Machine learning is one of the most trending technologies because of its applications like finance, healthcare, self-driving cars, speech recognization, virtual assistance, etc. Here, I will discuss how Machine learning can be applied to the healthcare domain. Many reasons are already known, like,
Tracking the performance of doctors, To detect Breast cancer,
To record the impact of medicines, etc. Right now, the world is fighting with the deadly virus COVID-19.
This story is all about the ML for COVID-19.
As per my thinking, Two applications of ML for COVID-19 can be,
The danger of disease by the past conditions of the patients and Risk of improving disease by geographical locations.
Here, let us take an example of the Predicting risk of developing COVID-19 in patients.
Machine Learning can be significant for predicting the patient’s health
conditions after some time. In advance patient care, doctors can count the risk factors, and accordingly, they can plan the precautions. As well as, they can prioritize the patients in risky conditions of mass infections.
In Coronavirus infection, it is highly required to go through the patient’s information to predict the risk of the infection or risk of death for any patient. Another application is developing the disease is from a geographical location. The risk of developing the disease is more when a person is coming from the clustered zones.
Now back to the other application that is predicting the risk of death of the patient.
If the patient had any serious disease in the past, then the patient may fall into the critical stage. Age is also a factor to affect the patient more because older selves have low immunity power. Let’s not drive into the diseases that affect the covid patients as it is out of the scope of this post. So, Here, it is required to prioritize the patients and give them the medications according to the risk.
Here, I simply want to explain what I am talking about. For the sake of explanation, I made a raw dataset to predict the condition of COVID patients in the future.
Note: The dataset is not related to the real world. I made this for only explanation purposes.
Here, the data set include the condition of the patient in the next 12 days after admitting. Which are classified into three classes
1. Dead
2. Stable
3. Critical
Attributes are given as follows which will lead to any of the 3 classes.
1. Age
2. HighBP
3. Cardiac
4. Diabetes
5. Liver
6. Cancer
7. None
So this is a labeled dataset. As we know, Supervised learning is used for the labeled dataset.
Supervised-learning: It is the type of machine learning where the dataset is labeled i.e the dataset also includes the outputs with inputs. Supervised learning will map the inputs to the outputs based on the given data. Here, the output data is the stage (after 12 days), and input data is other attributes.
In the real-world, this dataset can be generated by recording the medical data of past patients. This dataset then will be fed to the machine learning algorithm(Supervised algorithms), where training and testing of the data will be done. These algorithms will help to predict the condition of the patient in the future. Once the algorithm will become accurate, most of the problems of doctors will be automatically solved.
For example, If a particular patient’s condition is predicted critical by the ML algorithm, doctors can easily differentiate the stable patient and critical patient and can give extra care and the appropriate dose of the medicine to the critical patient.
As seen in the dataset, the output is categorical data. So, for that, I have used Logistic regression which is the algorithm of supervised learning. Any other algorithm can also be used that falls under supervised learning.
For a simple implementation of this problem where I have used the logistic regression, you can check my GitHub account-https://github.com/Farnaz08/MLforCOVID.git
