Dr Andreas Karwath

Supervisor Details

Contact Details

School of Medical Sciences, University of Birmingham

Scientific Background

Research Interests

I am an Associate Professor of Health Data Science at the University of Birmingham, UK with a background is in Computer Science, Machine Learning and AI. My research focuses on clinical AI for cross-sectional and longitudinal patient data, multi-modal data integration, cheminformatics, sub-phenotype discovery, and risk prediction. I leads the analytical work package in the Wellcome Leap ACORN study (https://www.acornstudy.co.uk), aiming to predict the risk of opioid use disorder and chronic pain after major surgery. I am a task leader in the EU project PINK in the AI-based generation of sustainable chemicals (https://pink-project.eu) and, as Deputy Director of the Solihull Innovation Hub, I aim to dive the adoption of analytics within NHS workflows.

Scientific Inspiration

My scientific inspirations are scientists who treat rigour as non-negotiable yet never lose sight of the essential question: will this help patients or improve clinical care or why are we doing this? I have been lucky to work with many such inspiring scientists, whose clarity of purpose steered complex methods towards practical solutions. They taught me to begin with the actual problem, design methods that are explainable and fair, and judge success by overall benefit rather than publication metrics. That ethos guides my work today, from how I frame research questions to how we evaluate and translate our models.

Supervision Style

In three words or phrases: supportive, organised, collaborative, guiding, inclusive, detail-oriented, outcome-oriented, technique-orientated.

Provision of Training

I will provide a clear vision for your PhD, initially focusing on core technical skills, and later on overall research outcomes and your future career. After an initial training period with closer supervision, you will take ownership and work increasingly independently. Your training will be supervised primarily by me, with support from an experienced post-doc or PhD student where appropriate.

Progression Monitoring and Management

"I like to be kept in the loop and updated and expect to see evidence of developments and results on a weekly basis. I will set out my expectations clearly beforehand (may it be for weekly goals, monthly meetings or other deadlines). If these are not met, we will always discuss a plan of action together. I have high expectations, and I expect self-motivation, curiosity, and sustained engagement with the project. If time allows, I am always available for advice and guidance.

Communication

Sometimes I do email my team/PhD students late in the day or on weekends. However I will not expect you to do the same. I am happy to discuss any issues that are impacting your ability to fulfil your potential or my/our expectations. Overall, I have an open-door policy, and you are always welcomed to speak/contact to me when I am available.

Meetings

PhD Students can expect scheduled meetings with me at least once per month. These meetings will be face-to-face. I am usually contactable for an instant response on every working day.

Work Patterns

The timing of work in my lab is completely flexible, and (other than attending pre-arranged meetings), I expect students to manage their own time.

Notice Period for Feedback

I need at least 1 month's notice to provide feedback on written work of up to 5,000 words.

MIBTP Project Details

Primary supervisor for:

AI for Rapid Infectious Disease Testing

See the PhD Opportunities section to see if this project is currently open for applications via MIBTP.

Please Note: The main page lists projects via BBSRC Research Theme(s) quoted and then relevant Topic(s).

AI for Rapid Infectious Disease Testing

Secondary Supervisor(s): Prof James Tucker

University of Registration: University of Birmingham

BBSRC Research Themes:

Renewable Resources and Clean Growth (Industrial Biotechnology)

Project Outline

The aim of this project is to develop AI methods that design highly specific, ultrafast nucleic-acid assays for point-of-care diagnostics. Building on the University of Birmingham’s work on isothermal exponential amplification reaction (EXPAR) [1,2], we will use modern transformer models to identify optimal recognition sequences within viral genomes that maximise speed while minimising false positives and cross-reactivity.

This research project focuses on better characterising the nucleic acid sequence and structural features that drive efficient amplification, and on translating those insights into robust in-silico design tools.

Methods include transformer-based sequence modelling, contrastive training, and structural causal models to prioritise causal motifs rather than mere correlates [3]. We will also apply cross-layer transcoders (CLT) to decompose model representations to guide candidate selection.

Workplan: curate datasets; develop and evaluate models; simulate assay performance under noise; deliver a ranked design pipeline. Promising designs may be evaluated with the University’s spin-out ExGenDx.

The project is entirely computational but with opportunities for collaboration with wet-lab partners working on EXPAR assay optimisation.

Candidate profile: strong coding skills (Python/ML); interest in transformer architectures and AI; background in molecular biology or chemistry advantageous.

Impact: The tools created will accelerate development of rapid, sensitive assays for future pandemics and point-of-care testing across infectious diseases and beyond.

[1] L. Cheng et al., Arch. Microbiol., (2023) 205:239, 10.1007/s00203-023-03579-9

[2] J. G. Carter et al., PNAS, 2021, 118, e2100347118, 10.1073/pnas.2100347118

[3] Mayank et al., CSBJ, (2025), 27, 10.1016/j.csbj.2025.03.021

APPLY HERE