Survey of Machine Learning Techniques in Drug Discovery

Area of Science:

• Computational Science
• Pharmaceutical Sciences
• Artificial Intelligence

Background:

• Drug discovery is a critical, yet expensive and time-consuming, pharmaceutical industry process.
• Machine learning (ML) and deep learning (DL) techniques are achieving state-of-the-art results across various scientific domains.
• AI, particularly ML, presents a significant opportunity to enhance the efficiency and effectiveness of drug discovery.

Purpose of the Study:

• To survey the current landscape of machine learning applications in drug discovery.
• To analyze trends and patterns in academic and industrial ML adoption for pharmaceutical research.
• To discuss the future potential of ML in advancing drug discovery pipelines.

Main Methods:

• Conducted a comprehensive literature search across scientific databases (e.g., ScienceDirect, Arxiv).
• Analyzed publications and information from startup companies in the ML and drug discovery sectors.
• Identified and categorized patterns related to ML techniques within the drug discovery domain.

Main Results:

• Distinct patterns were observed between general ML fields and specialized drug discovery applications.
• Keywords such as 'prediction,' 'brain,' 'discovery,' and 'treatment' are characteristic of drug discovery literature.
• There is a consistent year-over-year increase in the number of publications combining ML and drug discovery.

Conclusions:

• Machine learning techniques are increasingly integral to both academic and industrial drug discovery efforts.
• The survey identified key patterns indicating the growing synergy between AI and pharmaceutical research.
• Future applications of ML in drug discovery hold significant promise for innovation and efficiency.