Amir H Assadi | University of Wisconsin-Madison (original) (raw)

Papers by Amir H Assadi

Nature Scientific Reports, 2019

Individuals with autism spectrum disorder struggle with motor difficulties throughout the life sp... more Individuals with autism spectrum disorder struggle with motor difficulties throughout the life span, and these motor difficulties may affect independent living skills and quality of life. Yet, we know little about how whole-body movement may distinguish individuals with autism spectrum disorder from individuals with typical development. In this study, kinematic and postural sway data were collected during multiple sessions of videogame play in 39 youth with autism spectrum disorder and 23 age-matched youth with typical development (ages 7-17 years). The youth on the autism spectrum exhibited more variability and more entropy in their movements. Machine learning analysis of the youths' motor patterns distinguished between the autism spectrum and typically developing groups with high aggregate accuracy (up to 89%), with no single region of the body seeming to drive group differences. Moreover, the machine learning results corresponded to individual differences in performance on standardized motor tasks and measures of autism symptom severity. The machine learning algorithm was also sensitive to age, suggesting that motor challenges in autism may be best characterized as a developmental motor delay rather than an autism-distinct motor profile. Overall, these results reveal that whole-body movement is a distinguishing feature in autism spectrum disorder and that movement atypicalities in autism are present across the body. Autism spectrum disorder (ASD) is a neurodevelopmental disability that affects 1 in 59 children in the US 1 and has staggering public health costs, estimated at $461 billion annually by 2025 2. In addition to the primary social communication and repetitive-behavior deficits 3 , individuals with ASD face a diversity of co-occurring motor impairments, including challenges with postural stability 4 , manual motor functions 5-7 , and motor anticipation 8,9. While motor challenges are not part of the diagnostic criteria for ASD, motor challenges in the first two years of life have been consistently found to be an early indicator of later ASD symptoms and diagnosis 10-14 , and motor challenges in youth with ASD have been found to be associated with core ASD symptoms 6,15,16. Motor difficulties in individuals with ASD appear to persist across their life span 7,17,18 and have been linked to poorer independent-living skills (i.e., dressing, toileting, self-grooming, household tasks, and finances) across multiple age groups 7,19. For example, manual motor skills were associated with poorer adaptive daily living skills both concurrently and five-to-nine-years later, even after controlling for age and IQ 7. This link between motor challenges and independent-living skills is important because it suggests that targeting motor skills through intervention could potentially remove barriers to independent living in order to enhance quality of life in ASD. Yet, we know little about how whole-body movement may be atypical in ASD, with the majority of studies examining just one motor domain at a time. It is important to examine which aspects of movement most robustly differ in ASD compared to typical development (TD), as this information would allow us to detect and treat motor challenges in ASD. A key step toward this goal is to determine which motor domains are the most distinctive in ASD compared to TD. Anzulewicz and colleagues 20 found that fine motor skills while playing two tablet games were shown to reliably distinguish between 3-6 year-olds with ASD and 3-6 year-olds with TD. Specifically,

Understanding the structure of relationships between objects in a given database is one of the mo... more Understanding the structure of relationships between objects in a given database is one of the most important problems in the field of data mining. The structure can be defined for a set of single objects (clustering) or a set of groups of objects (network mapping). We propose a method for discovering relationships between individuals (single or groups) that is based on what we call the empirical topology, a system-theoretic measure of functional proximity. To illustrate the suitability and efficiency of the method, we apply it to an astronomical data base.

