Assistant Professor of Computer Science

Reza's Picture

BIO

Dr. Reza Sadeghi is an Assistant Professor of Computer Science at Marist College. Before Marist College, he experienced lectureing various Applied Data Science and Computer Science courses at four higher eduction institute of Sacred Heart University, University of New Haven, Bay Path University, and Providence College. Dr. Sadeghi received his PhD in Computer Science from Wright State University. His main specialty is machine learning, and he loves to research in the fields of health care, Web, and learning analytics. Throughout his PhD, he was a graduate research assistant in the Data Science for Healthcare Lab. He was also research trainee at the division of Sleep and Circadian Disorders at Brigham and Women's Hospital, Harvard Medical School. He contributes to the Institute of Electrical and Electronics Engineers (IEEE), IEEE Engineering in Medicine and Biology Society (EMBS), and the American Thoracic Society (ATS).

Office_Hourse

  • Spring 2024 at Hancock Center 3040, Main Campus, Marist College
  • CMPT 120L office hours Tuesday 1:00 PM-2:00PM
  • CMPT 308N office hours Thursday 1:00 PM-2:00PM
  • MSCS 542L office hours Wednesday 2:00 PM-3:00PM
  • DATA-492L office hours Thursday 2:00 PM-3:00PM
  • AI-health for all research group office hours Wednesday 1:00 PM-2:00 PM
  • Notice: Students must contact me through my academic email at least 24 hours ahead to schedule a 15-minute appointment.

Teaching

  • Spring 2024
  • CMPT 120L, Introduction to Programming (Sectoin 112)
  • CMPT 308N, Database Management (Section 111)
  • DATA-492L, Deep Learning (Section 200)
  • MSIS 537L, Database Management I (Section 537)
  • MSCS 542L, Database Management Systems (Section 256)
  • Fall 2023
  • CMPT 120L, Introduction to Programming (Sectoin 114)
  • CMPT 308N, Database Management (Section 113)
  • CMPT 308N, Database Management (Section 201)
  • MSCS 542L, Database Management Systems (Section 256)
  • Samples of student final projects in the above course:
  • Task Management System using Python
  • Spring 2023
  • CMPT 120L, Introduction to Programming
  • CMPT 308N, Database Management
  • MSCS 542L, Database Management Systems
  • CS 617, Artificial Intelligence (Section A)
  • Samples of student final projects in the above courses:
  •  College Data Management System
  • Late Spring 2022
  • CS 504, Introduction to Programming Using Scripting (Section D)
  • CS 551, Introduction to Object-Oriented Programming with Java (Section B)
  • CS 603, Advance Database Design (Section D)
  • CS 617, Artificial Intelligence (Section A)
  • CS 617, Artificial Intelligence (Section B)
  • Samples of student final projects in the above courses:
  •  Phone Book Management System using Python & comma-separated values files
  • Summer 2021
  • ADS654, Deep Learning (Section Z1)

Current Research



Reza's Picture 		Machine learning for healthcare

Stable and accurate clinical decision support systems are essential tools for providing fast diagnosis and high quality treatment of healthcare issues. Such systems are effective as they rely on the risk factors of health issues and are validated over big healthcare data. Machine learning techniques can both extract the risk factors from electronic health records and bring light to complex relationships among the risk factors. Along this theme, I employed distinct machine learning methodologies as simple as logistic regression up to convolutional neural network and probabilistic graphical models to estimate sleep quality, manage dementia, identify disease risk factors, assess alcohol withdrawal, and predict early hospital mortality.


the graph of internet 		Web mining

World Wide Web (WWW) is the biggest source of available information. Web mining tries to extract desired data from this massive information in an efficient way. The information is retrieved from three distinguished levels of content, structure, and usage of accessible documents. From the perspective of web content mining, efficient algorithms can convert huge source of texts, images, sounds, and videos into usable information. However, each source of information implies different concepts in distinct context. As a result, separating the accessible documents into distinguished categories, known as web structure mining, is essential. The speed of producing new information is too fast such that expert knowledge is not enough to categorize the huge amount of WWW information. One of the best methods to address this issue is to consider the navigational behavior of users known as web usage mining. I am following these three web mining levels through text mining, web robots, and web structure characterization.


Reza's Picture 		Structure learning

Learning Bayesian network structures from data is a NP hard problem since the number of candidate graphs grows super-exponentially. Through this research, the meaningful structure of data is investigated to discover unknown relations in biomedical fields. This research utilizes both advantages of constant-based and score-based methods. Bayesian network learning is commonly executed as structure learning and parameter learning, respectively. The structure learning point to learn the structure of a directed acyclic graph. However, the parameter learning tries to address learning the local distribution showed by the structure of learned directed acrylic graph via the structure learning. Both learning stages can handle by unsupervised and supervised learning methods.

