Disease Prediction Using Machine Learning
Disease Prediction Using Machine Learning
Disease Prediction Using Machine Learning
http://doi.org/10.22214/ijraset.2020.6077
International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.429
Volume 8 Issue VI June 2020- Available at www.ijraset.com
Abstract: Machine learning has various applications and one of them is healthcare. There should be much more advanced
medical facilities so as to provide the best possible treatment for the patients[3].Also there are many machine learning algorithms
(such as KNN, Random Forest and Decision Tree Classifier algorithms and many more) which were selected and on the given
data many algorithms were applied so as to produce the best results. We can say that when machine learning implemented in
healthcare can lead to a high increase in patient satisfaction. so this research paper, will try to implement functions of machine
learning in health facilities in a particular system[8]. Instead of directly performing treatment for the patient, if the disease is
predicted beforehand using certain machine learning algorithms then the entire process of treatment can be made much more
efficient[12]. There are also some cases which occur when early diagnosis of a disease is not performed or carried out. Hence
disease prediction is a really important step while treating the patient. As it is said “Prevention is better than cure”, the right
prediction of disease would definitely lead to an early prevention of that particular disease[19].
Index Terms: KNN, healthcare, Logistic Regression.
I. INTRODUCTION
Today healthcare industry has become a big money making business. The healthcare industry uses and produces quite a large amount
of data which can be used to extract information about a particular disease for a patient. This information of healthcare will
further be used for effective and best possible treatment for patient’s health. This area also needs some improvement by using the
informative data in healthcare sciences. But a major challenge is to extract the information from the data because the data is present
in a huge amount so some data mining and machine learning techniques are used. The expected result and of this project is to predict
the disease beforehand so that the risk of life can be prevented at an early stage and save life of people and the cost of treatment can
be reduced to a particular extent. In India also we should adopt the non-manual system of medical treatment which is the best for
improving and understanding the human health. The main motive is to use the concept of machine learning in healthcare to
improvise the treatment of patients. Machine learning has already made it much more easier to identify and predict various
diseases[7]. Predictive analysis of the disease with the help of many machine learning algorithms helps us to predict the disease
and helps in treating the patients in an effective manner. Disease prediction using machine learning also uses the patient history
and health data by applying vari- ous concepts like data mining and machine learning techniques and also some algorithm. Many
works have also applied data mining techniques to the pathological data for prediction of some particular diseases. These approaches
were intended to before- hand predict the re-occurrence of certain diseases[15]. Also, some approaches tried to do prediction while
controlling the disease. The recent work of deep learning was in disparate areas of machine learning which have driven a shift to
machine learning models that can learn and understand the hierarchical representations of raw data with some pre-processing. With
the development of this concept called big data technology, more attention is paid to disease prediction.
C. Logistic Regression
Logistic regression is a regression analysis to study when the dependent variable is binary in nature. Like all regression analyses,
logistic regression is also a predictive analysis. Logistic Regression is used when a particular variable or target is categorical in its
type. It uses maximum estimation as a method of approximation.
D. Decision Tree
Decision tree is a supervised learning algorithm which has a predefined target variable that is used in classification prob- lems. It
works for concepts like categorical and continuous input and output variables for the model. In this technique, we will split the
population of sample into some homogeneous sets based on most significant input variables.
E. Naives Bayes
It is basically a classification technique based on Bayes’ Theorem with some changes like as follows. In simple terms, a Naive
Bayes algorithm assumes that the presence of a certain feature in a class is not related to the presence of any other feature. For
example, a fruit may be called as an apple if it is red, round, and is somewhat about 3 inches in its diameter. But if these features
depend on each other then all of these features will independently contribute to the probability that this fruit is an apple only and that
is the reason why it is known as ‘Naive’.
F. Linear Discriminant Analysis algorithm
Linear Discriminant Analysis or Normal Discriminant Anal- ysis or Discriminant Function Analysis is a dimensionality reduction
algorithm or a technique which is commonly used for the supervised learning of various processes that support machine learning in
our particular model.
G. Random Forest
Random decision tree is a type of ensemble learning method for classification. They are used for correction for the habit of
overfitting of the training set.
H. Classification Problems
These are used for studying the differences in groups i.e. separating two or more classes. It is used to highlight the These are used for
studying the differences in groups i.e. separating two or more classes. It is used to highlight the features in higher dimensions of
space into a lower dimension of space.
LDA is represented in a simple manner which can be easily understood by everyone. The model comprises of the statistical
properties of our data that is calculated for each and every class of our model. The same properties are used and calculated for the
Gaussian for the case of multiple variables. Thus these problems are essential to be used.
Table II. Results Obtained Using Embedded Method For Feature Selection
Algorithms Accuracy Scores Standard Deviation No. Of Features
LOGISTIC REGRESSION 0.988790 0.003105 124
CART 0.963035 0.006168 61
KNN 0.960116 0.008465 52
NA¨IVE BAYES 0.969986 0.004774 52
SVM 0.956250 0.006451 52
LDA 0.950578 0.006168 77
RANDOM FOREST 0.808566 0.042578 52
A. Data Preprocessing
Machine learning always faces a challenge whenever there is an inadequate dataset. To handle the missing data is an important step
to make sure that the models and algorithms of machine learning produce the most accurate results. This includes creating a
category object, fitting the encoder of the system to the ‘prognosis’ column present, and then to apply the encoder in our column to
transform these categories into final integers. We also handled the missing values of the columns with less than 50.
