UROP Research Mentor Project Submission Portal: Submission #1299

serial: '1299'
sid: '21091'
uuid: 745b457d-9fa3-4e83-b304-122a2608750a
uri: /urop-research-mentor-project-submission-portal
created: '1755697813'
completed: '1755700209'
changed: '1756135805'
in_draft: '0'
current_page: ''
remote_addr: 144.174.214.39
uid: '0'
langcode: en
webform_id: urop_project_proposal_portal
entity_type: node
entity_id: '1116'
locked: '0'
sticky: '0'
notes: ''
metatag: meta
data:
  roundtable_info: {  }
  approximately_how_many_hours_a_week_would_the_research_assistant: '10'
  are_you_currently_looking_for_students_: 'Yes'
  confirmation_1: '1'
  contact_email_fsu_email: ''
  contact_email_fsu_email2: ''
  contact_email_fsu_email_if_affiliated_: tuy.nguyen@fsu.edu
  faculty_advisor_confirmation: ''
  faculty_advisor_name: ''
  faculty_advisor_s_fsu_email: ''
  fsu_college: 'FAMU-FSU College of Engineering'
  fsu_department_if_applicable_: 'Electrical and Computer Engineering'
  headshot_optional_: '62811'
  if_the_project_location_is_off_campus_does_the_student_need_to_p: 'Bus service to FSU students is offered through the Seminole Express which will be operational Monday through Friday. Students can get on the bus at University Center C. Use the Seminole Express app to find the most accurate and up-to-date bus times.'
  mentoring_philosophy: 'My mentoring philosophy centers on empowering mentees through a combination of shared experience and a supportive, secure environment. I believe in being open about my own successes and, more importantly, my failures, to normalize the learning process and build a foundation of trust. By creating a space where mentees feel comfortable taking risks and making mistakes, I encourage them to step out of their comfort zones without fear of judgment. This safety net is crucial for genuine learning and personal growth. Instead of providing all the answers, I challenge my mentees to tackle complex problems, guiding them to find their own solutions. I see these challenges not as hurdles but as opportunities for them to develop resilience, critical thinking skills, and a deeper understanding of their field. Ultimately, my goal is to equip mentees with the confidence and skills to navigate their careers independently. This approach fosters a partnership built on mutual respect and a shared commitment to continuous improvement.'
  mentor_handbook_and_faqs: '1'
  name_of_other_faculty_collaborator_if_applicable_: ''
  number_of_assistants_needed_faculty_postdoc_max_6_graduate_stude: '3'
  other_faculty_collaborator_s_preferred_pronouns: ''
  overall_research_project_description: |-
    Cardiovascular diseases continue to be a leading cause of morbidity and mortality worldwide, underscoring the critical importance of accurate and timely diagnosis in healthcare. Among the diagnostic modalities, the electrocardiogram (ECG) signal is a fundamental tool for monitoring cardiac activity. Its intricate waveform provides valuable insights into the heart's electrical activity, aiding clinicians in identifying abnormalities and making informed decisions about patient care. The advent of machine learning (ML) has revolutionized medical diagnostics, offering the potential to enhance the accuracy and efficiency of ECG signal classification. ML models, when trained on vast datasets, can discern subtle patterns and anomalies in ECG signals that may elude conventional diagnostic methods. This transformative capability has paved the way for more precise and timely cardiac diagnoses, contributing to improved patient outcomes. However, the utilization of ML models in healthcare raises concerns about data privacy and security. To address these challenges, federated learning (FL) emerges as a promising paradigm. FL enables the training of ML models across decentralized devices without sharing sensitive data centrally. In the context of ECG signal classification, the integration of FL not only safeguards patient privacy but also facilitates collaborative model training across diverse healthcare institutions 

