores must go in the same direction but the amount of evidence provided in Table 3 is much lower than that of Table 2 and the evaluation has to take that into account. In other words, we cannot draw the same conclusions from a test set of 20 examples with respect to a test set of 100 examples and this should be represented in the scores. As it can be seen from the comparison of the scores in Table 3 with respect to those in Table 2, all the standard metrics provide the same exact scores despite having \(1/5\) of the data. Different from them, the proposed method shows a great reduction in value, correctly reflecting the reduced size of the test set. For example, in case (b), the amount of evidence in favour of \(H_1\) is decisive in Table 2 (\(\log B_{10} = 10.67\)) but only worthy of a bare mention in Table 3 (\(\log B_{10} = 1.84\)).
4.2 Realdata application
The results shown in Table 4 about SMR5 and SMR7 have prediction accuracy of 61 % in both cases. The analysis of the confusion matrix by means of the test of independence reveals that SMR5 does not provide relevant information about ADHD diagnosis, while SMR7 provides strong evidence in support of \(H_1\). The predictions in the latter case are therefore statistically dependent on the true class labels and a positive answer to the question, whether the classifier learned to discriminate the classes, can be given. Notice that \(MCC,\, \kappa \) and \(J\) show little increase from SMR5 to SMR7, making it difficult to detect such difference.
A substantially similar result can be obtained on SMR2 vs. PHEN, in Table 4. The prediction accuracy is again at the same level in both cases: while the SMR2 is found to obtain positive evidence, PHEN has a \(\log (B_{10}) = 9.58\), which is decisive evidence for statistical dependence between predicted and true class labels. Other standard metrics, i.e. \(MCC,\, \kappa \) and \(J\), shows a small increase in value but the absence of interpretation guidelines, as those in Table 1, makes it difficult to understand the practical meaning of those changes.
Furthermore, the comparison of SMR7 vs. SMR2 shows another example for how the prediction accuracy may be misleading. The former has the lower prediction accuracy, but strong evidence (\(\log (B_{10})=4.44\)) that the classifier might have learned to discriminate the classes, while the latter has a slightly higher prediction accuracy, but only positive evidence (\(\log (B_{10})=2.98\)).
As a general comment, the ranking of relevance for diagnosis of the four different data sources is in agreement when considering the proposed method based on Bayesian inference and the Matthews correlation coefficient or the \(\kappa \)statistic. The main difference is that the result of the proposed method has a direct interpretation in terms of evidence, while the significance of the differences in the values of the standard metrics across the confusion matrices remains to be determined.
5 Discussion
In this work, we propose a novel method for the evaluation of classification results that overcome the limitations of commonly adopted metrics. The proposed method is based on the Bayesian inference framework and provides a measure of evidence in the data that can be easily interpreted by means of standard guidelines. This differs from standard metrics where guidelines for interpretation are not available due to the lack of a statistical foundation.
Additionally, in Sect. 4.1, we show that the proposed method agrees with standard metrics in many cases. But it is the only one able to provide the correct answer in more extreme cases, where standard metrics are either undefined or misleading.
In Sect. 4.2, on real data, we show that the proposed method distinguishes between data sources that are of importance for the discrimination between the diagnostic groups of ADHD from those who are not. This is sometimes in contrast with prediction accuracy that may lead to incorrect conclusions (see SMR7 vs SMR2).
The accurate detection of data sources which are irrelevant to diagnosis can lead to their exclusion from diagnosis protocols and therefore to improve the costbenefit tradeoff. The proposed Bayesian test of independence is an effective tool for such task.
Declarations
Acknowledgments
The authors are grateful to the ADHD200 Initiative for providing the ADHD200 data and setting up the Global Competition. We would like to thank the NeuroBureau Initiative for providing the preprocessed version of the ADHD200 data and specifically Cameron Craddock for the Athena pipeline and Carlton Chu for the Burner pipeline.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.
Authors’ Affiliations
References
 Berger JO, Pericchi LR (1996) The intrinsic Bayes factor for model selection and prediction. J Am Stat Assoc 91(433):109–122View ArticleMATHMathSciNetGoogle Scholar
 Casella George, Moreno Elias (2009) Assessing robustness of intrinsic tests of independence in twoway contingency tables. J Am Stat Assoc 104(487):1261–1271View ArticleMathSciNetGoogle Scholar
 Fawcett T (2006) An introduction to ROC analysis. Pattern Recognit Lett 27(8):861–874View ArticleMathSciNetGoogle Scholar
 Geurts P, Ernst D, Wehenkel L (2006) Extremely randomized trees. Mach Learn 63(1):3–42View ArticleMATHGoogle Scholar
 Kass RE, Raftery AE (1995) Bayes factors. J Am Stat Assoc 90(430):773–795View ArticleMATHGoogle Scholar
 Landis RJ, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33(1):159–174View ArticleMATHMathSciNetGoogle Scholar
 Olivetti E, Greiner S, Avesani P (2012) ADHD diagnosis from multiple data sources with batch effects. Front Neurosci 6 (70)Google Scholar
 Olivetti E, Greiner S, Avesani P (2012) Induction in neuroscience with classification: issues and solutions. Springer, Berlin (Lecture Notes in Compuer Science)Google Scholar
 Pekalska E, Paclik P (2002) A generalized kernel approach to dissimilaritybased classification. J Mach Learn Res 2:175–211MATHMathSciNetGoogle Scholar
 Smith SM, Fox PT, Miller KL, Glahn DC, Fox MM, Mackay CE, Filippini N, Watkins KE, Toro R, Laird AR, Beckmann CF (2009) Correspondence of the brain’s functional architecture during activation and rest. Proc Natl Acad Sci USA 106(31):13040–13045View ArticleGoogle Scholar
 Okeh UM, Okoro CN (2012) Evaluating measures of indicators of diagnostic test performance: fundamental meanings and formulars. J Biom Biostat 3(1):2Google Scholar
 Van Rijsbergen CJ (1979) Information retrieval, vol 30. Butterworths, LondonGoogle Scholar
 Youden WJ (1950) Index for rating diagnostic tests. Cancer 3(1):32–35View ArticleGoogle Scholar