V. RESULT
The main aim of was to understand and improve the process of disease prediction and do a comparative study of algorithms in order
to find the best suited algorithm. Apart from the comparison of accuracy scores of algorithms, data visualization was also done so as
to understand the data and its trends in detail. The final comparison of the results obtained by the different algorithms employed for
prediction of disease prediction from hospital data, revealed that CART model or simply said, decision tree gave the highest
performance, followed by logistic regression, KNN, Na¨ıve Bayes, SVM, Random Forest, and LDA. RFE method of feature
selection even though employed on only two models, made a major impact. The second preferred feature selection method was
embedded method, whereas Pearson Correlation made the least impact. While analyzing the results we must also keep in mind the
extent by which the features were reduced to obtain a certain level of accuracy. If we think from that perspective then embedded
method made a major difference by reducing the number of features from 132 to 52 and still obtaining relatively good accuracy for
all algorithms. Pearson Correlation is a less precise method for feature selection, hence its performance was lower than that of the
other two methods which have higher precision.
The final comparison of the results obtained by the different algorithms employed for prediction of disease prediction from hospital
data, revealed that Logistic Regression gave the high- est performance, followed by Naïve Bayes, CART or simply said, decision
tree, KNN, LDA,SVM, and Random Forest. RFE method of feature selection even though employed on only two models, made a
major impact. The second preferred feature selection method was embedded method, whereas Pearson Correlation made the least
impact. While analyzing the results we must also keep in mind the extent by which the features were reduced to obtain a certain
level of accuracy. If we think from that perspective then embedded method made a major difference by reducing the number of
features from 132 to 52 and still obtaining relatively good accuracy for all algorithms. Pearson Correlation is a less precise method
for feature selection, hence its performance was lower than that of the other two methods which have higher precision.
VI. CONCLUSION
With our proposed system, comparatively a good and higher accuracy is achieved. This is then used by researchers, physicians or
doctors in order to provide the best treatment and medical care for the patients. Hence machine learning when used in healthcare can
lead to an effective treatment and the patient is also well taken care of. Here we try to implement some of the functions of machine
learning in healthcare into our system. Instead of direct diagnosis, when a disease is predicted for a patient then machine learning is
implemented using certain machine learning algorithms and then healthcare can be made smarter and better. When we compare the
different algorithms used for disease prediction from our dataset and the output we expect we get the best accuracy with Logistic
Regression algorithm and KNN algo- rithm, whereas LDA algorithm had the lowest performance when compared to the other
algorithms. Machine Learning (ML) provides methods, processes and certain techniques that can help solving the issue of diagnostic
problems in a simpler and modernized variety of medical domains. ML is nowadays being used for the prediction and analysis of the
clinical works. ML is also currently being used for the process of data analysis, such as detection of errors in the dataset and for
dealing with incorrect data present in our system. It is a debatable topic that the perfect use and implementation of ML algorithms
can act as great source of help in the integration of computer systems in the field of healthcare to facilitate and enhance the work of
doctors and finally leading to improve the efficiency and quality of our medical care for the respective patients.
VII. FUTURE SCOPE
As nowadays we can clearly witness the increase in use of computers and technology to consider a huge amount of data, computers
are being used to perform various complex tasks with commendable accuracy rates. Machine learning (ML) is a collection of
multiple techniques and algorithms which permit computers to execute such complex tasks in an simplified manner. It is also used
in both academics which is for students or learners and also in industry to make accurate predictions and use these diverse sources
of dataset and information. Till date we can say we have grown in the fields of big data, Machine learning, and data sciences etc and
have been a part of one of those industries which were able to collect such data and the staff to transform their goods and services
in a desired manner. The learning methods developed for these industries and researches offer excellent potential to further
improvise medical research and clinical care for the patients in the best possible manner. Machine learning uses mathematical
algorithms and procedures which are used to describe the relationship between variables used in the model and the others. Our paper
will explain the process of training the model and learning a suitable algorithm to predict the presence of a particular disease from the
sample of the tissue based on its features. Though these algorithms work in different and unique manners depending on the way in
which they are developed and used by the researchers. One way is to consider their supreme goals. The goal of our paper and
statistical methods is to reach to a conclusion about the data which are collected from a wide variety of samples from our
population. Though many techniques, like linear and logistic regression, are able to predict the diseases. For example, consider a
case where, if we can create a model which described and understood the relationship between clinical variables and their
transience then we can follow the organ transplant surgery i.e. we would need the factors and features which differentiate low
mortality rate from high if we can develop such outcomes and reduce mortality rate to a desired rate in the near future also nothing
can be said to be better than such situations.
REFERENCES
[1] D. W. Bates, S. Saria, L. Ohno-Machado, A. Shah, and G. Escobar, “Big data in health care: using analytics to identify and manage high-risk and high-cost
patients,” Health Affairs, vol. 33, no. 7, pp. 1123–1131, 2014.
