Hello and welcome to Glass Box Medicine!
What’s Glass Box Medicine?
Machine learning is beginning to transform and disrupt medicine. Computer vision, natural language processing, and other areas of data science have been applied successfully to real clinical problems, demonstrating that data scientists and clinicians can achieve impressive feats when they work together. At the same time, there are vast differences in knowledge, experience, and culture between these disciplines that can make collaboration difficult. This blog hopes to facilitate collaboration between medicine and data science by posting articles that make each field more intuitive and accessible to the other. I’ll be writing two types of articles:
- “About Medicine”: these articles will aim to make medicine less of a “black box” for data scientists, by describing what happens in the clinic and on the wards, including where data comes from and how model results might be used. Articles will also explain key medical concepts that are relevant to building models and collaborating with clinicians. Example topics: understanding medical abbreviations and jargon, basics of common diseases, interpretation of medical images
- “About Machine Learning”: these articles will aim to make machine learning and other data science topics less of a “black box” for clinicians by describing the model development and testing process, as well as key data science concepts relevant for collaboration with data scientists. Example topics: basic machine learning for medicine, data types/data preparation, capabilities and limitations of modern methods
Rachel Ballantyne Draelos: I am an M.D. & Computer Science Ph.D. student in the Duke Medical Scientist Training Program (MSTP). I started the MD/PhD program in fall 2014. I’m interested in machine learning methods development, particularly for medical imaging, genomics, electronic health records, and natural language.
I’m looking forward to writing articles for you! If you have any requests for future article topics, you can email me via the ‘Contact’ tab.