PROSTATE CANCER PREDICTION USING BI-LSTM

Abstract

The study and classification demonstrated here using ensemble based Bi-LSTM network is developed for prostate cancer detection utilizing micro-array data. The ensemble based Bi-LSTM network includes two main phases such as informative gene selection and classification. The simulation evaluation revealed that the ensemble based Bi-LSTM network achieved better results and accuracies for the given dataset.

Specification

Description:The little gland called the prostate, which is situated in front of the rectum and beneath the bladder in males and is essential to the production of seminal fluid, can develop cancer of the prostate type. One of the most prevalent forms of cancer in males, it sometimes goes years without showing any symptoms. Age, family history, and specific genetic predispositions are risk factors. While many men may not have symptoms in the early stages, symptoms can include pelvic discomfort, blood in the urine or semen, and trouble urinating. Digital rectal exams (DREs), imaging investigations, and prostate-specific antigen (PSA) blood tests are commonly used in the diagnosis process. Depending on the stage of the malignancy, there are many treatment choices such as radiation therapy, surgery, active surveillance, hormone therapy, or chemotherapy. Early detection and treatment are crucial for improving outcomes and survival rates , C , Claims:1) Prostate Cancer Prediction Using Bi-LSTM (100) comprises:
i) Data Pre-Processing (101) involves Sequence Cleaning, Handling missing data, Encoding if required, Normalization, Dimensionality reduction, Format conversion if required. This step improves the quality and consistency of the data, making it suitable for tasks like feature extraction and classification.
ii) Ensemble feature optimization (102) involves using multiple algorithms or methods to select features, then combining their results to create a more robust and effective feature set for model training. This approach can help capture the complementary strengths of different methods, leading to improved predictive performance.
iii) Informative gene selection (103) is a critical process in bioinformatics and computational biology, especially for tasks like gene expression analysis, disease classification, and biomarker discovery. The goal is to identify a subset of genes that provide the most relevant information for a specific biological question or model.
iv) Bidirectional Long Short Term Memory (Bi-LSTM) (104) network is employed for detecting prostate cancer from the micro-array data of gene expression.

2) Prostate Cancer Prediction Using Bi-LSTM (100) as claimed in claim 1 uses Ensemble feature optimization techniques (102) FOA, PSO, ABC and GWO then combining their results to create a more robust and effective feature set for model training.
3) Prostate Cancer Prediction Using Bi-LSTM (100) as claimed in claim 1 then makes Informative gene selection (103) for selecting the best suitable genes for the Prostate Cancer classification.
4) AI Enabled Marine Animal Recognition System (100) as claimed in claim 1 will then use Bi-LSTM for classification of the data as Prostate Cancer or Normal subject.

Documents