Events Archive 2023 | Emory School of Medicine

2023 Events

2023 Fall/Winter Seminar Series

Automated Gender Bias Identification in Textbooks

Date: Dec. 19, 2023 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Sudeshna Das, Postdoctoral Researcher
Abstract:
Gender bias is the inclination towards or unfair discrimination against individuals of one gender in favor of individuals of other gender(s). The presence of gender bias is seen in text from various domains, such as education, healthcare, and social media. In this talk, I will describe my work on automated gender bias detection in textbooks. I will start by discussing the nature of gender biases seen in textbooks, deep neural network-based methods for gender bias detection, and the importance of linguistic markers of gender in detecting gender bias. To that end, I will present a model that is designed to capture these linguistic markers of bias, along with global context dependence and locality of gender bias. I will also talk about the need for a gender-agnostic view of gender bias.

Analysis of Wearable Device Data Using Functional Data Models

Date: Dec. 5, 2023 | 12:00PM-1:30PM, BMI Classroom 4004 or on Zoom
Speakers: Julie Wrobel, PhD
Abstract:
The ability of individuals to engage in physical activity is a critical component of overall health and quality of life. Establishing normative trends of physical activity is essential to developing public health guidelines and informing clinical perspectives regarding individuals’ levels of physical activity. Beyond overall quantity of physical activity, patterns regarding the timing of activity provide additional insights into latent health status. Wearable accelerometers, paired with statistical methods from functional data analysis, provide the means to estimate diurnal patterns in physical activity. Using methods we developed for separating amplitude and phase variability in exponential family functional data, we uncover the distinct phenotypes, or chronotypes, that give rise to differences in these patterns in physical activity, as well as how daily patterns in physical activity change with age.

Prediction of Hypoxic-Ischemic Encephalopathy Using Events in Fetal Heart Rate and Uterine Pressure

Date: Nov. 7, 2023 | 12:00PM-1:00PM, BMI Classroom 4004 or on Zoom
Speakers: Johann Vargas-Calixto, PhD. Candidate
Abstract:
In developed countries, visual assessment of intrapartum fetal heart rate (FHR) and uterine pressure (UP) signals is the standard approach for identifying infants at risk of hypoxic-ischemic encephalopathy (HIE). However, this method suffers from significant intra- and inter-observer variability, leading to diagnostic uncertainty and variable intervention rates. Our research takes advantage of a large set of approximately 41,000 births at Kaiser Permanente Northern California hospitals, acquired between 2011 and 2019, to address these challenges. We aim to demonstrate that FHR events, including baseline, accelerations, and decelerations, as well as UP contractions, display different properties in infants with and without HIE. Using these events, we developed a random forest classifier for the early detection of HIE. We also explored the influence of clinical risk factors, specifically nulliparity, on the performance of the HIE classifier. Preliminary results suggest a strong potential for early HIE detection. In turn, early detections could lead to timely clinical interventions. Our ongoing efforts are focused on further improving our classifier performance based on intrapartum FHR and UP signals. Eventually, we expect that a system that provides a timely detection of HIE could lead to the reduction of its catastrophic consequences.

2023 Spring Seminar Series

Portable and Integrative Biomechanics Labs for Precision Rehabilitation

Date: May 23, 2023 | 1:00PM-2:00PM, on Zoom
Speakers: Eni Halilaj, PhD
Abstract:
Musculoskeletal conditions are the leading cause of disability worldwide, costing the United States’ economy $950 billion per year in direct and indirect costs. In this talk, I will share our work on key challenges toward precision rehabilitation for musculoskeletal injuries and diseases. I will first describe how we are combining deep learning with physics-based modeling to democratize motion capture for all researchers, clinicians, and patients, making it accessible from personal smartphones and inexpensive wearable sensors. These portable tools will strengthen the feedback loop between research and clinical practice, enabling large-scale research-grade data to be collected outside of specialized gait laboratories. I will then talk about the importance of understanding interactions between mechanical loading (e.g., in response to exercise, injury, assistive wearables) and biological factors (i.e., deep patient phenotyping) when personalizing rehabilitation protocols and assistive devices. Our ultimate goal is to build smart rehabilitation technologies that can capture meaningful biomechanical outcomes in natural environments, reason about what these biomechanics mean for a particular patient and guide the delivery of personalized feedback in real time. Toward this vision, we are studying flexible sensors that can be embedded into smart garments for continuous monitoring and haptic feedback. In this last part of my talk, I will share our collaborative work on characterizing capacitive sensing as a biomechanics-monitoring tool.

