Postdoctoral Researcher in machine learning and NLP for clinical trial data 80-100 %

Department of Clinical Research
Employment upon agreement
The Faculty of Medicine at the University of Bern is an environment for high-quality, future-oriented research. Strong connections between basic research, engineering sciences, and university hospitals enable a unique setting for translational and patient-centered clinical research. The faculty prioritizes cross-disciplinary research and digitalization , fostering innovation in medical science. It is one of the largest medical faculties in Switzerland and is affiliated with the country's largest hospital complex.

The Department of Clinical Research (DCR) is a joint initiative of the University of Bern's Faculty of Medicine and its university hospitals, including Inselspital and the University Psychiatric Services (UPD). It supports and professionalizes clinical and translational research collaborations.
Our specialized divisions assist researchers throughout the entire research process, from project conception to result dissemination. We provide tailored educational programs and events on all aspects of clinical research, equipping researchers and students with the skills to conduct efficient and impactful studies. Our mission prioritizes patient-centered research, ensuring that patient perspectives are integral to our work.

The Medical Data Science group, led by Assistant Professor Benjamin Ineichen, a medical doctor with a PhD in neuroscience/pharmacology, is part of the DCR at the University of Bern. The group, known as the STRIDE-Lab , is a multidisciplinary team with expertise in medicine, neuroscience, statistics, and computer science. It focuses on bridging the gap between preclinical and clinical research and eventually drug approval, to advance therapy development for human diseases, with a focus on neuroscience. Using evidence synthesis and data science, the lab

Developing drugs for clinical applications is challenging, with only about 5% of therapies receiving regulatory approval ( Ineichen et al., PLoS Biology, 2024 ). While some failures are due to the complexity of innovative therapies, others stem from adjustable factors in drug testing, such as outcome measures, trial duration, and model selection (Berg et al., eBiomedicine, 2024). The impact of these factors is difficult to assess in individual trials but can be uncovered through large-scale clinical trial data analysis (Ineichen et al., Nature Reviews, 2024). Our approach combines expertise in medicine, evidence synthesis, and natural language processing (NLP) (Doneva et al., EMNLP, 2024) with Bern's extensive clinical trial landscape and modern data science infrastructure. The goal is to identify the key factors driving successful drug approvals and use this knowledge to optimize clinical trial design. Your work will contribute to:" target="_blank">Ineichen et al., PLoS Biology, 2024). While some failures are due to the complexity of innovative therapies, others stem from adjustable factors in drug testing, such as outcome measures, trial duration, and model selection ( Berg et al., eBiomedicine, 2024 ). The impact of these factors is difficult to assess in individual trials but can be uncovered through large-scale clinical trial data analysis ( Ineichen et al., Nature Reviews, 2024 ).
Our approach combines expertise in medicine, evidence synthesis, and natural language processing (NLP) ( Doneva et al., EMNLP, 2024 ) with Bern's extensive clinical trial landscape and modern data science infrastructure. The goal is to identify the key factors driving successful drug approvals and use this knowledge to optimize clinical trial design. Your work will contribute to:

1. Developing TrialSim, a digital platform that integrates multimodal deep learning to analyze large-scale clinical trial data from two sources:

• Clinical trial registries
• Electronic health record (EHR) data from trials in Bern and international sources. TrialSim should combine imaging and textual data while addressing modern privacy requirements in clinical research.

2. Understanding factors for successful approval by applying advanced statistical modeling and machine learning to analyze outcome measures, adverse events, and trial design, identifying what contributes to drug approval.

3. Simulating clinical trials using TrialSim to generate predictive analytics and scenario planning for drug development.
You will work at the interface of medicine and computer science, leveraging the large volume of clinical data available in Bern as well as from publications. Additionally, you will:

• Provide technical support for NLP and machine learning projects within the group.
• Collaborate with researchers in related fields, such as computer vision, benefiting from the University of Bern's strong digital research environment ( see digitalization strategy ).
• Contribute to publications and conferences.
• Foster a positive and collaborative team culture.

We are looking for candidates with a high enthusiasm for the projects we work on, including for drug development, clinical trials, health data, and statistical modelling, enjoying interdisciplinary work at the intersection of medicine and computer science.
Academic qualifications:

• PhD degree in computer science/informatics, machine learning, computational linguistics, statistics, software engineering, or a related field.

Technical qualifications:

• Proficiency in Python and/or R and common ML frameworks and tools, libraries, data structures, and data modeling.
• Hands-on experience developing Transformers-based NLP models using ML frameworks like Hugging Face and PyTorch using health-related data.
• Hands-on experience working with open-source large language models (LLMs) (e.g. BERT, LLaMA, Mistral) : prompt engineering, (parameter efficient) fine-tuning and multimodal model developing to incorporate tabular and/or image (and potentially other) data.
• Skilled in MLOps, including model deployment, scaling, and monitoring using tools like Docker and MLflow. Experienced in optimizing and serving large language models (LLMs) using techniques like model quantization and efficient inference.
• A proven publication record with at least one first author publication in peer-reviewed international journals.
• Experience with privacy-preserving techniques, including data anonymization and federated learning for training on sensitive health data.
• Experience with agentic LLM systems is a nice to have.

Additional skills for UI and Backend:

• Experience with front-end frameworks, such as React, Angular.js, and Vue.js
• Knowledge of multiple front-end technologies and libraries like HTML, CSS, and JavaScript
• Knowledge of back-end frameworks like Django and Flask, building APIs and integrating with databases such as PostgreSQL.
• Familiarity with common programming design patterns and best practices
• Experience with common web development tools and services, such as version control software, package managers, and CI/CD pipelines
• Experience developing desktop and mobile applications

Additional skills:

• Excellent organizational and supervisory skills, you can work in a team as well as independently.
• Strong ability to multi-task, to meet timelines, to work accurately; you are stress resistant.
• The ambition to write and publish at least one peer-reviewed publication per year

• Purposeful work aimed at improving animal welfare and advancing treatment for neurological (and other) diseases.
• A small multidisciplinary team with expertise in medicine, neuroscience, statistics, and computer science.
• Flexible working hours.
• Opportunities for first- and co-authorships on peer-reviewed scientific articles whenever possible.
• Access to a dynamic machine learning community at the University of Bern, with a strong emphasis on digitalization.
• Collaboration within Switzerland's largest medical faculty and hospital complex, offering extensive networking opportunities.

If you have any inquiries, please contact Prof. Ineichen Benjamin, at E-Mail schreiben.

Are you interested? Then please send us your complete application to HR Administration
(E-Mail schreiben) by (March 28th, 2025), at the latest.

Required application documents:

• CV, including publications
• Motivation letter explaining your interest in this particular project and environment
• Academic transcript/record of grades

Note: Only complete applications will be considered. We will invite promising candidates for an interview.