As I am not from a medical field so I just wanted to know the views from domain experts that whether blood tests are actually helpful in predicting the chances of getting infected with COVID19.
Hoping for a healthy discussion.
Thank You.
please check the following links:
https://www.sciencedaily.com/releases/2020/10/201014160518.htm
Article Detection of COVID-19 Infection from Routine Blood Exams wit...
Mostly final decision by the doctors is given by evaluating the HRCT test. Not sure about blood, however, linked shared by Mohammed Salih Mahdi is a reasonable document.
Mohammed Salih Mahdi, I have gone through the papers before and that's what intrigued me to know from the domain expert that is there any relation of COVID19 infection with patient's blood test?
Yes, with no worries,. see e.g., Article Detection of COVID-19 Infection from Routine Blood Exams wit...
Article Routine blood tests as a potential diagnostic tool for COVID-19
yes we can build a ML model but we need to keep in view the other important attributes related to covid. i also thought of building a model but suitable dataset is not available. however you need to work on dataset more i.e preprocessing step. thankyou
Dear Sumit Banik ,
The more samples you have for both healthy and infected people, the more these samples contain enough data, and with the more effective preprocessing and feature engineering you deploy, you can create whatever ML system you want. Otherwise, do not even try!!.
Some of the hot research topics now revolve around getting good performance while using minimal data. 'Minimal Data Analysis' means to learn more from less data!!.
With Regards,
Mohammed
Thanks, Mohamed Elhadad . Do you have any idea from where could I obtain the blood test COVID data for further study?
Dear Sumit Banik
I think it can be valid. Also, you can read and benefit from this article
Article Novel Feature Selection and Voting Classifier Algorithms for...
Dear Sumit Banik ,
Everyone is floundering, and no one really knows what is works the best. It might be hard to find such data publicly available because of its value. Imagine the profit a person could gain if being able to introduce such a system and the loss that the companies that provide those different expensive tests. Anyhow, you can search for data here (https://datasetsearch.research.google.com).
Regards,
Mohamed
Yes. You can build . Try to add other relevant attributes. Refer this article. https://www.sciencedirect.com/science/article/pii/S2352914820305992#sec3
Sumit Banik
We can most definitely build an ML model for this. In my opinion, we'll need to add more than blood tests to the attributes as Mir Junaid Rasool mentioned earlier.
Height, weight, BMI, past respiratory conditions, high blood pressure, hypertension even attributes such as locations (since it's being handled differently in different parts of the world), I would also recommend adding the diet and lifestyle attributes.
Thank You Joseph Magiya for your valuable insights. Will keep this in mind.
It is possible to build a ML model to predict chances of contradicting COVID-19 based patient's blood tests. You will however need to understand which attributes your model will use from the blood sample tests. Since, the COVID-19 virus is still new, you will need a large data set to test on the accuracy and tune your model to accurately predict chances of contradicting COVID-19.
- Badges
- Science topic