The Combination of Multiple Classifiers Using an Evidential Reasoning Approach

In many domains when we have several competing classifiers available we want to synthesize them or some of them to get a more accurate classifier by a combination function. In this paper we propose a ‘class-indifferent’ method for combining classifier decisions represented by evidential structures called triplet and quartet, using Dempster's rule of combination. This method is unique in that it distinguishes important elements from the trivial ones in representing classifier decisions, makes use of more information than others in calculating the support for class labels and provides a practical way to apply the theoretically appealing Dempster–Shafer theory of evidence to the problem of ensemble learning. We present a formalism for modelling classifier decisions as triplet mass functions and we establish a range of formulae for combining these mass functions in order to arrive at a consensus decision. In addition we carry out a comparative study with the alternatives of simplet and dichotomous structure and also compare two combination methods, Dempster's rule and majority voting, over the UCI benchmark data, to demonstrate the advantage our approach offers. (A continuation of the work in this area that was published in IEEE Trans on KDE, and conferences)

Clinical and immunohistochemical assessment of vulval intraepithelial neoplasia following photodynamic therapy using a novel bioadhesive patch-type system loaded with 5-aminolevulinic acid

Background: The work in this study appraised photodynamic treatment (PDT) as a treatment method for vulval intraepithelial neoplasia (VIN) using a novel bioadhesive patch to deliver aminolevulinic acid. An analysis of changes in expression of apoptotic and cell cycle proteins (p53, p21, Mdm2, Blc-2, Bax, Ki-67) in response to PDT was evaluated. Methods: PDT was performed using non-laser light, either as a one or two-cycle treatment, with clinical and pathological assessment following after 6 weeks. Twenty-three patients with 25 VIN lesions underwent 49 cycles of PDT Patches were designed to conform to uneven vulval skin and contained 38 mg cm(-2) aminolevulinic acid. Assessment was carried out at 6 weeks post-treatment. Patient-based treatment assessment, along with clinical and pathological changes, were monitored. Immunohistochemical staining was used to elucidate a possible biomolecular basis for induced cellular changes. Results: Most patients (52%) reported a symptomatic response, with normal pathology restored in 38% of lesions. The patch was easy to apply and remove, causing minimal discomfort. Fluorescence inspection confirmed protoporphyrin accumulation. Pain during implementation of PDT was problematic, necessitating some form of local analgesia. Changes in expression of cell cycle and apoptotic-related proteins suggested involvement of apoptotic pathways. Down regulation of p21 and inverse changes in Bcl-2 and Bax were key findings. Conclusion: Treatment of VIN lesions using a novel bioadhesive patch induced changes in cell cycle and apoptotic proteins in response to PDT with possible utilisation of apoptotic pathways. The efficacy of PDT in treating VIN could be improved by a better understanding of these apoptotic mechanisms, the influence of factors, such as HPV status, and of the need for effective pain management.

Performing meta-analysis with incomplete statistical information in clinical trials

Background: Results from clinical trials are usually summarized in the form of sampling distributions. When full information (mean, SEM) about these distributions is given, performing meta-analysis is straightforward. However, when some of the sampling distributions only have mean values, a challenging issue is to decide how to use such distributions in meta-analysis. Currently, the most common approaches are either ignoring such trials or for each trial with a missing SEM, finding a similar trial and taking its SEM value as the missing SEM. Both approaches have drawbacks. As an alternative, this paper develops and tests two new methods, the first being the prognostic method and the second being the interval method, to estimate any missing SEMs from a set of sampling distributions with full information. A merging method is also proposed to handle clinical trials with partial information to simulate meta-analysis.