We are having a database having physiological data of people (sytolic pressure. diastolic pressure, heart rate, cholesterol, smoking habits etc.).
- What can Machine Learning do to such a database?
- How to apply Machine Learning to such a database?
- What tools can be used
- Do we have some out-of-the-box codes for applying it to such a database?
Hello Parag,
First I want to tell you that having an accurate and large health related dataset is a huge asset in today's world, where more and more people are health conscious and there is a huge gap that tech has to fill. Now let me answer your questions one by one.
1. What can Machine Learning do to such a database ?/ 2. How to apply Machine Learning to such a database ?
You can take one column such as 'heart rate' as the output column and using the rest of the columns you can predict how cholesterol levels relate to heart rate etc.
3. What tools can be used ?/ 4. Do we have some out-of-the-box codes for applying it to such a database ?
Yes, there are a lot of tools available for machine learning. You can use sklearn(python), liblinear/libsvm(c++) etc.
Hope this helps.
Applying "machine learning" to a database just for the sake of applying it seems like a waste of time and resources. Scope creep can be problematic if one leaves the project too open-ended. There are probably hundreds of ways you could apply machine learning to your particular physiological database but what is the end goal? There are lots of factors to consider, which depending on what type of data you have and how many patients are in the database, can impact what you do. I would first define a research goal and hypothesis. In your case, the aim would hopefully have some clinical utility. Also before diving into that, one should know what type of sampling was used to generate the data in the database. Is it representative of the overall population or a particular sub-population? Is it biased towards some particular type of condition, etc? Once you have a clearly defined project aim and figured out the limitations of your data etc, you can refine your search of relevant machine learning approaches, apply that approach on your retrospective data-set and then validate it on a prospective data-set.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude