Ali Heydari
Research Areas
Authored Publications
Sort By
Preview abstract
Blood biomarkers are an essential tool for healthcare providers to diagnose, monitor, and treat a wide range of medical conditions. Establishing personalized blood biomarker ranges is crucial for accurate dis-ease diagnosis and management. Current clinical ranges often rely on population-level statistics, which may not adequately account for the substantial influence of inter-individual variability driven by factors such as lifestyle and genetics. In this work, we introduce a novel framework for predicting future blood biomarker values and personalized reference ranges through learned representations from lifestyle data (physical activity and sleep) and blood biomarkers. Our proposed method learns a similarity-based embedding space that aims to capture the complex relationship between biomarkers and lifestyle factors. UsingUK Biobank (257K participants), our results show that our deep-learned embeddings outperform traditional and cur-rent state-of-the-art representation learning techniques in predicting clinical diagnosis. Using a subset of UK Biobank of 6440 participants who have follow up visits, we validate that the inclusion of these embeddings and lifestyle factors directly in blood biomarker models improves the prediction of future lab values from a single lab visit. This personalized modeling approach provides a foundation for developing more accurate risk stratification tools and tailoring preventative care strategies. In clinical settings, this translates to the potential for earlier disease detection, more timely interventions, and ultimately, a shift towards personalized healthcare.
View details
Preview abstract
Blood tests are an essential tool for healthcare providers to diagnose, monitor, and treat a wide range of medical conditions; however, quantitative approaches for personalizing such metrics are nascent and often ignore important factors such as lifestyle. Moreover, recent studies have shown that raw (untransformed) representations of health records are inadequate for constructing predictive models, especially when considering a single timepoint. In this work, we investigate the association of activity and sleep with blood test ranges, and based on our results, propose Proteus, a new deep metric learning algorithm that accounts for lifestyle. We show that Proteus significantly improves the performance of several downstream analyses, including the prediction of future health risk in currently-healthy patients using a single laboratory visit. Building upon our findings, we additionally introduce DeepRange, a novel lifestyle-informed algorithm which utilizes deep-learned embeddings for estimating personalized optimal blood test ranges. Our proposed methodology for personalized blood test ranges and single-visit health risk prediction can be readily implemented and has the potential to significantly improve health outcomes by enabling early intervention and personalized treatment.
View details
Preview abstract
We propose a novel formulation of the triplet objective function that improves metric learning without additional sample mining or overhead costs. Our approach aims to explicitly regularize the distance between the positive and negative samples in a triplet with respect to the anchor-negative distance. As an initial validation, we show that our method (called No Pairs Left Behind [NPLB]) improves upon the traditional and current state-of-the-art triplet objective formulations on standard benchmark datasets. To show the effectiveness and potentials of NPLB on real-world complex data, we evaluate our approach on a large-scale healthcare dataset (UK Biobank), demonstrating that the embeddings learned by our model significantly outperform all other current representations on tested downstream tasks. Additionally, we provide a new model-agnostic single-time health risk definition that, when used in tandem with the learned representations, achieves the most accurate prediction of subjects' future health complications. Our results indicate that NPLB is a simple, yet effective framework for improving existing deep metric learning models, showcasing the potential implications of metric learning in more complex applications, especially in the biological and healthcare domains.
View details
