ARTIFICIAL INTELLIGENCE TO CLASSIFY HUMAN LUNG CARCINOMA USING BLOOD PLASMA FTIR SPECTRA

Early diagnosis is pivotal for the survival rate in lung cancer patients. However, despite the great successes in various cancer types, effective and accepted screening techniques do not exist for lung cancer. Artificial Intelligence was employed to predict human lung carcinoma based on a model built using the biomarker module of MetaboAnalyst 4.0 software. FTIR spectra of human blood plasma collected from two groups (healthy and lung carcinoma patients) were used to build a model for classification. Using the model improved by removal of outlier samples, Linear SVM, PLS-DA and Random Forest algorithms were employed on unknown samples to predict the classification of lung carcinoma and healthy samples. Prediction accuracy (80-90 %) indicates that Artificial Intelligence based on FTIR spectra of human blood plasma can be developed as a widespread screening method for the identification of individuals at high risk. This screening method is fast, minimally invasive and cheap.

Automated summary

Early diagnosis is crucial for the survival of lung cancer patients, yet effective screening techniques are lacking. Artificial Intelligence was utilized to predict human lung carcinoma based on a biomarker model, showing promise in developing a fast, minimally invasive and cost-effective screening method.