Analysis of laboratory biochemical results using AI aided model for Hepatocellular Carcinoma Diagnosis and Prediction

Abstract

The aim of this study is to correlate between both Long noncoding RNAs (LncRNAs) and hepatocellular carcinoma (HCC) with an attempt of integrating them in the diagnosis protocol for HCC with the help of an artificial intelligence models. samples were collected and prepared by third partner SHEFAA ALORMAN hospital. Then RNA was extracted using specific extraction kits. Quantitative polymerase chain reaction (qPCR) was used later to determine specific RNA concentration in previously prepared samples. Different models were created using different trial of several data sets. The data sets were integrated into several algorithms, the non-AI traditional results showed low accuracy but by integrating the artificial intelligence in the diagnosis it enhanced the accuracy. The data was presented as mean ± SD where the results of LINC00853 were 29.92 ± 0.006711 with sensitivity and specificity of 97.14% and 95.71% respectively. The results of HULC were 20.49 ± 11.29 with sensitivity 95.71% and specificity 94.29%. Firstly, a model was built by using traditional data (ALT, AST, Total bilirubin and Serum albumin) and showed a higher accuracy than traditional results. Secondly by the evaluation of the value of addition of AFP to the previous parameters and it was d found that the accuracy increased when compared to the model that was trained with the traditional data only. Finally, in order to increase the accuracy of the data integration of novel LncRNAs (Highly up regulated in a liver cancer (HULC)-long intergenic non protein coding 853(LINC00853)) was done where it showed promising result. To conclude the novel LncRNAs biomarkers confirmed their potentiality as a diagnostic tool for HCC diagnosis and their integration in AI model increased the sensitivity and specificity for the diagnosis when compared to traditional data alone.