In-person social network(site, video, TEDx, press)
An average person spends two hours a day on social media, but almost half of us eat lunch alone. We created Connect to solve that problem. Since its creation in 2015, we have arranged nearly 10,000 platonic, in-person connections between students, alumni and staff at university campuses across the United States.
Coma prognostication(EMBC'15, NCS'16 (best poster), ACNS'17)
Cardiac arrest impacts over half a million people a year. Even if successfully resuscitated, patients can enter an indefinite coma. Predicting if patients will wake up from coma can prevent premature withdraw of care. We compiled the world's largest dataset of Electroencephelograms (EEG) from patients in coma after cardiac arrest, and are building algorithms that use EEG to rapidly predict coma outcomes.
Personalized medication dosing(ICM'14, EMBS'16, AAAI'18)
Mis-dosing sensitive medications can have dire consequences for patient care. We developed a personalized medication dosing algoithm that is robust to missing data, a common problem in intensive care settings. The approach was 29% more accurate than intensive care staff, and better able to distinguish outcomes than non-personalized models.
Digitizing paper records(BigData'17 (slides), code)
Clinical researchers, historians, educators and field researchers regularly capture data on paper spreadsheets. We built a tool that transcribes images of paper-based spreadsheets into electronic form. The open-source tool provides a generalized solution for spreadsheet transcription that maintains privacy, is up to 10 times faster and twice as cost effective as existing alternatives.
Determining the mood of a story(AAAI'17 (slides), press, video)
Inferring the emotional content of a story requires consideration of para-linguistic cues (e.g. pitch), linguistic content (e.g. vocabulary) and the physiological state of the narrator (e.g. heart-rate). We developed an algorithm that combines real-time auditory, text, and physiological signals to predict the mood (happy or sad) of people narrating a story.
Provider seniment and resource utilization(EMBC'18 CinC'15 (slides))
We used natural language processing to visualize the evolution of clinical language and sentiment with respect to several common population-level categories. We found a decrease in the complexity of language use over time for patients with poor outcomes. We also found greater positive sentiment for females, unmarried patients, and patients of African ethnicity.
Medical 'Big Data'(Scientific Data '16, data )
We provide a freely accessible, large, single-center database comprising information relating to patients admitted to critical care units at a large tertiary care hospital. Data includes vital signs, medications, laboratory measurements, observations and notes charted by care providers, and more. The database supports applications including academic and industrial research, quality improvement initiatives, and higher education coursework.
Biomarkers of schizophrenia(Front Hum Neurosci.'12)
Genetic factors contribute to the etiology of mental diseases, including schizophrenia. We developed a method to identify genetic variations associated with functional brain networks in schizophrenia. We found functional networks located in the thalamus, anterior and posterior cingulate gyri. The contributing genetic factors fell into two clusters centered at chromosome 7q21 and chromosome 5q35.
Mapping mental representations(Curr Biol. '13)
Humans develop rich mental representations that guide their behavior in a variety of everyday tasks. We developed a novel method to extract these complex representations through simple tasks. The extracted distributions allow us to predict the behavior of subjects to novel stimuli.
Skip a beat(App, press)
Copyright, Mohammad M. Ghassemi, 2017