Farzana Kabir Ahmad

The inference of breast cancer metastasis through gene regulatory networks

Understanding the mechanisms of gene regulation during breast cancer is one of the most difficult problems among oncologists because this regulation is likely comprised of complex genetic interactions. Given this complexity, a computational study using the Bayesian network technique has been employed to construct a gene regulatory network from microarray data. Although the Bayesian network has been notified as a prominent method to infer gene regulatory processes, learning the Bayesian network structure is NP hard and computationally intricate. Therefore, we propose a novel inference method based on low-order conditional independence that extends to the case of the Bayesian network to deal with a large number of genes and an insufficient sample size. This method has been evaluated and compared with full-order conditional independence and different prognostic indices on a publicly available breast cancer data set. Our results suggest that the low-order conditional independence method will be able to handle a large number of genes in a small sample size with the least mean square error. In addition, this proposed method performs significantly better than other methods, including the full-order conditional independence and the St. Gallen consensus criteria. The proposed method achieved an area under the ROC curve of 0.79203, whereas the full-order conditional independence and the St. Gallen consensus criteria obtained 0.76438 and 0.73810, respectively. Furthermore, our empirical evaluation using the low-order conditional independence method has demonstrated a promising relationship between six gene regulators and two regulated genes and will be further investigated as potential breast cancer metastasis prognostic markers.

A review of feature selection techniques via gene expression profiles

The invention of DNA microarray technology has modernized the approach of biology research in such a way that scientists can now measure the expression levels of thousands of genes simultaneously in a single experiment. Although this technology has shifted a new era in molecular classification, interpreting microarray data still remain a challenging issue due to their innate nature of “high dimensional low sample size”. Therefore, robust and accurate feature selection methods are required to identify differentially expressed genes across varied samples for example between cancerous and normal cells. Successful of feature selection techniques will assist to correctly classify different cancer types and consequently led to a better understanding of genetic signatures in cancers and would improve treatment strategies. This paper presents a review of feature selection techniques that have been employed in microarray data analysis. Moreover, other problems associated with microarray data analysis also addressed. In addition, several trends were noted including highly reliance on filter techniques compared to wrapper and embedded, a growing direction towards ensemble feature selection techniques and future extension to apply feature selection in combination of heterogeneous data sources.

Classifying breast cancer types based on fine needle aspiration biopsy data using random forest classifier

Breast cancer is a complex and heterogeneous disease due to its diverse morphological features, as well as different clinical outcome. As a result, breast cancer patients may response to different therapeutic options. Currently, difficulties in recognizing the breast cancer types lead to inefficient treatments. Generally, there are two types of breast cancer, known as malignant and benign. Therefore it is necessary to devise a clinically meaningful classification of the disease that can accurately classify breast cancer tissues into relevant classes. This study aims to classify breast cancer lesions which have been obtained from fine needle aspiration (FNA) procedure using random forest. Random forest is a classifier built based on the combination of decision trees and has been identified to perform well in comparison to other machine learning techniques. This method has been tested on approximately 700 data, which consists of 458 instances from benign cases and 241 instances belong to malignant cases. The performance of proposed method is measured based on sensitivity, specificity and accuracy. The experimental results show that, random forest achieved sensitivity of 75%, specificity of 70% and accuracy about 72%. Thus, it can be concluded that random forest can accurately classify breast cancer types given a small number of features and it works as a promising tool to differentiate malignant from benign tumor at early stage.

Short Term Traffic Forecasting Based on Hybrid of Firefly Algorithm and Least Squares Support Vector Machine

The goal of an active traffic management is to manage congestion based on current and predicted traffic conditions. This can be achieved by utilizing traffic historical data to forecast the traffic flow which later supports travellers for a better journey planning. In this study, a new method that integrates Firefly algorithm (FA) with Least Squares Support Vector Machine (LSSVM) is proposed for short term traffic speed forecasting, which is later termed as FA-LSSVM. In particular, the Firefly algorithm which has the advantage in global search is used to optimize the hyper-parameters of LSSVM for efficient data training. Experimental result indicates that the proposed FA-LSSVM generates lower error rate and a higher accuracy compared to a non-optimized LSSVM. Such a scenario indicates that FA-LSSVM would be a competitor method in the area of time series forecasting.

Designing a robot-assisted therapy for individuals with anxiety traits and states

The recent trend towards developing a new generation of robots capable of operating in human-centered environments, and participating in and assisting our daily lives has introduced the need for robotic systems capable to communicate and to react to their users in a social and engaging way. This type of robot could play essential roles to help individuals with severe cognitive problems. In this paper, several core components to design a robotic assisted therapy to support individuals with anxiety traits and states are presented.

A subject identification method based on term frequency technique

The analyzing and extracting important information from a text document is crucial and has produced interest in the area of text mining and information retrieval. This process is used in order to notice particularly in the text. Furthermore, on view of the readers that people tend to read almost everything in text documents to find some specific information. However, reading a text document consumes time to complete and additional time to extract information. Thus, classifying text to a subject can guide a person to find relevant information. In this paper, a subject identification method which is based on term frequency to categorize groups of text into a particular subject is proposed. Since term frequency tends to ignore the semantics of a document, the term extraction algorithm is introduced for improving the result of the extracted relevant terms from the text. The evaluation of the extracted terms has shown that the proposed method is exceeded other extraction techniques.

