SE PhD Prospectus Defense: Samad Amini

TITLE:   Transforming Dementia Diagnosis through AI

ADVISOR: Ioannis Paschalidis ECE, BME, SE, CDS

COMMITTEE: Rhoda Au BUMC,Vijaya Kolachalama BUMC, CDSPirooz Vakili SE, ME

ABSTRACT: Dementia, characterized by a progressive decline in cognitive function, poses significant challenges in our society. The development of reliable, affordable, and user-friendly strategies for dementia detection can play a vital role in enhancing clinical decision-making and improving dementia diagnosis.This prospectus focuses on developing predictive models based on deep learning techniques to identify dementia using diverse data modalities, including individual voice recordings, drawings, and physician notes. The prospectus begins by studying a cohort of 3423 individuals from the renown Framingham Heart Study (FHS). A deep Convolutional Neural Network is employed to analyze images derived from the Clock Drawing Test, enabling the assessment of dementia with high accuracy. Additionally, an automated screening system is developed utilizing Natural Language Processing (NLP) techniques to detect dementia from digital voice recordings obtained from neuropsychological exams of 1084 participants in the FHS. Furthermore, the prospectus leverages large language models to automate the cognitive impairment diagnosis process for 913 patients using physician notes in the Electronic Health Records data collected by the Massachusetts General Hospital.Through the integration of digital technologies, machine learning, NLP, and computer vision methodologies, this research addresses the pressing need for reliable and accessible dementia diagnosis.

When 11:00 am to 1:00 pm on Wednesday, July 5, 2023
Location 665 Commonwealth Ave, Room 1646