Can models faster reach clinical settings? An epidemiological perspective
Abstract: Advancements in data availability and computational capabilities have spurred the development of numerous in-silico models targeting medical applications. Many of these models are designed to aid in clinical decision-making and diagnostic processes. However, the journey to validate and integrate these models into real clinical settings is arduous and resource-intensive. Regulatory bodies, recognizing the novelty and diverse applications of these models, proceed cautiously, complicating the process of medical approval. Nevertheless, despite the challenges in direct implementation, alternative pathways exist for these models to contribute to improved clinical outcomes. This talk aims to explore how machine learning (ML) can be leveraged to enhance epidemiological studies and ultimately lead to better patient outcomes, with a particular focus on intervertebral disc degeneration and back pain.
Intervertebral disc degeneration presents a complex challenge, with a significant genetic component contributing to its etiology. Traditional genetic analysis has faced limitations, often due to insufficient statistical power. Here, ML-driven automated MRI phenotyping emerges as a promising solution, offering increased analytical power and insights into the genetic underpinnings of the condition. By translating these genetic frameworks into polygenic risk scores, we can improve existing outcome prediction tools, such as the STarTBack screening tool, thereby improving their predictive capabilities and facilitating more personalized medical decision-making. Notably, the enhanced predictability achieved through ML phenotyping serves as validation, affirming the accuracy of phenotype extraction and reinforcing the correlation between phenotype and outcome.
Biosketch: Roger Compte Boixader is the Early Stage Researcher 6 – PhD candidate, Disc4All Project. He studied a Biotechnology BSc and a Bioengineering MSc at Insitute Quimic de Sarrià (IQS) at Ramon Llull University in Barcelona. He undertook a MSc thesis at Massachusetts Institute of Technology (MIT) in Boston where he developed methodologies to study osteoarthritic knee cartilage function recovery from mechanical stimuli and the administration of medication. He is now currently finishing his PhD under the Disc4All Marie Curie ITN researching the genetic and molecular patterns of intervertebral disc degeneration and back pain.
