NeuroInteract: An Innovative Deep Learning Strategy for Effective Drug Repositioning in Schizophrenia Therapy

Schizophrenia (SCZ) is a serious physiological and neurological disorder that affects an individual’s perception of factuality. It expresses different symptoms such as thinking, aberrant behavior, delusions, and hallucinations. An efficient approach for inferring potential indications for drugs is through drug repositioning. In this context, drug repositioning imparts a valuable strategy to gain safer, faster, and potentially efficient treatment options to improve schizophrenia therapy. Current treatments are insufficient and existing drug repositioning methods are unsuccessful in solving the drug-disease interactions’ difficulties, including long-term efficacy, drug synergy, and capturing genetic variations. Also, existing methods are restrained because of the incapacity to efficiently integrate heterogeneous biomedical data, which results in suboptimal predictions. This research introduces a NeuroInteract model using deep learning in order to predict candidate drugs for SCZ therapy. The proposed model enhances the accuracy of drug repositioning through the collection of various data sources such as genetic information and drug-disease associations. The novelty of the proposed model is the utilization of the heterogeneous data network that is integrated with the progressive optimization model for the purpose of improving prediction accuracy. The developed method imparts effective learning from various data characteristics through the integration of various types of neural network layers such as fully connected layers, convolutional layers, recurrent layers, and graph convolutional layers. The collected data from DrugBank 5.0 and repoDB undergoes a process of data integration, which aids in generating precise predictions for candidate drugs for repositioning. A data pre-processing technique is employed to improve the data quality. After data pre-processing, the proposed method effectively extracts the meaningful features and finds the spatial dependencies to predict the potential candidate drugs for SCZ treatment. Also, it efficiently handles sequential dependencies and genetic information. The oppositional crossover boosted meerkat optimization (OCMO) algorithm is deployed to optimize the performance of the model. The OCMO optimizes the learning process and enhances the model accuracy by dynamically adjusting its search strategy. Ultimately, comprehensive experimental analyses are conducted using several estimation parameters. The proposed method gains greater effectiveness and outperforms existing methods in drug repositioning. The developed method reaches an accuracy of 98.84% and a hit rate of 98.76%. These experimental findings ascertain the ability of NeuroInteract to find promising drugs for repurposing, furnishing a robust and more cost-effective model for SCZ treatment.