International Research Experience for Students

• Virtual Training Series: April 13-May 29, 2026
  • Through six (6) 90-minute weekly Zoom sessions, students prepare for their summer research experience. These Zoom sessions consist of lectures/activities centered around: program logistics and expectations, conducting scientific research, reading the scientific literature, living in Sweden, Swedish culture (inside and outside the lab), AI, data science, and algorithm bias.
• Travel to Stockholm & Welcome Weekend: June 4-7, 2026
  • Students travel from their home airport to Stockholm. After arriving in Sweden, the program director hosts a welcome weekend in Stockholm to build comradery within the cohort and introduce students to the city. The welcome weekend consists of ice breaker activities, a city tour, shared meals, and a tour of the Nobel Prize Museum. The program director hosts an orientation session at KI on the Monday following arrival in Sweden. During this orientation, the PI gives students a tour of KI and SciLifeLab.
• Research Program: June 8- August 14, 2026
  • During their time in Stockholm, IRES students are embedded members in one of seven host research groups. Students work on research projects under the mentorship of professors, postdocs, and/or graduate students in each host lab. Example project descriptions for each of the research groups can be found below in the ‘Current Program Projects’ section. During the summer, students also participate in a professional development workshop series run by the program director to provide students with the skills necessary to succeed in their summer research project and beyond. Workshop activities include: scientific communication, research methods and why/how to apply to graduate school
• Biomedical Engineering Society Meeting (optional): October 21-24, 2026 in Orlando, Florida.
  • Following the summer program, students have the option to present their work at the BMES annual meeting in October. This is an opportunity for students to network with researchers and speak with potential graduate school mentors. Additionally, many graduate programs host receptions during the conference directed at recruiting undergraduate students to their programs. IRES students attending the conference can speak with faculty members at these receptions to learn more about what each program has to offer.

• $7000 stipend.
• Free round-trip airfare to Stockholm, Sweden.
• Free housing at KI Residence Solna during the summer.
• Professional development workshops.
• Tour of the European Centre for Disease Prevention and Control in Stockholm.
• Live in Stockholm for the summer with the exciting opportunity to travel to various European destinations during your weekends off!
• An opportunity to present at the annual Biomedical Engineering Society (BMES) meeting (October 2026; Orlando, Florida). Costs covered: conference registration, travel and accommodation.

• Full-time undergraduate student at a primarily undergraduate institution (PUI) in the United States. Note: A PUI is an accredited institution that focuses on undergraduate education and awards associate's, bachelor's, and/or master's degrees. The National Science Foundation (NSF) defines a PUI as an institution that awards <20 PhDs in NSF-supported fields over two academic years. Examples of PUIs: University of San Diego, Harvey Mudd College, Olin College, Bucknell University, Cooper Union, Cal State LA. Examples of non-PUIs: All universities in the University of California System, all Ivy-league institutions, most public state Universities. If you are unsure about your eligibility, please contact the program director (contact information below).
• Expected Bachelor's graduation date in Fall 2026 or later (i.e. seniors graduating in Spring 2026 are ineligible).
• U.S. citizen or permanent resident.
• Pursuing a degree in biomedical engineering, bioengineering, computer science, bioinformatics, computational biology, or related fields.
• Applicants must be available during all program dates.
• Participants must be 18 years of age by June 2026.

• Major components of the application include:
  • Personal and academic information.
  • Personal statement discussing the student’s personal and professional motivations for participation in the program, interest in AI/data science, perspectives on international collaboration/travel, future goals, how participation in this program would help the student achieve their future goals, and research faculty project/mentor preferences (if any).
  • Statement of preparedness detailing how the student’s academic/personal background and previous research experiences (not required) have prepared them for the program.
  • GPA statement describing whether or not the student thinks their GPA is an accurate reflection of their academic abilities. Although GPA is somewhat indicative of aptitude/work ethic, it will not be a main criterion for acceptance.
  • Unofficial transcript.
  • Letter of recommendation.
• Only complete applications will be reviewed.
• APPLY NOW

Research Advisor: Magnus Boman, KI
Self-learning algorithms for interpreting organ-agnostic CT scans: solid tumor radiomics for precision medicine.

Objective: To efficiently find lung cancer metastases using CT scans of the abdomen with self-correcting deep learning algorithms.

