Background: Initiating tumors begin to develop years before cancer diagnosis. At this early stage, cancer can be asymptomatic; however, early cancer signals can be detected using serum RNA levels. For future cancer patients, the accurate detection of cancer signals could be life-saving, allowing for early and effective treatments to prevent cancer formation. In this project, you will have the opportunity to look for early signals of cancer, using a longitudinal machine learning technique. You will have access to serum RNA profiles from the Janus Serum Bank, https://www.kreftregisteret.no/en/Research/Janus-Serum-Bank/, a unique resource that contains human serum samples dating as far back as 1972, as well as cancer outcome data thanks to Norway’s centralized cancer registration system.
Goal: Develop diagnostic models to detect early signals of cancer based on serum RNA levels and machine learning models.
Opportunity for ML algorithm development: You will have the chance to improve our existing Python ML algorithm by adding new features and or optimizing the run-time with fast languages (Julia, C++, etc), and parallelization strategies.
Outcome:
-
Cancer prediction models that can detect cancers years in advance using a simple blood sample.
-
Hands on experience with machine learning and high-dimensional omics data.
-
Good potential to coauthor a scientific paper.
Required Background:
• Data science and or statistics
• Python Programming
Institutions: University of Oslo, Simula Research Laboratory, Cancer Registry of Norway
Supervisors: Gabriel Balaban (SRL), Sinan Umu (CRN), Molly Maleckar (SRL), Trine Rounge (UiO, CRN)