Minding the Gap: Why Realizing the Potential of Social Determinants of Health is Important for Biomedical Research in Addressing Mental Health Disparities and Youth Suicide

Date: May 9, 2023 | 1:00PM-2:00PM, on Zoom
Speakers: Yunyu Xiao, PhD
Abstract:
In this talk, we will dive into the alarming public health crisis of suicide, a pervasive issue that disproportionately impacts marginalized communities. We will embark on a journey to uncover the hidden epidemiological patterns of health disparities and suicide, and unveil the powerful Social Determinants of Health (SDoH) models rooted in social sciences.
As we explore the SDoH framework, you will learn how its integration into marketing and biomedical research is critical to promoting mental health equity and fostering change. We'll delve into various cutting-edge methodologies for identifying, studying, and addressing SDoH, opening up a world of possibilities for revolutionizing mental health care.
Finally, we will reveal the remarkable results achieved through interdisciplinary collaboration, harnessing the power of electronic health records and large-scale population-based studies. Together, we will unlock the potential of SDoH to transform the landscape of mental health equity, and work towards a future where everyone has the opportunity to thrive.

Unlocking the Secrets of Early Brain Development: Advancements in Neuroimaging, AI, and Multi-Omics

Date: April 25, 2023 | 1:00PM-2:00PM, on Zoom

Speakers: Yun Wang, PhD

Abstract:

Studying early brain development is crucial for understanding the foundations of cognitive and emotional functioning, and for identifying potential interventions to promote mental health. The primary research objective is to unravel the complexities of early neurodevelopment and identify key factors contributing to risk and resilience for mental illness. The laboratory harnesses artificial intelligence (AI) to enhance medical imaging analysis in fetuses and infants and employs a multi-omics approach to investigate intricate relationships between maternal factors, multi-omics data, brain development, and behavioral outcomes, such as social deficits, emotional dysregulation, and other developmental or psychiatric disorders. To address the shortcomings of traditional neuroimaging techniques, the lab is developing an advanced, accurate, and scalable AI-driven analysis platform specifically designed for fetal and infant brain imaging. Utilizing AI's capabilities in multi-omics, the interconnected factors influencing multi-omics data and their subsequent effects on brain development and behavior are explored. This pioneering approach will lay the foundation for novel treatments and interventions, fostering healthy development and reducing the prevalence of neurodevelopmental and psychiatric disorders. In essence, the lab's work focuses on revolutionizing fetal and infant neuroimaging through AI integration and in-depth multi-omics data analysis, providing a comprehensive understanding of early brain development, the dynamic interplay of key determinants, and their implications for behavioral outcomes and disorders.

Enhancing Diagnostic Precision through AI-Driven Imaging Signatures

Date: April 18, 2023 | 12:00PM-1:00PM, in BMI classroom 4004 or on Zoom

Speakers: Saima Rathore, PhD

Abstract:

As modern medical imaging becomes increasingly multi-parametric, interpreting these images in neurodegenerative and oncologic diseases has become more complex. This underscores the importance of developing advanced methods that condense the vast amount of information these images convey into a concise, comprehensive, and easily interpretable panel of measurements that provide highly informative insights. Artificial Intelligence (AI) methods have revolutionized the integration of diverse imaging data, revealing unique and individualized imaging patterns with diagnostic and prognostic value. They have enabled researchers to move beyond group comparisons and instead focus on identifying imaging biomarkers that offer personalized value. This talk will review several research directions around this topic, emphasizing the use of AI-based advances in imaging-based biomarkers for faster, safer, and more effective implementation of biomarker-based classification, risk assessment, diagnosis, and prognosis in neurodegenerative and oncologic diseases. Additionally, this talk will discuss how AI-based methods have been instrumental in breaking down broader diagnostic categories into more distinct subgroups and how they can non-invasively estimate molecular characteristics and predict therapy responses. The discussion will also demonstrate how introducing precise AI-based biomarkers for clinical outcomes is enhancing precision diagnostics and paving the way for a revolutionary approach to precision medicine that can deliver more personalized and effective patient care.

