Computational models for predicting anticancer drug efficacy: A multi linear regression analysis based on molecular, cellular and clinical data of oral squamous cell carcinoma cohort. (September 2019)

• Record Type: Journal Article
• Authors: Robert, Beaulah Mary; Brindha, G.R.; Santhi, B.; Kanimozhi, G.; Prasad, Nagarajan Rajendra
• Is Part Of: Computer methods and programs in biomedicine. Volume 178(2019)
• Page Start: 105
• Page End: 112
• Publication Date: 2019-09
• Subjects: Oral squamous cell carcinoma -- Multi linear regression -- Precision medicine -- Computational models -- Drug efficacy prediction; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2019.06.011
• Languages: English
• ISSNs: 0169-2607

Abstract

Highlights: The enhanced MLR-WLS model was found to be the most effective and accurate method for selection of anticancer drugs based on tumor features of the patients in a personalized manner. Increasing more features like the mode of action of the drugs, the chemical structure of anticancer agents as additional features will improve the robustness of the developed model. The enhanced MLR_WLS model can also be used for the efficacy prediction of individual drugs. Abstract: Background and objectives: The computational prediction of drug responses based on the analysis of multiple clinical features of the tumor will be a novel strategy for accomplishing the long-term goal of precision medicine in oncology. The cancer patients will be benefitted if we computationally account all the tumor characteristics (data) for the selection of most effective and precise therapeutic drug. In this study, we developed and validated few computational models to predict anticancer drug efficacy based on molecular, cellular and clinical features of 31 oral squamous cell carcinoma (OSCC) cohort using computational methods. Methods: We developed drug efficacy prediction models using multiple tumor features by employing the statistical methods like multi linear regression (MLR), modified MLR-weighted least square (MLR-WLS) and enhanced MLR-WLS. All the three developed drug efficacy prediction models were then validated using the data of actual OSCC samples (train-test ratio 31: 31) and actual Vs hypothetical samples (train-test ratio 31: 30). The selected best statistical model i.e. enhanced MLR-WLS has then been cross-validated (CV) using 341 theoretical tumor data. Finally, the performances of the models were assessed by the level of learning confidence, significance, accuracy and error terms. Results: The train-test process for the real tumor samples of MLR-WLS method revealed the drug efficacy prediction enhancement and we observed that there was very less priming difference between actual and predicted. Furthermore, we found there was a less difference between actual apoptotic priming and predicted apoptotic priming for the tumors 6, 8, 21 and 30 whereas, for the remaining tumors there were no differences between predicted and actual priming data. The error terms (Actual Vs Predicted) also revealed the reliability of enhanced MLR-WLS model for drug efficacy prediction. Conclusion: We developed effective computational prediction models using MLR analysis for anticancer drug efficacy which will be useful in the field of precision medicine to choose the choice of drug in a personalized manner. We observed that the enhanced MLR-WLS model was the best fit to predict anticancer drug efficacy which may have translational applications.