Development of AI-Driven Advanced Lung Cancer Screening Models and Risk Factor Analysis
Principal Investigator
Name
IVAN MACIA
Degrees
Ph. D.
Institution
Fundación Centro de Tecnologías de Interacción Visual y Comunicaciones - Vicomtech
Position Title
Director of Digital Health and Biomedical Technologies, PI
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-1372
Initial CDAS Request Approval
Nov 13, 2023
Title
Development of AI-Driven Advanced Lung Cancer Screening Models and Risk Factor Analysis
Summary
Lung cancer stands as a significant global public health challenge, with an estimated 2.2 million new diagnoses and up to 1.8 million deaths reported worldwide in 2020. The age-standardized five-year overall survival rate is typically low, falling between 10% and 20% in most countries. Although screening programs have been established to detect lung cancer at its early stages, diagnosing patients in a timely and treatable manner remains a complex undertaking.
Our research project is driven by the hypothesis that a comprehensive analysis, harnessing clinical information and advanced AI models, including Machine Learning and Deep Learning, can reveal critical risk factors and represent the intricate web of their relationships within the realm of lung cancer for better understanding, allowing to derive personalized predictions of having risk of lung cancer in actionable timeframes. We believe that through acquiring a deep understanding of this complex landscape, we can create a path toward more effective risk-based screening programs and refine the accuracy of patient stratification into high and low-risk categories, as inferred by the models, ultimately resulting in improved outcomes for individuals at risk.
To achieve our research objectives, we proposed a 2 phased approach. In the initial phase, our focus centers on constructing various AI-based models, evaluating and comparing their predictive capabilities for lung cancer incidence across different time horizons, including 1 year, 3 years, and 5 years. In the following research phase, our objective is to assess and explain the relative importance risk factors for the most successful models, carefully studying their role in the development of lung cancer. We also aim to unravel novel or less known risk factors, for a better understanding of those leading to disease in different population groups.
Our research project is driven by the hypothesis that a comprehensive analysis, harnessing clinical information and advanced AI models, including Machine Learning and Deep Learning, can reveal critical risk factors and represent the intricate web of their relationships within the realm of lung cancer for better understanding, allowing to derive personalized predictions of having risk of lung cancer in actionable timeframes. We believe that through acquiring a deep understanding of this complex landscape, we can create a path toward more effective risk-based screening programs and refine the accuracy of patient stratification into high and low-risk categories, as inferred by the models, ultimately resulting in improved outcomes for individuals at risk.
To achieve our research objectives, we proposed a 2 phased approach. In the initial phase, our focus centers on constructing various AI-based models, evaluating and comparing their predictive capabilities for lung cancer incidence across different time horizons, including 1 year, 3 years, and 5 years. In the following research phase, our objective is to assess and explain the relative importance risk factors for the most successful models, carefully studying their role in the development of lung cancer. We also aim to unravel novel or less known risk factors, for a better understanding of those leading to disease in different population groups.
Aims
Leveraging state-of-the-art AI methodologies, our aims are two-folded:
- Develop novel AI-based lung cancer risk models that outperform current approaches in lung cancer prediction, and that may enable better personalized risk-based lung cancer screening strategies.
- Identify and gain insights into risk factors leading to the development of lung cancer according to the model predictions
Collaborators
Alba Garin - Senior Researcher (Vicomtech) agarin@vicomtech.org
Xabier Calle - Postdoctoral Researcher (Vicomtech) xcalle@vicomtech.org
Isabel Amaya - Predoctoral Researcher (Vicomtech) iamaya@vicomtech.org
Eduardo Alonso - Predoctoral Researcher (Vicomtech) ealonso@vicomtech.org