    This research investigates the synergy between AI and cybersecurity to revolutionize ECG signal classification. By harnessing the power of ML models and the privacy-preserving capabilities of FL, we aim to significantly improve the accuracy and security of cardiac diagnoses. This project seeks to address the following research questions: (1) Can we develop innovative ML models, particularly convolutional neural networks (CNN), to enhance the accuracy and efficiency of ECG signal classification? (2) How can we effectively integrate an FL system to address privacy concerns, ensuring decentralized model training and safeguarding sensitive patient data? (3) What insights can we glean from the nuances of accuracy variations between local and server-side implementations, providing valuable guidance for the deployment of FL-enhanced models in real-world healthcare settings? (4) How does our proposed method compare to existing approaches in terms of classification performance?
  please_add_any_additional_information_here: 'PI Tuy Nguyen is an expert in the area of design, implementation, optimization, and applications of artificial intelligence and cybersecurity in both software and hardware. PI Nguyen has contributed significantly to academia, publishing 22 journal articles, 32 conference papers, and securing two patents. Additionally, his commitment to nurturing future talent is evident in his mentorship, which has resulted in 25 collaborative research papers and one patent with students, particularly one journal paper and three conference papers with undergraduate research teams. PI Nguyen is currently mentoring three Ph.D. students in his research team.'
  please_provide_a_link_to_your_publications_a_video_clip_or_a_web: 'https://www.iiilab.org/publications'
  please_select_the_choice_that_most_accurately_describes_your_exp: 'Partially Remote'
  please_select_the_location_of_your_project_: 'Center for Advanced Power Systems, 2000 Levy Ave, Tallahassee, FL 32310; and FAMU-FSU College of Engineering, 2525 Pottsdamer St, Tallahassee, FL 32310'
  position_availability_for_student_research: 'During Business Hours'
  position_title: Faculty
  primary_research_mentor_name: 'Tuy Nguyen'
  project_keywords: 'Federated learning; Machine Learning; Electrocardiogram Signal Classification; Data Security and Privacy'
  relevant_student_major_s_: 'Open to all majors.'
  research_mentor_preferred_pronoun2: ''
  research_mentor_pronouns: ''
  research_mentor_supervisor_if_different_from_above_: ''
  research_tasks_for_student_research_assistant_s_: |-
    We propose the following solutions to address the research questions.

    - Data Collection and Preprocessing 
    The PTB diagnostic ECG database comprises 14,552 binary-classified ECG signal samples from Physionet's PTB diagnostic database. Recorded at 125Hz, each sample represents the heart's electrical activity. Preprocessing ensures uniformity, involving handling missing values, normalizing amplitudes, and aligning signals. The dataset is split for training and testing, with measures like data augmentation and addressing class imbalances implemented.

    - ECG Classification with Stacked CNN Architecture
    The increased depth enables the stacked CNN to excel in learning intricate temporal patterns, automatically extracting nuanced representations from raw data. This depth is crucial for discerning diverse cardiac conditions, whereas shallower architectures of normal CNNs may struggle to capture complex variations. The mathematical formulation of a stacked CNN's forward pass, involving layer-wise computations, further enhances its capacity to model intricate temporal dependencies, contributing to superior ECG classification performance.

    - Communication and Iterative Processes: 
    To extend our model to a FL system, we consider a scenario where multiple clients collaborate without sharing raw ECG data. Each client has its dataset and trains the model locally. In the proposed iterative process, the global model parameters are communicated to all participating nodes, initiating local training and subsequent model aggregation. Convergence will be monitored throughout the process by assessing changes in the loss function and performance metrics.

    - Preliminary results: Our initial efforts in the domains of biomedical and biometric image security, AI for smart healthcare systems, cybersecurity, and security in IoT communications are integral to the successful realization of this project. More information about our research can be found at https://www.iiilab.org/publications.
  roundtable_times_and_zoom_links: ''
  skills_that_research_assistants_may_need_: 'Hands-on experience with at least one programming language such as Python, Julia, or C/C++ is required. Knowledge of machine learning, federated learning, and ECG signal classification is a strong plus.'
  title_of_the_project: 'Federated Learning for Electrocardiogram Signal Classification'
  update_url: 'https://cre.fsu.edu/urop-research-mentor-project-submission-portal?element_parents=elements/research_mentor_information/headshot_optional_&ajax_form=1&_wrapper_format=drupal_ajax&token=o9GF5G-acvC19i7mBwg_WmPLWjtDM_KLZApbVJJnAUc'
  urop_performance_evaluation: '1'
  urop_poster_presentation: '1'
  when_potential_research_assistants_are_reaching_out_via_email_2: ''
  when_potential_research_assistants_are_reaching_out_via_email_wh: ''
  when_students_are_reaching_out_via_email_what_is_your_preferreda: Dr.
  would_you_like_to_participate_in_the_urop_research_mentor_round2: 'No'
  would_you_like_to_participate_in_the_urop_research_mentor_roundt: ''
  year: '2025'