Differential gene expression analysis in treatment of Parkinson's disease using the moduli s... more Differential gene expression analysis in treatment of Parkinson's disease using the moduli space of triangles

2017 International Conference on Computational Science and Computational Intelligence (CSCI)

Here we discuss the design of mathematical models for predicting air quality based on the empiric... more Here we discuss the design of mathematical models for predicting air quality based on the empirical topology of regional atmospheric chemistry data. First, Markov chain model trained on historical local data predicts air pollution levels, evaluated by a fuzzy comprehensive decision maker. Secondly, a singular vector machine (SVM) is trained to identify conditions leading to higher pollution levels. Additionally, we describe local-to-global methods for estimation of air quality over a region containing non-uniformly-distributed data points. Together these techniques forecast air quality over an entire region, based on the current state of the atmosphere (and historical training data) from heterogeneously-distributed geographically-disjoint measurement locations

ArXiv, 2008

The history of research on eigenvalue problems is rich with many outstanding contributions. Nonet... more The history of research on eigenvalue problems is rich with many outstanding contributions. Nonetheless, the rapidly increasing size of data sets requires new algorithms for old problems in the context of extremely large matrix dimensions. This paper reports on a new method for finding eigenvalues of very large matrices by a synthesis of evolutionary computation, parallel programming, and empirical stochastic search. The direct design of our method has the added advantage that it could be adapted to extend many algorithmic variants of solutions of generalized eigenvalue problems to improve the accuracy of our algorithms. The preliminary evaluation results are encouraging and demonstrate the method's efficiency and practicality.

The ubiquitous role of the cyber-infrastructures, such as the WWW, provides myriad opportunities ... more The ubiquitous role of the cyber-infrastructures, such as the WWW, provides myriad opportunities for machine learning and its broad spectrum of application domains taking advantage of digital communication. Pattern classification and feature extraction are among the first applications of machine learning that have received extensive attention. The most remarkable achievements have addressed data sets of moderate-to-large size. The 'data deluge' in the last decade or two has posed new challenges for AI researchers to design new, effective and accurate algorithms for similar tasks using ultra-massive data sets and complex (natural or synthetic) dynamical systems. We propose a novel principled approach to feature extraction in hybrid architectures comprised of humans and machines in networked communication, who collaborate to solve a pre-assigned pattern recognition (feature extraction) task. There are two practical considerations addressed below: (1) Human experts, such as pla...

Neural Networks for Signal Processing X. Proceedings of the 2000 IEEE Signal Processing Society Workshop (Cat. No.00TH8501), 2000

Prospects in topology: proceedings of a conference …, 1995

Algebraic Geometric Invariants for Homotopy Actions Amir H. Assadi 1 Introduction Let X be a para... more Algebraic Geometric Invariants for Homotopy Actions Amir H. Assadi 1 Introduction Let X be a paracompact topological space, and let'H (X) denote the monoid of self-homotopy equivalences of X. The homotopy equivalences which are ho-motopy equivalent to the identity map of ...

Scientific Reports

Neurocomputing, Jun 1, 2001

An interesting problem in analysis of video data concerns design of algorithms that detect percep... more An interesting problem in analysis of video data concerns design of algorithms that detect perceptually significant features in an unsupervised manner, for instance methods of machine learning for automatic classification of human expression. A geometric formulation of this genre of ...

Chem Phys Lipids, 2002

We have argued the importance of a differential geometric approach to describing the nonlinear fe... more We have argued the importance of a differential geometric approach to describing the nonlinear features of massive data sets. Based on biological models, one would expect the behavior of the brain to be comparable with a nonlinear dynamical system. Preliminary results from an investigation of the nonlinearity of brain activation as measured by fMRI studies motivate a careful study as we have outlined. From identifying features of data (i.e. characterizations of linearity or nonlinearity) in the two-dimensional slices of massive data sets, it is possible to construct a Riemannian curvature tensor with which one can describe global behavior of points in a data set. We demonstrate results of feature extraction based on local principal component analysis and spline fitting on both synthetic and fMRI data

Neural Networks and Neurocomputing provide a natural paradigm for parallel and distributed proces... more Neural Networks and Neurocomputing provide a natural paradigm for parallel and distributed processing. Neurocomputing within the context of classical computation have been used for approximation and classification tasks with some success. In this paper we propose a model for Quantum Neurocomputation and explore some of its properties and potential applications to pattern recognition.