Dementia outcomes through predictive modeling techniques

 
   Lead Investigator:    Amor Bhalla
   Institution      :    Arizona State Univeristy
   E-Mail           :    amol.bhalla@asu.edu
   Proposal ID      :    325



   Proposal Description:
   Our research is focusing on the behavior characteristics of the participants, specifically developing a novel approach for diagnosing cognitive decline by using machine learning and data mining approaches. A data mining approach requires a tremendous amount of data, so specifically, we would like the entire UDS data (28,444 patients with all their epochs) for the B9 form and the C1 form. The C1 form will be used (MSE values) to compare the model's predictive capabilities. Our study will then use data from the B9 form,  which is clinicians judgment of symptoms. The significance is that the B9 form contains information that includes memory complaints, cognitive symptoms, behavior symptoms, motor symptoms and overall summary of symptom onset. We will use the variables from the B9 form to develop a model that predicts a binary outcome of normal and abnormal cognitive decline. Our model's results will be compared to information provided by the C1 form that contains data of the Mental Status Exam. This will allow us to validate our model. The significance is that we want to determine features that may lead to a change from normal to abnormal, assuming that the B9 form contains participants that have transitioned from normal to abnormal diagnosis over several epochs.