General
On Wednesday July 5th,VU Campus Center Artificial Intelligence & Health is hosting a research visit of Madalina (Ina) Fiterau, an assistant professor at the University of Massachusetts. She will give a research talk on AI and Health from 10:00 - 11:00. Location: NU-5A47. Please find more details on the talk below. Hope to see many of you there!
Title research talk
Modeling the Evolution of Chronic Diseases from Heterogeneous, Multimodal Data
Abstract
Chronic conditions such as congenital heart disease, Alzheimer's disease and osteoarthritis affect a significant segment of thepopulation. According to a survey by the European Commission, more than one-third (35.2%) of people in the EU reportedhaving a long-standing (chronic) health problem in 2021. Longitudinal studies that monitor subjects over extended periods oftime help determine the relationships between risk factors and disease evolution, which is essential in quantifying theeffectiveness of treatment and palliative care. The studies comprise multimodal data such as demographics, time series,medical images, and genetic information. All are collected across multiple institutions, multiple patient populations andmultiple visits. The collection process induces heterogeneity at all levels: there is high irregularity, inter-subject variability, andpotentially changing collection protocols. Reliable disease trajectory models, constructed through retrospective statisticalanalysis of this multimodal longitudinal data, are necessary to inform patients and facilitate clinical decisions.
We address the methodological gap by tightly integrating multimodal data and leveraging the different sources of information,including domain expertise, to extract salient features. In the Information Fusion Lab, we develop hybrid models that optimizemulti-component objectives, specialized to the task and for the available data. Our models include hybrid layers, designed tocope with multiple inputs of distinct types, such as attributes encoded as discrete features provided together with theirassociated images. In the talk, I will present mechanisms to conditionally route samples through the neural networksdepending on their cross-modal characteristics, models that leverage the intrinsic frequency of signals to learn sparseforecasting models from multivariate time series and weakly supervised deep learning architectures incorporating domain-specific heuristics. These techniques have enabled us to efficiently construct representations of images that adhere to specificpatterns, such as medical images of different organs. In the talk, I will demonstrate the performance of our models in attainingstate of the art results on tasks such as Alzheimer's disease forecasting, detecting heart conditions and in-hospital mortalityprediction.
Biography Madalina (Ina) Fiterau
Madalina (Ina) Fiterau is an Assistant Professor in theCollege of Information and Computer Sciences at UMass Amherst, leading theInformation Fusion Lab. Previously, she was a postdoc at Stanford University, having completed aPhD in MachineLearningfrom Carnegie Mellon University.Her current research is on hybrid models and on the development of new deep learning methodology to obtainsalient representations from multimodal biomedical data, including time series, text and images. She was awarded theMarr Prize for Best Paper at ICCV 2015, the Star Research Award at the Annual Congress of the Society of CriticalCare Medicine 2016 and theManning IALS Research Award in 2019.She is keenly interested in applying her ML research towards the advancement of healthcare.