Investigation of medication reviews and the identification of adverse drug reactions using machine learning algorithms

• Published in: Measurement. Sensors Vol. 33; p. 101240
• Main Authors: Mohanapriya, D., Chepur, Jyothi, Hirolikar, Deepali S., Sundareswari, K., Pokkuluri, Kiran Sree, Subbulakshmi, R.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.06.2024; Elsevier
• Subjects: Data mining; Drug detection; Drug suggestion; Lateral effects; Natural language processing; Text mining; Text mining; Drug detection; Natural language processing; Lateral effects; Data mining; Drug suggestion
• ISSN: 2665-9174; 2665-9174
• DOI: 10.1016/j.measen.2024.101240

The social networks for medicine allow users to share reviews on the drugs they have used. The medical and pharmaceutical industries can be guided by the arrangement of patient reviews of medications into ratings, whether good or unfavourable and the identification of the medications that cause an (Adverse Drug Reaction) ADR to better understand the patient's discomfort or satisfaction with particular medications. Unexpected pharmacological reactions from patient to patient can be caused by side effects. Pharmaceutical firms that produce the pharmaceuticals that are recommended for the prevention and treatment of sickness, however, occasionally cause negative lateral effects and lateral-effects that result in the demise of people who take the drugs. Pharmaceutical companies disclose the majority of side effects constructed on experimental tribunals, just one, which a small number are recognized. For better pharmacological side-effect prediction, feature sets are examined. When comparing Side-effects, machine learning (ML) techniques are utilized to determine which classifier has the highest level of accuracy. The goal of this study is to forecast how adenosine receptor-interacting medicines and therapeutic candidates would behave. In order to shape a machine learning prototypical, we used four distinct categorization methods: Naive Bayes(NB), Random Forest (RF), Decision Tree (DT), and Support Vector Machine (SVM). We then associated the findings and selected the one that produced the best results. In contrast to prior studies, we trained our model to distinguish between medications that interact with adenosine receptors and those that do not by using the drug side effect integrated into the drug fingerprint as a feature. In order to determine which medicine is most desired (and has the fewest side effects) among numerous, we rated the pharmaceuticals that interact with adenosine receptors depending on adverse effects. This aids in drug development. Drug side effects are typically not included in existing datasets, which focus instead on medicines, targets, and their interactions. For better pharmacological side-effect prediction, feature sets are examined. A comparative examination of Side-effects is conducted using machine learning (ML) techniques in order to determine which classifier has the highest level of accuracy. Using machine learning procedures and natural language processing techniques, this research proposes a model to forecast drug side effects based on review analyses.