PhD on AI-driven multimodal data fusion models in rare diseases

Project

The candidate will develop and benchmark zero-shot multimodal fusion models for rare disease prediction, using melanoma as a use case. The project integrates spatial and single-cell multi-omics data with histopathology and clinical information to learn robust cross-modal representations for diagnostic prediction. It pursues two integrated objectives: (i) to develop generative and explainable AI approaches for multimodal data fusion; and (ii) to extend existing multimodal frameworks with additional omics layers and evaluate their performance across heterogeneous datasets. At KU Leuven, the candidate will develop machine-learning strategies for multimodal representation learning and for the integration of the complex biological datasets generated within the consortium. 

Duration and start date: 48 months, full-time. The first 36 months are funded by the MSCA Doctoral Network SPACE-MEL; the appointment is extended to a full four-year doctoral trajectory at KU Leuven. Intended start date: begin 2027.

Profile

Eligibility conditions:

• Master’s degree in Bioinformatics, Computer Science, Artificial Intelligence, Bioscience Engineering or a related field, obtained by the start date of the contract.

• Applicants must be doctoral candidates, i.e. not already in possession of a doctoral degree, and must be admissible to the KU Leuven Doctoral School of Biomedical Sciences.

• Mobility rule: applicants must not have resided or carried out their main activity (work, studies) in Belgium for more than 12 months in the 36 months immediately before the recruitment date.

Required skills: 

• Strong programming skills in Python, and experience with at least one deep-learning framework (PyTorch preferred).

• A solid grounding in machine learning. Experience with representation learning, generative models, foundation models or multimodal integration is a strong asset.

• Experience in the analysis of high-dimensional biological data (single-cell or spatial transcriptomics, digital pathology or biomedical imaging) is an advantage.

• Comfortable working on Linux and HPC/GPU systems, with version control (git) and reproducible workflows (conda or containers, Snakemake or Nextflow).

• Able to work independently as well as within an interdisciplinary, international consortium.

• Proficiency in the English language is required, as well as good communication skills, both oral and written. Successful candidates will need to provide an English test (e.g. IELTS, TOEFL, Cambridge English). You may be exempt if you are a national of a majority native-English speaking country, or have qualifications / degree that has been taught and assessed in English. The supervisor may also confirm that a candidate has the required level of English.

Secondments: The project is carried out in close collaboration with the groups listed below, and visits to their laboratories are part of the training programme. These secondments take place within the contract period; a willingness to travel and to spend extended periods abroad is therefore essential.

• 4 months at University of Manchester, United Kingdom (supervisor: Dr. Martin Fergie)

• 3 months at Spotlight Pathology Ltd., United Kingdom (supervisor: Dr. Samantha Perona)

Offer

The doctoral candidate will receive a competitive salary in accordance with the MSCA Doctoral Networks program, comprising a living allowance and a mobility allowance. A family allowance is foreseen where applicable. 

KU Leuven additionally offers holiday pay, hospitalization insurance, reimbursement of certain commuting costs and access to its sports and childcare facilities.

Interested?

You can apply for this job no later than October 15, 2026 via the online application tool