CORDIS - EU research results

Cancer Long Survivors Artificial Intelligence Follow Up

Periodic Reporting for period 3 - CLARIFY (Cancer Long Survivors Artificial Intelligence Follow Up)

Reporting period: 2022-07-01 to 2023-06-30

It is estimated that more than 50% of adult patients diagnosed with cancer live at least 5 years in the US and Europe. This situation leads to a new challenge: to increase the cancer patients’ post-treatment quality of life. This project aims at identifying cancer survivors from 3 prevalent types of cancer: breast, lung and lymphomas with ongoing health and supportive care needs. We were able to determine the factors that predict poor health status after oncological treatments. For this aim, Big Data and artificial intelligence (AI) techniques were used to integrate in a digital platform that provides real-time analyses, CLARIFY Platform, all available patient´s information from different data sources that allows characterization of patient’s cohorts, survival, management and treatment outcomes, enabling physicians to make evidence-based decisions. Predictive relapse or toxicity-risk models based on statistical relational learning and explainable AI techniques, on top of the integrated knowledge graphs, were developed. With this tool not only do we help patients to improve their quality of life and global health status, but also the cost effectiveness of patient´s management and follow-up in the current overburdened healthcare systems.
The goal of the CLARIFY project is to transform cancer survivors’ care by providing enhanced, personalized treatment options. The current follow-up models have not progressed at the same pace as the advancements in cancer treatment effectiveness. Cancer survivors face various unmet physical, functional, and psychosocial needs that impact their overall quality of life and survival outcomes. All of these factors have an impact on the disease process. Therefore, it is necessary to take into account not only clinical data, but also quality of life, and activity data when the goal is improving the patient’s well-being and the disease outcome. CLARIFY Digital Decision Support Platform allows early identification and discovery of risk factors that may deteriorate a cancer survivor’s condition after the end of treatment. It stratifies survivors based on risk allowing clinicians to design personalized follow up and supportive care protocols. This is achieved through collecting and analyzing anonymized clinical and genomic data, data from wearable devices (circadian rhythm, physical activity), and electronic quality-of-life surveys. This information is integrated with existing biomedical knowledge from public repositories and biomedical literature. To analyze the vast and diverse data, CLARIFY employs state-of-the art predictive models and biomedical knowledge graphs.
We generated a comprehensive database, encompassing clinical, demographic, behavioral, and molecular data, enabling the identification of specific patterns. Noteworthy achievements include circadian rhythm analyses, creation of an annotated Spanish corpus. Especifically, we produced machine learning models predicting relapse in early-stage non small cell lung cancer patients with automatic explanations, fostering personalized and reproducible predictions. CLARIFY Platform was fully operational enhancing medical decisions during and after treatment, thus enabling clinicians to detect deteriorating health factors and implement personalized interventions through wearable devices and quality of life questionnaires. Therefore, CLARIFY Platform emerges as a groundbreaking decision support system, meeting clinicians' needs and enhancing healthcare accuracy, efficiency, and accessibility for cancer patients. Through personalized follow-ups based on risk factors, the Platform significantly contributes to improving the quality of life and well-being of long-term cancer survivors. These activities aligned at all times the Artificial Intelligence component with ethical standards, contributing to EU policy initiatives. The knowledge generated produced 35 publications, 20 scientific articles and 15 international presentations, and facilitated academic training.
High level architecture of the genomic data imputation pipeline
Conceptual Framework and Overall Pipeline CLARIFY predictive models
High level architecture of the core suite of explainable predictive models
CLARiFY Platform dashboard
CLARIFY pipeline
Cancer use case description
Knowledge Graph Creation Pipeline in CLARIFY