Past Research



Reza's Picture 		Sleep quality prediction using heart rate variability

Reza's Picture We employed machine learning techniques to enhance sleep quality prediction by selecting more effective features and reducing the dependability to the expert knowledge in feature computation. We designed several sleep quality methodologies by applying a great range of machine learning techniques over electronic health records, heart rate variability (HRV) features, and the raw Electrocardiogram (ECG) signals. In the process of predicting sleep quality from HRV, we came up with a clinical decision support system that processes the raw electrocardiogram signals independently from the prior knowledge of sleep experts. This system employs a convolutional neural network (CNN) to predict sleep quality based on heart activities during each night by analyzing images of two ECG signals during Polysomnography studies. To our knowledge, this is one of the first studies to predict sleep quality using HRV. The detailed report of this research is provided in “Predicting sleep quality in osteoporosis patients using electronic health records and heart rate variability”, which is presented in 42nd IEEE Engineering in Medicine and Biology Society (EMBC2020).


Reza's Picture

Identifying the disease-specific risk factors of daytime sleepiness

Reza's Picture Abnormal sleep quality or sleep quantity causes excessive daytime sleepiness (EDS), which is a highly prevalent condition in the older adult society. EDS is a symptom of several diseases, such as neurological disorders, e.g. dementia, and sleep breathing disorders, e.g. apnea. Distinguishing disease-specific risk factors of EDS can both reveal underlying reasons of abnormal sleep quality and enhance its prediction. To do so, we investigated EDS risk factors in two groups of patients one group with and one group without severe sleep apnea. The report of this research has been reflected in “Sleep Propensity and Sleep Apnea-Specific Hypoxia Are Associated with Excessive Daytime Sleepiness” and presented at Annals of the American Thoracic Society (ATS2020).


Reza's Picture

Sleep quality prediction in caregivers of people with dementia

Reza's Picture Most caregivers of people with dementia (CPWD) experience a high degree of stress due to the demands of providing care, especially when addressing unpredictable behavioral and psychological symptoms of dementia. Such challenging responsibilities make caregivers susceptible to poor sleep quality with detrimental effects on their overall health. Hence, monitoring caregivers’ sleep quality can provide important CPWD stress assessment. Most current sleep studies are based on polysomnography, which is expensive and potentially disrupts the caregiving routine. To address these issues, we propose a clinical decision support system to predict sleep quality based on trends of physiological signals in the deep sleep stage. This research has been elaborated on “Sleep quality prediction in caregivers using physiological signals” (Presentation slides / Source code) and published in Computers in Biology and Medicine. It is also featured in Using the E4 to assess sleep in caregivers of people with dementia.


Reza's Picture

Alcohol withdrawal prediction

Reza's Picture By the definition of the World Health Organization, alcohol use disorder refers to any form of alcohol consumption that causes health problems. Alcohol Withdrawal Syndrome (AWS) occurs roughly 4 to 72 hours following cessation or reduction of prolonged, heavy alcohol ingestion. Withdrawal delirium or Delirium Tremens is considered the most dangerous symptom of AWS which can lead to the death. During the initial 8 hours after the last drink, patients face with anxiety, insomnia, nausea, and abdominal pain. This condition is followed by high blood pressure, increased body temperature, unusual heart rate, and confusion. If this syndrome does not receive any treatment, the patients will suffer from hallucinations, fever, seizures, and agitation. As a result, there is an essential need to predict and treat this syndrome in the initial stages. The outcomes of this research are described in “Predicting alcohol withdrawal in intensive care units” and presented at The Symposium of Student Research, Scholarship, and Creative Activities 2020.


Reza's Picture

Mortality Prediction

Reza's Picture Early hospital mortality prediction is critical as intensivists strive to make efficient medical decisions about the severely ill patients staying in intensive care units. As a result, various methods have been developed to address this problem based on clinical records. However, some of the laboratory test results are time-consuming and need to be processed. To address this issue, we proposed a novel method to predict mortality using features extracted from the heart signals of patients within the first hour of ICU admission. In order to predict the risk, quantitative features have been computed based on the heart rate signals of ICU patients. Outcomes of the experiments indicates that heart rate signals can be used for predicting mortality in patients in the ICU, achieving a comparable performance with existing predictions that rely on high dimensional features from clinical records which need to be processed and may contain missing information. The outcomes of this research are reported in “Early Hospital Mortality Prediction using Vital Signals” (Presentation slides / Source code) and has been accepted in IEEE/ACM CHASE 2018 and published in the journal of Smart Health.


Reza's Picture

Tor structure characterization

Reza's Picture Tor is among most well-known dark net in the world. It has noble uses, including as a platform for free speech and information dissemination under the guise of true anonymity, but may be culturally better known as a conduit for criminal activity and as a platform to market illicit goods and data. Past studies on the content of Tor support this notion, but were carried out by targeting popular domains likely to contain illicit content. A survey of past studies may thus not yield a complete evaluation of the content and use of Tor. This work addresses this gap by presenting a broad evaluation of the content of the English Tor ecosystem. The outcomes of this research are reported in “A Broad Evaluation of the Tor English Content Ecosystem” and “Interaction of Structure and Information on Tor” is presented in "10th ACM Conference on Web Science" and "COMPLEX NETWORKS 2020", respectively.


