A5059480804- Radiomics and Machine Learning in Medical Imaging
- Distributed Control Multi-Agent Systems
- Lung Cancer Diagnosis and Treatment
- Cloud Computing and Resource Management
- Modular Robots and Swarm Intelligence
- Cloud Data Security Solutions
- Imbalanced Data Classification Techniques
- Machine Learning and Data Classification
- Adaptive Control of Nonlinear Systems
- Distributed and Parallel Computing Systems
- Molecular Biology Techniques and Applications
- Advanced Radiotherapy Techniques
- Reinforcement Learning in Robotics
- Mobile Ad Hoc Networks
- Advanced X-ray and CT Imaging
- AI in cancer detection
- Machine Learning in Healthcare
- Distributed systems and fault tolerance
- Head and Neck Cancer Studies
- Energy Efficient Wireless Sensor Networks
- Gastric Cancer Management and Outcomes
- Robotic Path Planning Algorithms
- University-Industry-Government Innovation Models
- Domain Adaptation and Few-Shot Learning
- Blockchain Technology Applications and Security
Washington University in St. Louis
2024-2025
Mallinckrodt (United States)
2022-2024
Alvin J. Siteman Cancer Center
2024
University of Pennsylvania
2015-2022
California University of Pennsylvania
2018-2021
University of New Mexico
2010-2020
University of California, San Francisco
2016
Pomeranian Medical University
2004
Machine learning algorithms that are both interpretable and accurate essential in applications such as medicine where errors can have a dire consequence. Unfortunately, there is currently tradeoff between accuracy interpretability among state-of-the-art methods. Decision trees therefore used extensively throughout for stratifying patients. Current decision tree algorithms, however, consistently outperformed by other, less-interpretable machine models, ensemble We present MediBoost, novel...
Machine learning is proving invaluable across disciplines. However, its success often limited by the quality and quantity of available data, while adoption level trust afforded given models. Human vs. machine performance commonly compared empirically to decide whether a certain task should be performed computer or an expert. In reality, optimal strategy may involve combining complementary strengths humans machines. Here, we present expert-augmented (EAML), automated method that guides...
Significance As machine learning applications expand to high-stakes areas such as criminal justice, finance, and medicine, legitimate concerns emerge about high-impact effects of individual mispredictions on people’s lives. a result, there has been increasing interest in understanding general models overcome possible serious risks. Current decision trees, Classification Regression Trees (CART), have played predominant role fields due their simplicity intuitive interpretation. However, trees...
Abstract Background: Prostate cancer (PCa) progression has been shown to be influenced by the relationship of extracellular matrix (ECM) proteins within stroma, and immune tumor cells. However, current standard care for PCa patients does not take these factors into account. Previously, our team identified sets ECM glycans using matrix-assisted laser desorption/ionization (MALDI) imaging that were associated with PCa. We also found preliminary evidence MALDI can guide pathomic textural...
We show the Lyapunov stability and convergence of an adaptive decentralized coverage control for a team mobile sensors. This new approach assumes nonholonomic sensors rather than usual holonomic found in literature. The kinematics unicycle model nonlinear law polar coordinates are used order to prove controller applied over feasibility algorithm verified through simulations Matlab. Furthermore, some experiments carried out using four Pioneer 3-AT robots sensing piecewise constant light...
Abstract Background and purpose Chest wall toxicity is observed after stereotactic body radiation therapy ( SBRT ) for peripherally located lung tumors. We utilize machine learning algorithms to identify predictors develop dose–volume constraints. Materials methods Twenty‐five patient, tumor, dosimetric features were recorded 197 consecutive patients with Stage I NSCLC treated , 11 of whom (5.6%) developed CTCAE v4 grade ≥2 chest pain. Decision tree modeling was used determine syndrome CWS...
We evaluate radiomic phenotypes derived from CT scans as early predictors of overall survival (OS) after chemoradiation in stage III primary lung adenocarcinoma. retrospectively analyzed 110 thoracic acquired between April 2012−October 2018. Patients received a median radiation dose 66.6 Gy at 1.8 Gy/fraction delivered with proton (55.5%) and photon (44.5%) beam treatment, well concurrent chemotherapy (89%) carboplatin-based cisplatin-based (36.4%) doublets. A total 56 death events were...
Abstract In this paper we show the Lyapunov stability of an adaptive and decentralized coverage control for a team mobile sensors. This new approach assumes nonholonomic sensors rather than holonomic ones usually found in literature. Furthermore, sufficient conditions are provided to guarantee ultimate bound system when presence time‐varying sensory functions. The convergence feasibility verified through simulation experimental results.