Project Summary: Since there is a shortage of radiologists, automating some tasks with AI, such as the measurement of tumor burden, is among the most promising opportunities for time-saving. For tasks like segmentation of organs, self-learning and self-correcting approaches like U-net have been shown to emulate, and in some cases supersede, human efforts. In collaboration with attending radiologist Dr. Vitali Grozman, Dr. Boman has automated adrenal gland measurements using a neural network U-net approach. Usefulness of this new technique is prominent: metastases are sometimes missed by radiologists, growth can be slow, and many adrenal glands have a complex shape and take time to evaluate. The goal with the adrenal study was to efficiently find metastases after lung cancer, and the next steps are to examine other structures commonly affected by metastases including the liver, kidneys, and skeleton. This project centers around developing computational methods to achieve this by using imaging data from the cutting- edge technology of photon-counting CT.

Research Advisor: Carsten Daub, KI and SciLifeLab
Computational multi-modal data integration for the characterization and risk assessment of breast cancer.

Objective: Developing a multi-modal data analysis method to improve breast cancer characterization.

Project Summary: Breast cancer characterization and corresponding risk assessment is routinely performed on stained breast tissue sections together with expert pathologist image annotations and clinical characteristics. Spatial Transcriptomics (ST) is a technology developed in Sweden measuring gene expression by sequencing thousands of small areas in a tissue section providing spatially-resolved gene expression data. This technology shows great potential as a tool for pathologists to assist with the clinical evaluation of cancer samples, however sequencing depth issues with ST exist. To solve these depth issues, the integration of ST data with single cell (SC) sequencing data within individual datasets has allowed for the mapping of SC clusters onto tissue structures. Accordingly, this project will build on this work by developing a multi-modal breast cancer data integration model of in-house and publicly available ST image and sequencing data, SC sequencing data, and corresponding image/clinical annotations. The overall aim of this project is to assess how each of the considered modalities contributes to explaining breast cancer sub-type and cancer onset relative to the distance to the tumor.

Research Advisor: Simon Elsässer, KI and SciLifeLab
Multiplexed quantitative ChIP-seq for the profiling of epigenetic signatures in circulating nucleosomes in human plasma.

>Objective: Evaluating diagnostic potential of circulating nucleosomes with machine-learning approaches.

Project Summary: Epigenetic alterations are associated with, and likely modulate, many common diseases. New approaches are needed to survey systemic epigenetic alterations associated with disease. Cell-free/circulating nucleosomes (cfNucs) in the bloodstream, derived from cells in the body, provide an accessible biomarker type that can be used for early detection of disease. Intriguingly, it has been found that pathological processes can change the sequence content and post-translational modification (PTM) patterns of cfNucs. Collectively, these modifications define functional states that regulate how the cell uses the underlying DNA. Based on a multiplexed quantitative ChIP-seq approach extensively validated on cellular chromatin in Dr. Elsässer’s prior work, this project aims to develop a multiplexed platform for profiling epigenetic signatures on cfNucs in human blood plasma for novel cancer biomarker detection.

Research Advisor: Juliette Griffie, SciLifeLab and Stockholm University
Developing machine learning-based image analysis tools to investigate autoimmunity with advanced light microscopy.

Objective: Develop an algorithm dedicated to high content image processing in order to extract fundamental information about T cell activation in the context of autoimmunity.

Project Summary: Advanced light microscopy encompasses a number of imaging approaches that produce high content datasets. These include super-resolution techniques which aim to circumvent the diffraction limit, high content screening, and lattice light sheet microscopy which generates live cell movies. While these techniques have the potential to investigate key cellular processes and diseases at the molecular level, they require the development of dedicated analytics to extract information from the complex images generated. In this vein, this project focuses on developing and applying machine learning (ML) techniques to single molecule super-resolution microscopy data to map protein aggregates in T cells in order to better understand autoimmune diseases. Specifically, this work builds off of a previous study which mechanistically showed the importance of PTPN22, a protein implicated in many autoimmune disorders, in late T cell activation. The current study will examine the role of PTPN22 throughout T cell activation by probing its submicron spatial organization in relation to other key protein markers. This work will provide novel insight into cell fate and, more broadly, pave the way towards patient disease phenotyping in a fully data/image driven manner.