Multi-domain data integration for precision health

Date: March 21, 2023 | 1:00PM-2:00PM, on Zoom

Speakers: Xiao Li, PhD

Abstract:

The large amount of biomedical data derived from molecular profiling, wearable sensors, and electronic health records are rapidly transforming our healthcare systems. Our lab focuses on integrating these multi-domain data to promote precision health. Recently, we conducted a secondary analysis of dietary, metabolic, and molecular data collected from 609 participants before, during, and after a 1-year weight-loss intervention with either a healthy low-carbohydrate (HLC) or a healthy low-fat (HLF) diet. Through systematic analysis of multidomain datasets, we identified dietary, molecular, and metabolic factors, common or unique to the HLC and HLF diets, providing a roadmap for developing individualized weight-loss strategies. As a widely deployed technology that has been used by tens of millions of people worldwide, the utility of wearable technologies in health management has been another major focus of the lab. In the current pandemic, we investigate the use of wearable devices for the early detection of COVID-19 in a retrospective manner and also present an approach for using wearable device-detected physiological parameters for real-time health monitoring and surveillance. To meet the big data challenges, we further developed the Personal Health Dashboard (PHD), which utilizes state-of-the-art security and scalability technologies to provide an end-to-end solution for big biomedical data analytics. Supported by PHD framework, we are currently investigating the potential of wearable technologies in precisely predicting and managing human diseases and health.

Wearable Sensors and Artificial Intelligence Technologies for Chronic and Infectious Disease Monitoring

Date: March 14, 2023 | 12:00PM-1:00PM, BMI Classroom 4004 and on Zoom

Speakers: MD Mobashir Hasan Shandhi, PhD

Abstract:

The majority of the health care costs related to the treatment of chronic and infectious diseases are attributed to direct care costs (e.g., hospital admissions and readmissions). The prevalence of chronic diseases and associated costs in the United States is growing at an alarming pace. The COVID-19 pandemic has further impacted the health of high-risk individuals by increasing the likelihood of more severe illness for those with underlying health conditions and associated healthcare costs. There have been ample efforts from researchers and clinicians to develop remote healthcare systems and wearable devices to manage patients with chronic and infectious diseases in home settings, which has reduced the burden on inpatient care facilities and gained further momentum during the COVID-19 pandemic. Yet, there is a lack of reliable wearable devices that can provide clinically acceptable information to healthcare professionals, as well as a lack of emphasis on validating wearable and artificial intelligence technologies in representative populations to enable a reliable and equitable remote health management system. This talk will present the challenges and potential solutions for developing tools (i.e., wearable sensors and computational algorithms) for reliable and equitable remote patient monitoring systems for chronic and infectious diseases. More specifically, the presenter will share the development and validation of wearable sensors and computational algorithms for cardiovascular health monitoring, particularly for patients with heart failure, and the development of an intelligent allocation method of diagnostic testing for COVID-19 using commercial wearables.

Integrating AI with Care: Lessons Learned in Public Health

Date: March 7, 2023 | 12:00PM-1:00PM, BMI Classroom 4004 and on Zoom

Speakers: Azra Ismail

Abstract:

There has been growing interest in applying Artificial Intelligence (AI) in public health, motivated by scarce and unequal access to care in communities globally. Many of these efforts promise improved health outcomes but also risk amplifying existing inequities. This calls for the human-centered design of AI to ensure alignment with the needs and experiences of communities and health workers targeted. This talk presents lessons learned from AI integration in public health, in the context of maternal and child care delivery in India. I will share three studies on the integration of machine learning and conversational agents with existing (1) data flows and practices, (2) organizational structures and goals, and (3) care workflows. My work centers care to draw attention to how AI is embedded in a broader ecology with multiple stakeholders, and tensions that may arise when targeting equity with AI. Finally, I chart a path forward for greater human-AI collaboration in healthcare, to improve care experiences for marginalized communities and workers.

Leveraging Machine Learning and Clinical Decision Support for Delirium Prediction

Date: March 2, 2023 | 1:00PM-2:00PM

Speakers: Siru Liu, PhD

Abstract:

This talk will introduce a three-stage process to develop and implement a clinical decision support tool to predict new onset delirium in hospitalized adult patients. The tool used explainable machine learning and deep learning with an AUC of 0.927 and 0.952, respectively. In the second stage, the tool was developed through a user-centered design and evaluated for usability. In the final stage, the tool was implemented on an EHR vendor and evaluated through clinician interviews and questionnaires. The tool has the potential to reduce delirium by identifying high-risk patients, reducing the frequency of neurological assessments for low-risk patients, and assisting in discharge or transfer.

