PhD Studentship: Leveraging Precision Medicine to Predict an Adverse Drug Reaction

The tragedy of adverse drug reaction (ADRs) is synonymous with thalidomide yet in the context of mental health (MH) patients, intolerable side effects of psychiatric medication, leading to poor concordance, with the likelihood of further deterioration and possible involuntary treatment, is emerging as a very serious health issue. An estimated 970 million people worldwide have a lived experience of a MH condition, with depression (D) and anxiety (A) higher than psychosis (P), a figure which is increasing. Medication intended to alleviate mental distress is very often accompanied by ADRs which can be stigmatising, depressing and dangerous.

These ADRs have an impact on concordance and can undermine trust in the healthcare team. This, coupled with the risks posed by depression, anxiety, or psychosis (DAP), can make involuntary treatment more likely.  ADRs are not the sole reason for discontinuing a MH medication however ADRs of psychiatric medication are known to have significant negative impacts on physical health and quality of life, and further patient involvement is essential to understand the links between side effects and concordance more fully. ADRs are not always predictable from clinical trials. Long latency and novel effects especially in populations requiring MH treatment where patients may find it more difficult to raise concerns about medication with prescribers.

There is a need to study the experiences of risk populations (patients with a lived experience of DAP) using semi-automated machine learning natural language processing (NLP) to code this qualitative data in an unbiased way. The research project will curate and analyse evidence from suspected ADR reports, drug pharmacology and pharmacoepidemiology databases. ML-algorithms will be trained and applied to identify patterns and connections. The power of this research approach will guide regulatory and clinical decision-making for patient benefit and suggest alternative strategies that improve the patient’ experience – to avoid the most troubling ADRs experienced at a personalised level.

The researcher will screen known national and international databases for ADR reports and pharmacological interactions. Based on the unbiased NLP encoded ADRs we will deploy our ML-algorithm to integrate open sourced, fully anonymised, patient data that is available on ADRs (Yellow Card Scheme and others) and pharmacological data (ChEMBL and others). Other data sources are available e.g., WHO, FDA, and EudraVigilance registries to further expand upon the findings into global outcomes will enable the generalisation and scalability of the findings. A subset of anonymised patient health records, assay, imaging, and genome sequencing data will be overlaid from genomic datasets to enable pharmacoepidemiology links to be identified. There will be a focus on predictive ADR signal detection and confirmation of new signals identified.

For more information, please see the webpages below:

• Main MIBTP page - https://warwick.ac.uk/fac/cross_fac/mibtp/
• Birmingham MIBTP page - https://www.birmingham.ac.uk/research/activity/mibtp
• University of Birmingham Press release - Major investment in doctoral students to support next generation of researchers - University of Birmingham
• UKRI press release - : Over 4,700 newly funded post-graduate places in UK universities to create new generation of engineers and scientists - GOV.UK

Funding notes:

This studentship is funded by the MITBP.

A high 2:1 or 1st in a relevant discipline (chemical sciences, pharmacy, biomedical sciences, epidemiology, or computer science) is a requirement.

References:

Sandhu, D., Antolin, A.A., Cox, A. R., Jones, A.M. Identification of different side effects between PARP inhibitors and their polypharmacological multi-target rationale. Brit. J. Clin. Pharmacol. 2022, 88, 742-752. https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.15015.

Advert information