Reza's Picture

Web Robot Detection

Reza's Picture The accurate detection of web robot sessions from a web server log is essential to take accurate traffic- level measurements and to protect the performance and privacy of information on a Web server. Moreover, the irrecoverable risks of visits from malicious robots that intentionally try to evade web server intrusion detection systems, covering-up their visits with fabricated fields in their http r est packets, cannot be ignored. To separate both types of robots from humans in practice, analysts turn to heuristic methods or state-of-the-art soft computing approaches that have only been tuned to the specification of a kind of web server. Noting that the landscape of web robot agents is ever changing, and that behavioral patterns and characteristics vary across different web servers, both options are lacking. To overcome this challenge, my colleagues and I proposed several methods based on Fuzzy Rough Set, Markov Clustering, Self-organizing Map concepts. The report of this research is reflected in “A soft computing approach for benign and malicious web robot detection” (Source code) and “Detection of Web site visitors based on fuzzy rough sets” published in Expert Systems with Applications journals (Impact factor: 3.928) and Soft Computing (Impact factor: 2.472), respectively.


Reza's Picture

Outlier Detection

Reza's Picture Event handlers have wide range of applications such as medical assistant systems and fire suppression systems. These systems try to provide accurate responses based on the least information. Support vector data description (SVDD) is one of the appropriate tools for such detections, which should handle lack of information. Therefore, many efforts have been done to improve SVDD. Unfortunately, the existing descriptors suffer from weak data characteristic in sparse data sets and their tuning parameters are organized improperly. These issues cause reduction of accuracy in event handlers when they are faced with data shortage. Therefore, my colleagues and I proposed several methods based on Fuzzy Rough Set, Bat Algorithm, and Chaos theory. The results of this research is presented in “Weighted support vector data description based on chaotic bat algorithm” and “Automatic support vector data description” (Source code) published in Applied Soft Computing journals (Impact factor: 3.541) and Soft Computing (Impact factor: 2.472), respectively.


Reza's Picture

Dynamic Facility Location

Reza's Picture Determination of facilities, such as factories or warehouses, location and availability conditions is one of the important and strategic decisions for an organization to make. Transportation costs that form a major part of goods price are dependent to this decision making. There are verity of methods have been presented to achieve the optimal locations of these facilities which are generally deterministic. In real world accurate estimation of the effective parameters on this optimal location for single or multiple time periods is difficult and merely impossible.In this research, my colleagues and I tried to achieve an efficient model with consideration of uncertainty demand over different time periods on the basis of previously presented models that we call stochastic dynamic facilities location problem. In order to do so we use stochastic constrain programming which convert the stochastic model to a deterministic one. The results of this research is presented in “Dynamic Facility Location with Stochastic Demand” and Stochastic Facilities location Model by Using Stochastic Programming” published in Shiraz Journal of System Management.


Reza's Picture

Job Scheduling

Reza's Picture Nowadays the requests of managers and other persons who benefited by such projects for drop total project’s cost have increased considerably. Besides, the amount of changes exert on this factors can result to variation in initial estimation of them. In this way, however; by observing this types of changes in different circumstances, the ideal quality for projects is going to be considered. Moreover, generally in realty world, either falling in time consuming or the amount usage of resources for a task could not lead to decline of task quality. As a result, assigning fixed and deterministic values for assessing ideal quality leading to unpredictable outcomes. I this study, fuzzy logic developed remarkably to measure the quality for wide range of tasks and activities in various variation circumstances. At last, the presented model apply to real case study and the obtained values have proved the efficiency of proposed model in comparison to others in deterministic situations. The report of this research is reflected in “Solving the equilibrium problem of time, cost, resource and quality of project network by using expanded fuzzy logic set”.

Selected Publications

Research Service

Education

  • B.S., Computer Engineering- Software, 2012
  • Isfahan University of Technology (IUT), Department of Electrical and Computer Engineering, Isfahan, Iran
  • Project: Designing & implementation of online library management website in IUT High School

Future Direction

  • How do our neurons make decisions for us?
    https://zuckermaninstitute.columbia.edu/


  • How can we create a comprehensive wearable device?
    Reza's pics


  • Can organ failure be prevented?
    Reza's pics

Frequently Questioned Answers

  • How can I take a research-based course or an independent study credite with you?
  • You must earned an A in one of my courses to take any research-based courses with me.
  • How can I become a member of AI-health for all research group ?
  • Send an email with the title of "AI-health for all research group" that contains your resume, your research interest, your availability, and a time spot from research group office hours
  • What should I do if you are in my thesis committee?
  • Please submit your thesis draft at least two weeks before your thesis defense.