Screening of Key Transcripts from Expression Data Using Applied Artificial Intelligence for Cancer Prediction

Abstract

Abstract Research on the effects of artificial intelligence (AI)-driven solutions in the field of oncology is still being conducted all around the world. Applications of AI to identify the transcripts that cause cancer are being investigated, particularly when employing ensemble learning techniques. Ensemble feature fusion is the process of distributing the feature selection process and combining the local features that were chosen to create a smaller global feature set. This article addresses the use of ensemble feature fusion in the field of differential transcript expression analysis to screen for important transcripts linked to a disease. Owing to the necessity and significance of research in cancer diagnosis, an ensemble feature fusion approach experimental case study has been carried out using 109 liver cancer samples obtained from Mitranscriptome dataset. It was found that the expression data were used to filter the most pertinent transcripts, which allowed for the best possible differentiation between sample type. Accordingly, a set of 26 significant liver cancer causing transcripts has been screened using unanimous voting-scheme, giving an accuracy of percentage of 96. During generalization testing, cancer prediction classifiers constructed with this essential transcript collection shown excellent discriminating power and performed well in differentiating between normal and malignant cells. By resolving the “high dimension-low sample (High p Low n)” issue that is typically present in the expression data, this improves the predicting ability of cancer diagnostic systems. In the field of oncology, artificial intelligence-powered solutions will facilitate the development of applications that prioritize Sustainable Development Goal 3: Good Health and Well-Being.

Keywords

• Ensemble classifier
• Transcripts
• Cancer prediction
• Mitranscriptome