Introduction to Proposal:
Acute inflammation is a quick response to injurious bacteria that cause to deliver mediators of host defence leukocytes and plasma proteins to the particular injured part of the body. There are several components which cause acute inflammation but the three main components are
- Alterations in vascular calibre: increase blood flow/pressure.
- Circulation: structural changes that allow plasma protein to leave the circulation.
- Emigration: accumulation in the injury and their activation from the microcirculation.
The two main disease caused by the acute inflammation are
Nephritis is the expansion of one or both of the kidneys. The organs that filter the blood in kidney and get rid of excess liquid and chemicals. It can be caused by different conditions. The symptoms may develop as the Nephritis get worse but in about 60% of adult and 89% of children it comes and goes with only a small disturbance in their life. The symptoms for Nephritis are normally headache, visual disorder, high blood pressure, urine with blood traces, and the most common is the reduction in the urine volume and loss of appetite
It is also called as inflammation of urinary bladder. It can be cause by the infected bacteria that ascend urethra to the bladder. The symptoms unlike nephritis develop very quickly and about 2 in 5 women and 1 in 5 men are at the risk of this disease and develop urinary tract infection. The symptoms for Cystitis are normally frequent urination, and often it feels like burning while urination and the person feels cramps after urination. But it can be avoided by drinking excess of liquid and avoid irritants as perfumed soap. If cystitis left untreated it may develop into scarring and small stones which create problems while urination. The temperature rise but not above 38C. It also may contain blood particle. The treatment may take several weeks and there are chances the disease may develop again. One should expect the disease might turn into lengthened form.
Machine Learning problem:
The main concept of this data set is to create a algorithm for the expert system which perform the diagnosis of the two diseases discussed above i-e Nephritis and Cystitis in urinary system. It will be the example to diagnosing the cystitis and nephritis and the better understanding of both the disease is just been discussed above.
This data set was first created by a medical expert team as a data set to check the expert system which then perform the diagnosis of the two disease in urinary system. The data set in this particular assignment is been picked randomly from different sources for the practical experiment of the problem in machine learning for this assignment. The first research done on this particular disease is done by J.Czerniak, in 2002 9th international conference proceedings and is been published by Kluwer academic publishers in 2003.
Machine learning is a new concept to solve the complex problems and to make decision in short span of time based on the data taken earlier it helps to narrow down the problem and in the same way it help to get to know the solutions. The machine learning algorithms prove to be quite efficient for the doctors to decide whether or not the person suffers from a particular disease and also to suggest them the best possible cure for it. In this particular problem the doctor ask some question and then based on the answer he then compare it with the dataset he had with him and then made a decision.
INTRODUCTION TO WEKA:
Weka is a combination of machine learning algorithms which help to solve the complex data mining problems. It is developed in java and can be run on any platform using java virtual machine. Weka contain tools for data classification, association rule, visualization clustering and regression. It is also helpful for developing new algorithm in machine learning. This particular software is widely used in educational institutions to analyze the data in different rules.
DATA COLLECTION ANALYSIS:
This particular data is about the disease which is very common in children and sometime in adults. The data is been collected from internet and the link is been added as a reference. The main concept behind this data is to collect and prepare a algorithm for the expert system so that it can be helpful for future use in research and practical life. After running the algorithm it will help the medical expert as what disease the person is actually suffering from, and help to suggest the patient the correct medicine. For this assignment i have picked 100 instance and all the result and implementation are based on these. But it might not the right analysis as the data set is not large the bigger the data set the chances of getting the right solution is higher as the algorithm works good for the large amount of data more accurately.
LIST OF ATTRIBUTES:
I have use the following attributes in my data
- Lumbar pain.
- Urine Pushing.
- Micturition pains.
- Urethra Outlet.
- Urinary Bladder.
- Pelvis Origin.
This attribute is of numeric type and it keep the record of the patient temperature. For example 37.8
This attribute is of nominal type and it keep the record of sickness occurrence. It keep the record in yes or no form.
This attribute is of nominal type and it keep the record of lumbar pain in yes or no format.
This attribute is of nominal type and it keep the record of continuous urination in yes or no form.
This attribute is of nominal type and it keep the record of the pain while urination. It keep the data again in yes or no format.
This attribute is of nominal type and it keep the record of the burning while urination. It also keep the record in yes or no form
This attribute is the decision making attribute which tells the disease of a person and based on this is prescribed for medicine. It is a nominal attribute and keep the record in yes or no format.
This attribute is also a second decision making attribute which further illustrate the problem and help to narrow it down. It is also a nominal attribute and keep the record in yes or no format.
Further I would like to explain it in more detail as this is a classification problem and the attributes are categorical and integer and I have used C4 algorithm (J48) for this data set. A decision tree is a predictive machine learning tool which decide the target value of a new data set. The internal node of a decision tree donates to different attributes and the branches of the node tell us the value that attribute can have in the sample. And while in the end it tell us the decision of the dependent instance.
The J48 decision tree classifier follows the simple algorithms. It first creates a decision tree based on the values of the attributes in the dataset. And if there is any value of which there is no uncertainty that is for which the data instance falling in its category have the same value for the target instance, then it terminate that branch and assign to it the target value that it have obtained.
IMPLEMENTATION OF THE ALGORITHM IN WEKA:
In fig 2 we have a result of classification using algorithm j48. As shown in the figure we have 100 instances with correctly Instances are 100 and Incorrect Instances are 0.
From fig 3 as we can see the tree which illustrate the problem so that its easy to decide what is the actual problem. The tree are very much efficient to represent the rules as it use the previous data and can be wrong sometime and so they cannot guarantee the 100% correct result for the real world.
As in our tree if the temperature is greater or equal to 37.9 then it's not a urinary bladder problem (Cystitis) which is 60.0 out of 100. And if the temperature is less than 37.9 and had a lumbar pain as well then he is suffering from pelvis origin problem which is 34.0 out of 100 and if he is not suffering from lumbar pain then he is suffering from urinary bladder problem which is 6.0 out of 100 instances.
- Machine learning, Tom M. Mitchell, McGraw-Hill, ISBN: 0070428077
- Data Mining, Second Edition, Ian H. Witten & Eibe Frank, ISBN: 0-12-088407-0