Research Advisor: Adil Mardinoglu, SciLifeLab and KTH – Royal Institute of Technology
Comprehensive multi-omics analysis to identify biomarkers in health and disease.

Objective: This project aims to integrate advanced multi-omic data to identify distinctive biomarkers that differentiate healthy individuals from those at risk of chronic diseases.

Project Summary: Chronic diseases are a rising global health concern. Early identification through biomarkers can be pivotal in disease management and prevention. This project aims to delve deep into this concept by harnessing the power of multi- omics. In this study, 1000 adult volunteers from a health clinic in Turkey (part of an existing collaboration) will be profiled, excluding those with diagnosed serious chronic diseases. Blood, urine, saliva and fecal matter samples from these volunteers have been collected and will be analyzed as part of the study. Blood samples have been subjected to genomic, transcriptomic, proteomic, and metabolomic analyses, urine samples to metabolomic and proteomic analyses, and saliva/fecal samples have been subjected to metagenomic microbiome analyses. This dataset will be compared with available datasets from individuals with chronic diseases, such as non-alcoholic fatty liver disease and type 2 diabetes, using systems biology tools (i.e. co-expression networks, transcriptional regulatory networks, protein-protein interaction networks, genome-scale/metabolic networks) to identify potential biomarkers.

Research Advisor: Fredrik Strand, KI
Multi-modal survival analysis for breast cancer by Pathology Language and Image Pre-Training fine- tuning.

Objective: The objective of this project is to improve breast cancer prognosis using a multi-modal AI approach to enhance the accuracy and reliability of breast cancer survival predictions.

Project Summary: This project proposes a comprehensive approach to breast cancer prognosis by integrating multi-modal AI techniques, specifically Pathology Language and Image Pre-Training (PLIP) fine-tuning and self-supervised learning. This integration aims to exploit the synergies between these techniques for improved model performance, interpretability, and robustness. This study involves the integration of clinical data, pathology parameters, and radiology images into a unified dataset. Subsequently, PLIP will be fine-tuned on this integrated dataset using transfer learning techniques tailored for breast cancer survival analysis. Self-supervised learning methods, such as contrastive learning, may be employed to enhance the model's understanding of latent representations in the data, leveraging both labelled and unlabeled information. The project may also include investigating multi-modal fusion techniques to effectively integrate features from different modalities. Interpretability methods will be implemented to visualize the contributions of both PLIP and self-supervised learning representations, providing insights into the factors influencing survival predictions. The performance of this algorithm compared with baseline parametric approaches will be evaluated on a validation set during development and on a withheld test set after model freeze. The anticipated outcomes include improved prognosis accuracy, enhanced robustness, increased interpretability, and, possibly, improved generalization to diverse patient populations and datasets.

Research Advisor: Carl Johan Sundberg, KI
Development of an in-silico model for the identification and examination of the preventive medicine capabilities of physical activity.

Objective: The objective of this project is to integrate multi-modal data collected from humans and develop an in-silico model of exercise to understand physical activity’s role in preventing non-communicable diseases.

Project Summary: Individuals who maintain an exercise program have a decreased chance of developing several chronic diseases, such as type 2 diabetes, cardiovascular disease, and some types of cancers. Given that exercise has been shown to influence gene expression, protein abundance, posttranslational modifications and epigenetic markers across the body, the ability to fully elucidate the link between the molecular changes that accompany exercise and its preventive medicine capabilities is difficult. As such, this project aims to consolidate multi-modal data collected in Dr. Sundberg’s lab and from publicly available datasets into a computational model to identify the biological mechanisms behind exercise’s health benefits. In this study, muscle, adipose and blood samples from 48 trained and untrained volunteers have been collected and have been profiled to varying degrees. Muscle has been subjected to transcriptomic, proteomic and metabolomic analyses, adipose samples to transcriptomic analysis, blood samples to proteomic analysis, and body composition data has been collected on a subset of volunteers. In collaboration with Dr. Mardionoglu’s group, this multi-modal dataset will be compared with available datasets from individuals with lifestyle-related diseases, such as type 2 diabetes, using systems biology tools (i.e. co-expression networks, transcriptional regulatory networks, protein- protein interaction networks, and genome-scale/metabolic networks) to identify key drivers promoting metabolic health.