Synergy network based inference for breast cancer metastasis

Breast cancer is a world wide leading cancer and it is characterized by its aggressive metastasis. In many patients, microscopic or clinically evident metastases have already occurred by the time the primary tumor is diagnosed. Chemotherapy or hormonal therapy reduces the risk of distant metastasis by one-third, but it is estimated that about 70% to 80% of patients receiving treatment would have survived without it. Therefore, being able to predict breast cancer metastasis can spare a significant number of breast cancer patients from receiving unnecessary adjuvant systemic treatment and its related expensive medical costs. Current studies have demonstrated the potential value of gene expression signatures in assessing the risk of post-surgical disease recurrence. However, most of these studies attempt to develop genetic marker-based prognostic systems to replace the existing clinical criteria, while ignoring the rich information contained in established clinical markers. Clinical markers, such as patient history and laboratory analysis, which are the basis of day-to-day clinical decision support, are often underused to guide the clinical management of cancer in the presence of microarray data. As a result, given the complexity of breast cancer prognosis, we proposed a novel strategy based on synergy network that utilize both clinical and genetic markers to identify the potential hybrid signatures and investigate their interactions which are associated with breast cancer metastasis. In this study, a computational method is performed on publicly available microarray and clinical data. A rigorous experimental protocol is used to estimate the prognostic performance of the hybrid signature and other prognostic approaches. The hybrid signature performs significantly better than other methods, including the 70-gene signature, clinical makers alone and the St. Gallen consensus criterion. At 90% sensitivity level, the hybrid signature achieves 77% specificity, as compared to 53% for the 70-gene signature and 43% for the clinical makers. The predicted results also showed a strong dependence of regulator genes that are related to cell death in cell development process. These significant gene regulators are useful to understand cancer biology and in producing new drug design.

Social spider optimisation algorithm for dimension reduction of electroencephalogram signals in human emotion recognition

Due to some limitations of current heuristics and evolutionary algorithms, this paper proposed a new swarm based algorithm for feature selection method called Social Spider Optimization (SSO-FS). In this research, SSO-FS is used in the EEG-based emotion recognition model as searching method to find optimal feature set to maximize classification performance and mimics the cooperative behaviour and mechanism of social spiders in nature. This proposed feature selection method has been tested on DEAP EEG dataset with six subjects and compared with the most popular heuristic algorithms such as GA, PSO and ABC. The results show that the SSO-FS provides a remarkable and comparable performance compared to other existing methods. Whereby, the max accuracy obtained is 66.66% and 70.83%, the mean accuracy obtained is 55.51±7.17 and 60.97±8.38 for 3-level of valence emotions and 3-level of arousal emotions classification respectively.

Filter-Based Gene Selection Method for Tissues Classification on Large Scale Gene Expression Data

DNA microarray technology is a current innovative tool that has offers a new perspective to look sight into cellular systems and measure a large scale of gene expressions at once. Regardless the novel invention of DNA microarray, most of its results relies on the computational intelligence power, which is used to interpret the large number of data. At present, interpreting large scale of gene expression data remain a thought-provoking issue due to their innate nature of “high dimensional low sample size”. Microarray data mainly involved thousands of genes, n in a very small size sample, p. In addition, this data are often overwhelmed, over fitting and confused by the complexity of data analysis. Due to the nature of this microarray data, it is also common that a large number of genes may not be informative for classification purposes. For such a reason, many studies have used feature selection methods to select significant genes that present the maximum discriminative power between cancerous and normal tissues. In this study, we aim to investigate and compare the effectiveness of these four popular filter gene selection methods namely Signal-to-Noise ratio (SNR), Fisher Criterion (FC), Information Gain (IG) and t-Test in selecting informative genes that can distinguish cancer and normal tissues. Two common classifiers, Support Vector Machine (SVM) and Decision Tree (C4.5) are used to train the selected genes. These gene selection methods are tested on three large scales of gene expression datasets, namely breast cancer dataset, colon dataset, and lung dataset. This study has discovered that IG and SNR are more suitable to be used with SVM while IG fit for C4.5. In a colon dataset, SVM has achieved a specificity of 86% with SNR while and 80% for IG. In contract, C4.5 has obtained a specificity of 78% for IG on the identical dataset. These results indicate that SVM performed slightly better with IG pre-processed data compare to C4.5 on the same dataset.

Combined WSD algorithms with LSA to identify semantic similarity in unstructured textual data

Semantically related sentence may not have any word in common. However, identifying the semantic similarity between words at sentence level possess difficult challenges such as polysemy, synonyms, heterogeneity and sparsity of unstructured textual datasets. It is assumed that sentences with similar text or words in common are semantically related. It means that the standard Information Retrieval (IR) measure based on word co-occurrence are not appropriate to tackle the aforementioned challenges of identifying semantics in unstructured text documents. Many semantic similarity measures have been proposed to resolve this non-trivial issues, but many existing studies did not properly utilize the combination of Corpus and Knowledge-based approach to solve the syntactic construct and the roles of Part Of Speech in identifying semantic similarities in sentences. In this research, we aim at proposing a method for measuring sentence semantic similarity identification that combines two algorithms from the knowledge-based Word Sense Disambiguation algorithms with Latent Semantic Analysis to identify the semantic similarity of sentences and to compare results with human evaluation.