Doctoral Student in Equitable AI for Breast Cancer Screening and Risk Prediction – Karolinska Institutet, Solna

Job Description

Do you want to contribute to top quality medical research?
To be a doctoral student means to devote oneself to a research project under supervision of experienced researchers and following an individual study plan. For a doctoral degree, the equivalent of four years of full-time doctoral education is required.
The research group
The doctoral student will be based at the Department of Oncology-Pathology, Karolinska Institutet, within the Computational Breast Imaging Group led by Associate Professor Fredrik Strand. The group is internationally recognized for the development and evaluation of AI for breast cancer imaging and precision medicine, combining expertise in radiology, machine learning, and biostatistics. It comprises PhD students, research engineers, research assistants, postdoctoral researchers, and collaborating senior clinicians in radiology, pathology, oncology, and surgery. The group has active national collaborations (KTH, Lund University, the Institute for Future Studies, and several groups at KI) and international collaborations (among others UC Berkeley, UCSF, University of Hawaii, and University of Barcelona). Available infrastructure includes high-performance computing, GDPR-compliant secure storage, and the VAI.B platform, which holds multi-region Swedish screening mammography data linked to cancer outcomes. The doctoral student will be supervised by Apostolia Tsirikoglou, PhD, research specialist and lead AI scientist in the group, as main supervisor, together with Associate Professor Fredrik Strand, head of the group and principal investigator of the project, as co-supervisor.
The doctoral student project and the duties of the doctoral student
This PhD project investigates whether AI systems used in population-based mammography screening perform equitably across population groups, and how inequities can be reduced. AI is entering breast cancer screening both as detection support and as risk prediction to guide screening intervals and supplemental imaging. Systems developed on specific populations may perform differently for groups that were underrepresented in their development, and screening outcomes already differ by country of birth and socioeconomic status. Evidence on equitable performance is therefore needed before AI is adopted at scale.
The project uses a large multi-region Swedish screening dataset with radiologist assessments, cancer outcomes from the National Quality Register for Breast Cancer, and demographic and socioeconomic data from Statistics Sweden. The doctoral student will evaluate commercial and in-house AI cancer detection systems and AI risk prediction models across subgroups defined by country of birth and socioeconomic status, compare AI with radiologist performance in the same groups, and, in a later phase, develop and evaluate fairness-aware models that reduce identified disparities while maintaining clinical performance.
The doctoral student is expected to take an active and increasingly independent role in study design, cohort construction and register linkage, statistical analysis, running and evaluating AI systems on imaging data, model development, writing scientific publications, and presenting results at international conferences. The doctoral education includes courses in epidemiology, biostatistics, machine learning, and AI fairness. Where needed, competence in deep learning model development will be built during the project under supervision.

Qualifications

• General eligibility: a second-cycle/advanced/master qualification (master's degree), or courses comprising at least 240 credits of which at least 60 credits at advanced/master level, or substantially equivalent knowledge acquired in Sweden or abroad.
• Specific eligibility: proficiency in English equivalent to the course English B/English 6 at Swedish upper secondary school.
• A master's degree (or equivalent) in bioinformatics, biostatistics, or a related quantitative field
• Programming skills for data processing and analysis
• Knowledge of statistical and epidemiological methods
• Strong analytical ability and interest in interdisciplinary research at the intersection of AI, medicine, and health equity
• Ability to work both independently and collaboratively in a multidisciplinary team
• Excellent written and spoken English

Skills Required

• Desirable/advantageous:
• Experience with population-based register data and record linkage, preferably Swedish national registers
• Experience with cohort construction and subgroup analyses in large health datasets
• Experience with large-scale processing of medical imaging data or evaluation of AI systems on clinical data
• Interest in health equity, fairness in AI, and responsible implementation of AI in healthcare

Additional Information

Always verify vacancy details and availability on the employer’s official website. Jobs Bio Help Learning is an independent job-discovery platform and does not process applications.