arXiv subjects

Svetomir N. Markovic

Publications and source records attributed to Svetomir N. Markovic.

5 recordsLinked to original sources

arXiv2025

Towards Spatially-Lucid AI Classification in Non-Euclidean Space: An Application for MxIF Oncology Data

Given multi-category point sets from different place-types, our goal is to develop a spatially-lucid classifier that can distinguish between two classes based on the arrangements of their points. This problem is important for many applications, such as oncology, for analyzing immune-tumor relationships and designing new immunotherapies. It is challenging due to spatial variability and interpretability needs. Previously proposed techniques require dense training data or have limited ability to handle significant spatial variability within a single place-type. Most importantly, these deep neural network (DNN) approaches are not designed to work in non-Euclidean space, particularly point sets. Existing non-Euclidean DNN methods are limited to one-size-fits-all approaches. We explore a spatial ensemble framework that explicitly uses different training strategies, including weighted-distance learning rate and spatial domain adaptation, on various place-types for spatially-lucid classification. Experimental results on real-world datasets (e.g., MxIF oncology data) show that the proposed framework provides higher prediction accuracy than baseline methods.

eess.IV↗

arXiv2025

Spatially-Delineated Domain-Adapted AI Classification: An Application for Oncology Data

Given multi-type point maps from different place-types (e.g., tumor regions), our objective is to develop a classifier trained on the source place-type to accurately distinguish between two classes of the target place-type based on their point arrangements. This problem is societally important for many applications, such as generating clinical hypotheses for designing new immunotherapies for cancer treatment. The challenge lies in the spatial variability, the inherent heterogeneity and variation observed in spatial properties or arrangements across different locations (i.e., place-types). Previous techniques focus on self-supervised tasks to learn domain-invariant features and mitigate domain differences; however, they often neglect the underlying spatial arrangements among data points, leading to significant discrepancies across different place-types. We explore a novel multi-task self-learning framework that targets spatial arrangements, such as spatial mix-up masking and spatial contrastive predictive coding, for spatially-delineated domain-adapted AI classification. Experimental results on real-world datasets (e.g., oncology data) show that the proposed framework provides higher prediction accuracy than baseline methods.

cs.LG↗

arXiv2024

Spatial Computing Opportunities in Biomedical Decision Support: The Atlas-EHR Vision

We consider the problem of reducing the time needed by healthcare professionals to understand patient medical history via the next generation of biomedical decision support. This problem is societally important because it has the potential to improve healthcare quality and patient outcomes. However, navigating electronic health records is challenging due to the high patient-doctor ratios, potentially long medical histories, the urgency of treatment for some medical conditions, and patient variability. The current electronic health record systems provides only a longitudinal view of patient medical history, which is time-consuming to browse, and doctors often need to engage nurses, residents, and others for initial analysis. To overcome this limitation, we envision an alternative spatial representation of patients' histories (e.g., electronic health records (EHRs)) and other biomedical data in the form of Atlas-EHR. Just like Google Maps allows a global, national, regional, and local view, the Atlas-EHR may start with an overview of the patient's anatomy and history before drilling down to spatially anatomical sub-systems, their individual components, or sub-components. Atlas-EHR presents a compelling opportunity for spatial computing since healthcare is almost a fifth of the US economy. However, the traditional spatial computing designed for geographic use cases (e.g., navigation, land-surveys, mapping) faces many hurdles in the biomedical domain. This paper presents a number of open research questions under this theme in five broad areas of spatial computing.

cs.CY↗

arXiv2022

SAMCNet for Spatial-configuration-based Classification: A Summary of Results

The goal of spatial-configuration-based classification is to build a classifier to distinguish two classes (e.g., responder, non-responder) based on the spatial arrangements (e.g., spatial interactions between different point categories) given multi-category point data from two classes. This problem is important for generating hypotheses in medical pathology towards discovering new immunotherapies for cancer treatment as well as for other applications in biomedical research and microbial ecology. This problem is challenging due to an exponential number of category subsets which may vary in the strength of spatial interactions. Most prior efforts on using human selected spatial association measures may not be sufficient for capturing the relevant (e.g., surrounded by) spatial interactions which may be of biological significance. In addition, the related deep neural networks are limited to category pairs and do not explore larger subsets of point categories. To overcome these limitations, we propose a Spatial-interaction Aware Multi-Category deep neural Network (SAMCNet) architecture and contribute novel local reference frame characterization and point pair prioritization layers for spatial-configuration-based classification. Extensive experimental results on multiple cancer datasets show that the proposed architecture provides higher prediction accuracy over baseline methods.

cs.CV↗

arXiv2017

T cell equation as a conceptual model of T cell responses for maximizing the efficacy of cancer immunotherapy

Following antigen stimulation, the net outcomes of a T cell response are shaped by integrated signals from both positive co-stimulatory and negative regulatory molecules. Recently, the blockade of negative regulatory molecules (i.e. immune checkpoint signals) demonstrates therapeutic effects in treatment of human cancer, but only in a fraction of cancer patients. Since this therapy is aimed to enhance T cell responses to cancers, here we devised a conceptual model by integrating both positive and negative signals in addition to antigen stimulation. A digital range of adjustment of each signal is formulated in our model for prediction of a final T cell response. This model allows us to evaluate strategies in order to enhance antitumor T cell responses. Our model provides a rational combination strategy for maximizing the therapeutic effects of cancer immunotherapy.

q-bio.CB↗