In the 21st century, people are living longer than ever before. Yet, while life expectancy has increased, our healthspan, the years lived in good health, has not kept pace. Many older adults spend their later years with age-associated diseases such as heart disease, diabetes, Alzheimer’s, and cancer. This reduces quality of life and places growing pressure on healthcare systems. In Norway, the elderly population is expected to double by 2050, posing major challenges for patient care.
This project explores how artificial intelligence (AI) can be used to prevent disease and improve outcomes for older patients. By combining multimodal data such as medical records, laboratory tests, lifestyle, genetics, etc., AI can reveal hidden patterns and provide more accurate predictions of individual health risks. To ensure our methods are interpretable and applicable, we use explainable AI (XAI), allowing clinicians to understand why a model makes a given prediction, as well as the risk factors driving it. Identifying risk factors is the first step for timing and personalized intervention, resulting in better health outcomes. Moreover, the transparency obtained using XAI is essential in healhtcare, where trust is fundamental.
In detail, a central focus is predicting post-surgical complications risk in elderly patients with colorectal cancer. Using XAI we aim to support doctors in making safer, more personalized surgical decisions that improve patient outcome. An important part of this work is the development of a clinically relevant biological age clock, a tool that measures biological age instead of chronological age. Biological age can be used as an early indicator of frailty and disease, guiding doctors adopting patient specific previentive strategies.
This research is carried out in close collaboration with clinicians, ensuring that the methods are tested, and applied in real healthcare settings. The goal is to close the gap between lifespan and healthspan, so that longer lives are also healthier ones.
This project aims to revolutionize the identification and treatment of AADs and ageing commorbidities by harnessing multi-omics datasets and AI to develop a sophisticated computational framework. The core objective is to integrate multilayered biomarkers with AI to gain a deep understanding of the biological mechanisms underlying AADs. This integrated approach is set to be incorporated into Agespan's software platform, enhancing the prediction of AAD risk, providing prognosis insights, and enabling personalized treatment plans tailored to each patient's unique biological profile.
The project has several key applications related to ageing and personalised medicine:
Firstly, it aims to construct an AI-driven framework for predicting AAD susceptibility by analyzing complex interactions between multi-omic and lifestyle factors, using extensive data from sources like the UK Biobank and Lifeline.
Secondly, it seeks to develop a surgery risk index using multi-omics data combined with lifestyle and clinical information from colorectal cancer patients, which will guide surgeons in assessing the benefits of prehabilitation and estimating postoperative risks on vulnerable and frail patients.
Thirdly, the project aims to improve the diagnosis and treatment of neuroendocrine carcinoma (NEC) by employing advanced ML techniques to classify NEC subtypes based on molecular phenotypes, ultimately informing treatment decisions and expanding therapy options.
Data sources include international biobanks (UK, Lifeline, Estonian), a NEC biobank at the University of Bergen, and clinical trial data from Bærum hospital.
