Projects

Statistical Prediction and Classification

Clinical Outcomes

Predicting clinical outcomes of ICU stratum patients utilizing retrospective datasets is a core focus of our work and drives the discovery, utility, and application of resulting models within hospital analytics platforms.

  1. Short and Long-term Survivability
  2. Neurological Status
  3. ICU and Hospital Length of Stay
  4. Discharge Location
  5. ICU and Hospital Readmission
  6. Improved APACHE

Disease Onset

Disease onset prediction involves data-driven understanding of patient characteristics that contribute to developing said disease during hospitalization. This enables early prediction of in-hospital ailments, providing clinicians sufficient time to provide interventions.

  1. Pulmonary Embolism
  2. Delirium
  3. Mechanical Ventilation (MV)
  4. MV Reintubation
  5. ICU and hospital readmission

Treatment Response

Understanding and predicting response to treatment/therapy is pivotal to creating patient specific precision care.

  1. Cardiac Arrest Treatments
  2. Traumatic Brain Injury Interventions
  3. Response to Medication
  4. Response to Procedures

Phenotype Discovery

Data-driven patient categorization

Unsupervised machine learning forcused on discovering clinically meaningful and data-driven physiologic sub-phenotypes to better categorize the heterogeneity within hospitalized populations.

  1. Out of Hospital Cardiac Arrest
  2. Moderate and Severe Traumatic Brain Injury
  3. Sepsis 3
  4. Acute respiratory distress syndrome

Optimizing clinical decision-making

Reinforcement Learning

Optimization of timing, dosing, and duration of medical interventions

  1. Medication Dose and Timing
  2. Surgical Treatments

Brain Mapping

Connectome Changes in Aneurysmal Subarachnoid Hemorrhage

Wearables

COMPUTATIONAL MOTOR SIGNATURES IN NEUROINTENSIVE CARE