Voice Emotion Recognition using CNN and Decision Tree

This paper presents the use of decision tree and CNN as classifier to classify the emotions from the English and Kannada audio data. The performance of CNN and DT are potential for various emotions. Comparative study of the classifiers using various parameters is presented. The performance of CNN has been identified as the best classifier for emotion recognition. Emotions are recognized with 72% and 63% accuracy using CNN and Decision Tree algorithms respectively. MFCC features are extracted from the audio signals and Model is trained, tested and evaluated accordingly by changing the parameters. Speech Emotion Recognition system is useful in psychiatric diagnosis, lie detection, call centre conversations, customer voice review, voice messages.

CXCR3 Binding Chemokines MIG, IP-10 and ITAC Are Predictors of Overall Survival in Newly Diagnosed Multiple Myeloma

Background. The chemokine receptor CXCR3 and its binding molecules MIG, IP-10 and ITAC have been associated with tumor progression, immune escape and angiogenesis in several human malignancies. In multiple myeloma (MM), CXCR3 binding molecules were shown to induce migration of MM cells without effecting proliferation. More recent results suggest a tumor suppressive activity of IP-10. Presently, information about the precise role of CXCR3 binding chemokines in MM is limited and evidence for their clinical significance is lacking. Therefore we aimed to evaluate the prognostic relevance of CXCR3 binding chemokines in patients with MM. Patients and Methods. Serum levels of MIG, IP-10 and ITAC were analyzed by FACS-CBA array in 65 newly diagnosed MM patients. Expression of CXCR3 and its binding molecules was also analyzed by quantitative PCR in 7 human MM cell lines (HMCLs) and in a publically available gene expression dataset (GSE2658). Further analysis of MIG serum levels was performed by ELISA in an extended cohort of MM (n=105) and MGUS patients (n=17), and in healthy volunteers (n=37). Results. Determination of serum levels by FACS-CBA revealed significant expression of MIG (range: 33.4 – 157 960 pg/ml) and IP-10 (12 - 4418.8 pg/ml), while ITAC (0 - 351.5 pg/ml) was only detectable in a subset (20 of 65) of patients. Interestingly, serum levels of all three molecules showed a positive correlation with each other (MIG vs. IP-10, R=0.38, P=0.002; MIG vs. ITAC, R=0.62, P<0.0001; ITAC vs. IP-10, R=0.41, P=0.0007). We also observed a significant correlation with beta 2 microglobulin (B2M) (MIG: R=0.45, P<0.0001; IP-10: R=0.36, P=0.003; ITAC: R=0.3, P=0.016) and a trend regarding ISS stage (MIG: R=0.23, P=0.06; IP-10: R=0.24, P=0.05; ITAC: R=0.11, P=0.39). Importantly, a significant association with overall survival (OS) was observed as well. Survival was significantly worse in patients with high compared to low MIG (median OS 25.3 months vs. not reached, P=0.003) and IP-10 (19.97 months vs. not reached, P=0.0006) as well as in patients with detectable compared to absent ITAC serum levels (19.97 vs. 65.8 months, P=0.019). In multivariate analysis, MIG (P=0.03) and ITAC (P=0.013) along LDH and calcium were revealed as independent predictors of survival. Expression of CXCR3 binding chemokines was rarely detected in HMCLs (1 of 7 expressed MIG, 3 of 7 IP-10 and 2 of 7 ITAC, respectively). In line with this, in-silico analysis of previously published primary MM cell samples (n=414) (GSE2658), showed a present detection call of MIG, IP-10 and ITAC in 51 (12.3%), 11 (2.7%) and 0 (0%) patients, respectively. In contrast, all three cytokines were detectable in 100% of bone marrow plasma cells of healthy donors, MGUS and smoldering MM patients in this dataset. Hence, CXCR3 binding chemokines are silenced in myeloma cells indicating that the increased serum levels of CXCR3 binding chemokines are derived from other cell types. As MIG serum concentration was identified as one of the most important predictors for OS, we studied the prognostic relevance of this molecule in an extended cohort (n=105) of MM patients by ELISA. Median MIG levels (161.3 pg/ml, range: 9.4-1966) were significantly elevated in newly diagnosed MM patients compared to MGUS (92.7 pg/ml, range: 6.29-1303.1) and healthy volunteers (106.2, range: 51–390.6 pg/ml). MIG levels were significantly correlated with B2M, ISS stage, calcium, albumin, LDH, hemoglobin and with age (R=0.466, P<0.001). Importantly, high MIG levels predicted adverse survival (17.0 months vs. not reached, P<0.001), which was upheld when age-adjusted cut-off levels were used. In accordance with our findings, in-silico analysis of MIG expression in purified plasma cells of MM patients (n=559) treated within the total therapy 2 and 3 protocol (GSE2658) revealed shorter OS in patients with a present compared to those with an absent detection call for MIG (P=0.004). Conclusion. Our findings depict MIG, IP-10 and ITAC as novel prognostic markers for shorter survival in newly diagnosed MM patients. High serum levels of CXCR3 binding chemokines in conjunction with silenced expression in MM cells may shield myeloma cells from immune attack as previously shown for T cell lymphomas. Further experiments will aim to confirm these initial results by extending our patient cohort and define the source as well as functional role of CXCR3 chemokines in MM. Figure 1 Figure 1. Disclosures No relevant conflicts of interest to declare.

Exploring homogeneity of correlation structures of gene expression datasets within and between etiological disease categories

The literature shows that classifiers perform differently across datasets and that correlations within datasets affect the performance of classifiers. The question that arises is whether the correlation structure within datasets differ significantly across diseases. In this study, we evaluated the homogeneity of correlation structures within and between datasets of six etiological disease categories; inflammatory, immune, infectious, degenerative, hereditary and acute myeloid leukemia (AML). We also assessed the effect of filtering; detection call and variance filtering on correlation structures. We downloaded microarray datasets from ArrayExpress for experiments meeting predefined criteria and ended up with 12 datasets for non-cancerous diseases and six for AML. The datasets were preprocessed by a common procedure incorporating platform-specific recommendations and the two filtering methods mentioned above. Homogeneity of correlation matrices between and within datasets of etiological diseases was assessed using the Box’s

Microarray Detection Call Methodology as a Means to Identify and Compare Transcripts Expressed within Syncytial Cells from Soybean (Glycine max) Roots Undergoing Resistant and Susceptible Reactions to the Soybean Cyst Nematode (Heterodera glycines)

Background. A comparative microarray investigation was done using detection call methodology (DCM) and differential expression analyses. The goal was to identify genes found in specific cell populations that were eliminated by differential expression analysis due to the nature of differential expression methods. Laser capture microdissection (LCM) was used to isolate nearly homogeneous populations of plant root cells.Results. The analyses identified the presence of 13,291 transcripts between the 4 different sample types. The transcripts filtered down into a total of 6,267 that were detected as being present in one or more sample types. A comparative analysis of DCM and differential expression methods showed a group of genes that were not differentially expressed, but were expressed at detectable amounts within specific cell types.Conclusion. The DCM has identified patterns of gene expression not shown by differential expression analyses. DCM has identified genes that are possibly cell-type specific and/or involved in important aspects of plant nematode interactions during the resistance response, revealing the uniqueness of a particular cell population at a particular point during its differentiation process.