Multimodal Screening Algorithms for Mitigating the Burden of Delayed Start of Care and the Risk of Negative Outcomes

Date: February 21, 2023 | 12:00PM-1:00PM

Speakers: Maryam Zolnoori, PhD

Abstract:

Delayed start of care is associated with a significant burden on patients, caregivers, and healthcare systems. Recognizing this burden, the National Institutes of Health introduced initiatives (e.g., screening algorithm) built on novel data science methods and data streams generated during routine clinical encounters as a solution with significant potential to improve the quality and safety of healthcare services and reduce healthcare costs. My program of research is dedicated to mitigating the burden of the delayed start of care through the use of multiple data science methods that take advantage of discrete data points from multiple sources (e.g., speech data, free-text clinical notes) for early identification of patients at risk of health deterioration and negative outcomes. In this talk, I present my journey in developing novel informatics solutions for mitigating delayed start of care, with a special focus on patients with mental health problems.

2022/2023 Winter Seminar Series

Contrasting Learning of Electrodermal Activity Representations for Stress Detection

Date: February 14, 2023 | 1:00-2:00pm

Speakers: Robert Lewis, PhD Candidate and Katie Matton, PhD Candidate

Abstract:

Electrodermal activity (EDA) is a biosignal that contains valuable information for health monitoring. Changes in EDA are related to sympathetic nervous system activity, which varies with changing levels of psychological arousal. Thus, a leading application of EDA is the estimation of stress and other related mental health constructs. However, wearable EDA measurements are challenging to analyze as the data tend to be noisy and sparsely labelled. In this talk, we will discuss our on-going work to address these challenges using contrastive learning approaches that are tailored to account for the specific properties of the EDA signal.

From Conventional Ophthalmology to Emerging Digital Ophthalmology

Date: December 13, 2022 | 1:00-2:00pm

Speakers: Siamak Yousefi, PhD

Abstract:

Large and complex imaging, visual field, and genetic datasets are becoming increasingly available to the eye and vision research community. However, analyzing such complex data is not trivial and typically requires advanced data mining and machine learning techniques to extract useful knowledge and making sense of findings. In this talk, I will briefly review applications of artificial intelligence models to automatically detect some of the ocular diseases. I will explore early and emerging artificial-intelligence-based models for forecasting, screening, diagnosis, and monitoring several retinal and corneal conditions with an emphasis on glaucoma as the second leading cause of blindness worldwide.

Toward Ubiquitous Intelligent Systems for Managing Neurological Disorders

Date: December 6, 2022 | 1:00-2:00pm

Speakers: Hyeokhyen Kwon, PhD

Abstract:

Neurodegenerative disorders, such as Alzheimer's disease or Parkinson's disease, negatively impact both patients and caregivers, leading to emotionally, financially, and physically exhausting experiences. In order to promote patients' active and independent management of these diseases, early detection and continuous monitoring of disease progression are crucial. The advancements in hardware and software frameworks for edge devices, along with AI/ML techniques utilizing on- and off-body sensors, have made it possible to continuously and passively quantify patients' physical, mental, and social behavior in both clinical and non-clinical settings. Through the application of ubiquitous intelligent systems, I will discuss the grand challenges in patient care and the latest advances in monitoring neurodegenerative disorders.

Wearable Sensors for Chronic and Infectious Disease Monitoring

Date: November 15, 2022 | 12:00-1:00pm

Speakers: Md Mabashir Hasan Shandhi, PhD

Abstract:

The majority of the health care costs related to the treatment of chronic and infectious diseases are attributed to direct care costs (e.g., hospital admissions and readmissions). The prevalence of chronic diseases and associated costs in the United States is growing at an alarming pace. The COVID-19 pandemic has further impacted the health of high-risk individuals by increasing the likelihood of more severe illness for those with underlying health conditions and associated healthcare costs. There have been ample efforts from researchers and clinicians to develop remote healthcare systems and wearable devices to manage patients with chronic and infectious diseases in home settings, which has reduced the burden on inpatient care facilities and gained further momentum during the COVID-19 pandemic. Yet, there is a lack of reliable wearable devices that can provide clinically acceptable information to healthcare professionals, as well as a lack of emphasis on validating technologies in representative populations to enable a reliable and equitable remote health management system. This talk will present the challenges and potential solutions for developing tools (i.e., wearable sensors and computational algorithms) for reliable and equitable remote patient monitoring systems for chronic and infectious diseases. More specifically, the presenter will share the development and validation of wearable sensors and computational algorithms for cardiovascular health monitoring, particularly for patients with heart failure, and the development of an intelligent allocation method of diagnostic testing for COVID-19 using commercial wearables.

Click here to view previous events.