Machine Learning and Quantitative System Pharmacology Strategies for the Prediction of Drug-Drug Interactions

Project Description
Drug-drug interactions (DDIs) can jeopardise the clinical management of multiple therapies and complicate the prescription of treatments for patients. Adverse drug reactions (ADR) due to DDIs can represent up to more than 20% of all reported ADRs. DDIs can be defined as the modulation of the pharmacologic activity of one drug by the prior or concomitant administration of another drug. The frequency of potential DDIs can vary between different group of patients and disease areas with potential severe impact on the effectiveness, toxicity and overall management of therapies. Due to the high number of possible drug combinations comparatively few DDI clinical studies are conducted, therefore there is no guidance on how to handle many DDIs and the identification of risk related DDIs is challenging. Advanced computational modelling can support the integration of multiple types of data (e.g. experimental and clinical) for the generation of predictive algorithms and recently a broad variety of methods based on machine learning and quantitative system pharmacology (QSP) have emerged. We represent one of the leading research group on DDIs, and we have developed award-winning, world-leading DDI resources for multiple disease areas to support a rationale risk assessment during the management of complex therapies.

Objectives
- To predict the risk related to DDI across key disease areas through machine learning strategies
- To refine the predictions generated using machine learning through a mechanistic QSP framework for the identification of drug characteristics influencing DDI magnitude in different groups of patients.

Novelty
The project would capitalise on unique resources and methodologies available across the two collaborative partners. We have one of the largest database on DDIs with around 60000 drug pairs and a rational classification of pharmacological and clinical information, constituting an unique platform to support the development of novel machine learning and QSP strategies. The integrated approach would represent an impactful strategy to rationalise the risk related to DDIs for drug combination that have not been tested clinically as well as future therapeutic agents to streamline drug development and the regulatory evaluation.

Timeliness
DDIs represent a public health priority considering the aging population and growing proportion of elderly individuals experiencing age-related comorbidities. Our integration of mechanism-based modelling and machine learning methodologies is an innovative strategy to predict patient risk related to DDIs

Experimental Approach
Machine learning is an effective approach to analyse large-scale heterogenous data to predict efficacy of drug combinations and improve drug design. QSP modelling supports a mechanistic understanding of drug distribution and actions allowing the prediction of clinical scenarios in specific subpopulation of patients.

To apply please send CV, a letter of motivation and the names of two referees who can send letters of recommendation to Dr Marco Siccardi ([Email Address Removed])

For application enquires please contact Dr Marco Siccardi ([Email Address Removed])