Experience

MGH Institute for Technology Assessment

• United States
• Hospitals and Health Care
• 1 - 100 Employee

• Machine Learning Intern - MGH/Harvard Medical School

  • Jul 2022 - Present

  The objective of this research paper was to assess how well meta-models, that use advanced ML techniques, can replicate the temporal outcomes of a complex simulation model. A meta-model is an approach to approximate complex and computationally intensive simulation with a statistical representation. In this project, we created a meta-model of the Opioid Policy Model(OPyM), which simulates the trajectory of the opioid overdose crisis in the United States. The meta-model was made from a Deep Neural Network(DNN). The DNN took 17 inputs and had 96 outputs - one output for each year(2002 - 2025) for each of the 4 categories: Total Opioid Overdose Deaths, Prevalence of Non-Medical use of Opioids, Prevalence of Opioid Use Disorder, and Prevalence of Illicit Opioid Use. After comparing the graphs of the metamodel and OPyM, we concluded that advanced neural networks can replicate the outcomes of complex simulation models. Show less



Ocho Cafe

• Westford, Massachusetts, United States

• Host

  • Jul 2021 - Aug 2021