Unlocking the Power of Machine Learning in Biostatistics Research: Innovative Approaches to PK/PD Modelling and Drug Repositioning

The utilisation of machine learning has the potential to drive unprecedented breakthroughs in biostatistics research, according to Professor Jinfeng Xu, Associate Professor in the Department of Biostatistics at the City University of Hong Kong. With machine learning techniques becoming more sophisticated and accessible, Professor Xu believes there is a tremendous opportunity to leverage these tools in biostatistics research to unlock new insights and accelerate progress towards improving human health.

In a contractual research collaboration with Janssen Research & Development LLC, a US-based pharmaceutical company, Professor Xu spearheaded the development of analytical tools for investigating the impact of shrinkage on population pharmacokinetic (PK) and pharmacodynamic (PD) models. PK/PD models provide critical information on how drugs are absorbed, distributed, metabolised, and eliminated by the body, making them indispensable in pharmaceutical research. This information is vital for determining the optimal dosing regimen for a drug, predicting its efficacy and safety, and identifying potential drug interactions.

Through their research, Professor Xu and the team discovered the potential of machine learning approaches in improving the accuracy of PK and PD models. By utilising machine learning techniques, they could develop more accurate PK and PD models, which could help improve the development of effective and safe medications.

The research team, led by Professor Xu, is currently developing a new approach to drug repositioning that involves identifying new therapeutic uses for existing drugs. This approach is critical in the pharmaceutical industry because it offers a faster and more cost-effective approach to drug development than traditional methods, by utilising existing knowledge about the safety and efficacy of approved drugs.

The main objective of their research is to create a network of connections between drugs and diseases by examining data from different sources, including electronic health records, clinical trials, and published literature. Nevertheless, integrating data from multiple sources presents significant challenges. The team found that machine learning methods are well suited for addressing them. With its ability to process large and diverse data sets, machine learning algorithms can extract meaningful patterns and insights from a variety of data sources. This can help establish a comprehensive network of relationships between drugs and diseases, which is critical for identifying new therapeutic uses for existing drugs.

Machine learning methods can also handle the challenge of individualised heterogeneity by allowing for personalised treatment effects. By analysing patient-specific data, such as genetic information, medical history, and clinical symptoms, machine learning algorithms can identify subgroups of patients likely to benefit from specific drug treatments and tailor treatment plans accordingly. This personalised approach can improve treatment outcomes and reduce the risk of adverse drug reactions.

Professor Xu and his team are captivated by the remarkable results of PK/PD modelling and drug repositioning research using machine learning. They are eagerly looking forward to new avenues and breakthroughs that machine learning can provide to advance our understanding of the intricate relationships between health, disease, and the environment.

Professor Jinfeng Xu, Associate Professor in the Department of Biostatistics at the City University of Hong Kong.

 

Find out more about the Department of Biostatistics at CityU.