Ph.d Pham Dinh Ba– Toronto University, Canada
In developed countries, doctors are not supposed to treat patients base upon what they learned from medical schools.
Doctors should make treatment decisions based upon results of studies evaluating whether treatment with antibiotics for 5 days is as likely to reduce the infection as treatment for 7 days. This is because you do not want to treat the infection longer than needed. One of the problems is that there are too many researches studies. Looking for the right studies to help you treat patients is hard because of the number of research studies is increasing, with more than 75 High-quality studies per day (>25,000 high-quality studies per year).
Questions such as “What drug is most related to pancreatic cancer?” or “What symptoms are associated with malaria?” are relevant to the treatment of patients but to answer them, we need to search many thousands or tens of thousands of research studies. Humans can do the search and make sense of studies but it would take a long time, such as weeks, months or sometime years. The answer must be correct because we can’t treat patients with some wrong answers, can we? Computers can do it but they need to be trained on how to read and “understand” written text in research studies. The computer tool cui2vec will let you interact with over 108,000 medical concepts (e.g., treatment, antibiotics, cancer, malaria) using their “embedding’s”. (1) The embedding’s of a word are clusters of words that tend to appear together and they can be used to give the word some meaning, and as such, the computer uses them to “understand” what written text is about. For example, the embedding’s of the word “Ha Noi” may include “capital”, “Vietnam”, “Hoan Kiem” and so on. When the computer reads the word “Ha Noi” it uses the other embedding words to “understand” that this is the city name, and a capital with landmarks. The embedding’s of the computer tool cui2vec were created using insurance claims for 60 million Americans, 1.7 million full-text PubMed articles, and clinical notes from 20 million patients at Stanford. That’s a huge database of over hundreds of millions of words that scientists at the Biomedical Informatics at Harvard Medical School has curated and analyzed to create the cui2vec tool.
With the computer tool cui2vec, anyone can sort concepts by similarity to ask the tool some medical questions. You can use cui2vec to help answering questions such as “What drug is most related to pancreatic cancer?” or “What symptoms are associated with malaria?”. The information from the tool will let scientists understand diseases and health conditions to further develop medicine. The information also helps clinicians in part on how to take care of patients and to tell healthy people what they need to do to stay healthy.
Work such as the development of the computer tool cui2vec is a small part of the research and training in BioMedical Informatics (BMI), the interdisciplinary field that studies and pursues the effective uses of computers and biomedical data for scientific inquiry problem solving, and decision making, motivated by efforts to improve human health. An interdisciplinary field of study, BMI brings together health services researchers, data analysts, computer scientists, clinicians, and mathematicians, among others to solve problem, generally big and real-world problems.
Vietnamese students often have a strong background in mathematics. If you’re fascinated about applying your mathematics skills, solving problems and help taking care of sick people, this may be a promising study area for you. If you’re interested, some members of the Student Union could help answer question about this study area. At the Union, we believe in sharing learning experience, life experience and the future of Vietnamese students, as we know some of us may become good scientist.
Let’s go exploring the world!
Pham Dinh Ba