Predicting Anti-Inflammatory Small Molecules with Machine Learning
Author Information
Author(s): Selvam Subathra, Balaji Priya Dharshini, Sohn Honglae, Madhavan Thirumurthy
Primary Institution: SRM Institute of Science and Technology
Hypothesis
Can machine learning effectively predict anti-inflammatory small molecules?
Conclusion
The AISMPred method successfully predicts anti-inflammatory small molecules, achieving high accuracy and AUC values.
Supporting Evidence
- The ET model achieved the highest accuracy of 92% on the independent test dataset.
- A total of 15 ML models were developed using various feature sets.
- The study highlights the potential of small molecules in treating inflammatory diseases.
- Machine learning models were validated using a 10-fold cross-validation approach.
Takeaway
This study created a computer program that helps find new medicines to reduce inflammation by using small molecules.
Methodology
The study used a dataset of 4300 compounds, applying machine learning classifiers and feature selection techniques to predict anti-inflammatory small molecules.
Potential Biases
Potential bias may arise from the dataset's composition and the feature selection process.
Limitations
The study may be limited by the dataset's size and the specific features selected for model training.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website