Radiation esophagitis is a clinically important toxicity seen with treatment for locally-advanced non-small cell lung cancer. There considerable disagreement among prior studies in identifying predictors of radiation esophagitis. We apply machine learning algorithms to identify factors contributing the development uncover previously unidentified criteria and more robust dosimetric factors.We used approaches grade ≥ 3 cohort 202 consecutive cancer patients treated definitive chemoradiation...
This study tackles interobserver variability with respect to specialty training in manual segmentation of non-small cell lung cancer (NSCLC). Four readers included for are: a data scientist (BY), medical student (LS), radiology trainee (MH), and specialty-trained radiologist (SK) total 293 patients from two publicly available databases. Sørensen-Dice (SD) coefficients low rank Pearson correlation (CC) 429 radiomics were calculated assess variability. Cox proportional hazard (CPH) models...
This is the first part of a paper that provides an overview some applications control theory to computing systems. With advent cloud and more affordable infrastructures, engineers are looking for provide rigorous tools analyze design systems algorithms. On other hand, have greatly benefited from distributed infrastructure. In I we will focus on at data center level.
Abstract We aim to determine the feasibility of a novel radiomic biomarker that can integrate with other established clinical prognostic factors predict progression-free survival (PFS) in patients non-small cell lung cancer (NSCLC) undergoing first-line immunotherapy. Our study includes 107 stage 4 NSCLC treated pembrolizumab-based therapy (monotherapy: 30%, combination chemotherapy: 70%). The ITK-SNAP software was used for 3D tumor volume segmentation from pre-therapy CT scans. Radiomic...
In this paper, we develop a decentralized probabilistic method for performance optimization of cloud services. We focus on Infrastructure-as-a-Service where the user is provided with ability configuring virtual resources demand in order to satisfy specific computational requirements. This novel approach strongly supported by theoretical framework based tail probabilities and sample complexity analysis. It allows not only inclusion metrics but incorporation security cryptographic algorithms...
No two robots are exactly the same-even for a given model of robot, different units will require slightly controllers. Furthermore, because change and degrade over time, controller need to time remain optimal. This paper leverages lifelong learning in order learn controllers robots. In particular, we show that by set control policies with (unknown) motion models, can quickly adapt changes or new robot unique disturbances. approach is completely model-free, allowing us apply this method have...
In this paper we present our multi-vehicle testbed that was designed for verification and validation of cooperative control algorithms involving environmental sensing. Two algorithms: prioritized multi-sensing behavior, a distributed adaptive algorithm nonholonomic sensor networks are qualitatively verified using testbed. The allows straightforward transition from simulation to experimenting on actual hardware has the flexibility interface various types sensors, vehicles, as well enable...
Background: Lung cancer is one of the most common cancers in United States and fatal, with 142,670 deaths 2019. Accurately determining tumor response critical to clinical treatment decisions, ultimately impacting patient survival. To better differentiate between non-small cell lung (NSCLC) responders non-responders therapy, radiomic analysis emerging as a promising approach identify associated imaging features undetectable by human eye. However, plethora variables extracted from an image may...
Security and resource optimization are two of the most critical concerns in cloud computing. A provider must ensure customers with appropriate security, while optimizing use resources. In this paper, we present a framework which optimizes both resources security provided to an infrastructure as service (IaaS) cloud. Our offers secure usage control sensitive data within virtual machines (VMs), dynamically instantiated allocated VMs. These then VMs using model based upon randomized algorithms....
In this paper we describe the development of a system that provides security provisioning and performance controls over content in cloud environment. Using an approach grounded Usage Management Control theory are able to successfully provision resources multiple systems. Since providing costs more therefore variable control model for different levels is proposed. This allocates Virtual Machines adaptively so desired measure lies between predefined upper lower bounds as agreed Service Level Agreement.
This is the second part of a paper that provides an overview some applications control theory to computing systems. With advent cloud and more affordable infrastructures, engineers are looking for provide rigorous tools analyze design systems algorithms. On other hand, have greatly benefited from distributed infrastructure. In paper, we will focus on at server level, specifically CPU utilization control.
In this paper we provide stabilization criteria for a class of nonlinear systems with time-varying state and input delays. The system is assumed to be separable into nominal linear part an additive perturbation satisfying the growth constraint. delays as well their time derivatives are bounded above. A state-feedback controller chosen close control loop. validation approach carried out through two simulation examples. first example uses theoretical model all conditions established in our...