[ONLINE] Ata Lab Seminar "Disease progression models"
Friday, May 29, 2020 06:10 pm
ATA Lab seminar
Disease progression models.
As humans, we like to predict all manner of things, not least of them being our physical health. And while there is much excitement in the world of artificial intelligence about the ever-improving accuracy with which we can predict things, we are often less concerned with domain-specific relevance and utility of the prediction. Simple binary questions such as “does this patient have disease X?” or even “will the patient acquire disease X in Y years?” have proven less interesting to basic researchers and health professionals than “how quickly will the patient’s health deteriorate?”, “when and in what order will future symptoms appear?” and “what are the connections among the observable biomarkers and between biomarkers and symptom onset?”
Disease progression models (DPMs) attempt to answer the more interesting questions. In this talk, we will focus on applications of DPMs to brain imaging and neurodegenerative disease. I will go over some recent imaging-DPM developments from simple Bayesian models describing temporal biomarker order to differential equation models linking brain structure, connectivity and prior neurobiological knowledge to dynamically predict the course of an individual’s neurodegeneration.
Conference ID: 812 6896 9014