[2] K.R.Lakshmi, Y.Nagesh and M.VeeraKrishna, ”Performance comparison of three data mining techniques for predicting kidney disease survivabil- ity”,
International Journal of Advances in Engineering Technology, Mar. 2014.
[3] Mr. Chala Beyene, Prof. Pooja Kamat, “Survey on Prediction and Anal- ysis the Occurrence of Heart Disease Using Data Mining Techniques”, International
Journal of Pure and Applied Mathematics, 2018.
[4] Boshra Brahmi, Mirsaeid Hosseini Shirvani, “Prediction and Diagnosis of Heart Disease by Data Mining Techniques”, Journals of Multidisci- plinary
Engineering Science and Technology, vol.2, 2 February 2015, pp.164- 168.
[5] A. Singh, G. Nadkarni, O. Gottesman, S. B. Ellis, E. P. Bottinger, and
J. V. Guttag, “Incorporating temporal ehr data in predictive models for risk stratification of renal function deterioration,” Journal of biomedical informatics,
vol. 53, pp. 220–228, 2015..
[6] S. Patel and H. Patel, “Survey of data mining techniques used in healthcare domain,” Int. J. of Inform. Sci. and Tech., Vol. 6, pp. 53- 60,March 2016.
[7] M. Young, The Technical Writer’s Handbook. Mill Valley, CA: Univer- sity Science, 1989.
[8] Hanka, R., Harte, T.P., Dixon, A.K., Lomas, D.J., and Britton, P.D. ”Neural networks inthe interpretation of contrast-enhanced magnetic res- onance images of
the breast”. InProceedings of Healthcare Computing, Harrogate, UK, 275-283, 1996.
[9] Hau, D., and Coiera, E. ”Learning qualitative models of dynamic systems”. MachineLearning, 26, 177-211, 1997.
[10] . Ifeachor, E.C., and Rosen, K. G. (eds.) Proceedings of the International Conference onNeural Networks and Expert Systems in Medicine and Healthcare,
Plymouth, UK, 1994.
[11] Innocent, P.R., Barnes, M., and John, R. ”Application of the fuzzy ART/MAP andMinMax/MAP neural network models to radiographic image classification”.
ArtificialIntelligence in Medicine, 11, 241-263, 1997.
[12] Jankowski, N. “Approximation and classification in medicine with IncNet neuralnetworks”. In [38].
[13] Karkanis, S., Magoulas, G.D., Grigoriadou, M. and Schurr, M. “De- tecting abnormalities incolonoscopic images by textural description and neural networks”.
In [38].
[14] Karkanis, S., Galoussi, K. and Maroulis, D. “Classification of endoscopic images based ontexture spectrum”.
[15] Pouloudi, A. ”Information technology for collaborative advantage in health care revisited”.Information and Management, 35, 6, 345-357, 1999.
[16] Pranckeviciene, E. “Finding similarities between an activity of the different EEGs bymeans of a single layer perceptron”. In [38].
[17] Prentza, A. and Wesseling, K.H. ”Catheter-manometer system damped blood pressuresdetected by neural nets”. Medical and Biological Engi- neering and
Computing, 33, 589-595, 1995.
[18] Reategui, E.B., Campbell, J.A., and Leao, B.F. ”Combining a neural network with case-based reasoning in a diagnostic system”. Artificial Intelligence in
Medicine, 9, 5-27, 1996.
[19] Ridderikhoff, J. and van Herk, B. ”Who is afraid of the system? Doctors’ attitude towardsdiagnostic systems”. International Journal of Medical Informatics 53,
91-100, 1999.
[20] Ruseckaite, R. “Computer interactive system for ascertainment of visual perceptiondisorders”.
[21] Schurr, M. “The Role of Machine Learning Methods in Endoscopic Techniques”.
[22] Strausberg, J. and Person, M. ”A process model of diagnostic reasoning in medicine”.International Journal of Medical Informatics, 54, 9-23, 1999.36. Zupan, B.,
Halter, J.A., and Bohanec, M. ”Qualitative model approach to computerassisted reasoning in physiology”. In Proceedings of Intelligent Data Analysis in
Medicineand Pharmacology-IDAMAP98,
Brighton, UK, 1998.37
[23] M Chen, Y. Hao, K. Hwang, L. Wang, L. Wang, ”Disease prediction by machine learning over big data from healthcare communities”, IEEE Access, vol. 5, no.
1, pp. 8869-8879, 2017. B. Qian, X. Wang, N. Cao,
H. Li, Y.-G. Jiang, ”A relative similarity based method for interactive patient risk prediction”, Springer Data Mining Knowl. Discovery, vol. 29, no. 4, pp.
1070-1093, 2015.
[24] IM. Chen, Y. Ma, Y. Li, D. Wu, Y. Zhang, C. Youn, ”Wearable 2.0: Enable human-cloud integration in next generation healthcare system”, IEEE Commun, vol.
55, no. 1, pp. 54-61, Jan. 2017.
[25] Y. Zhang, M. Qiu, C.-W. Tsai, M. M. Hassan, A. Alamri, ”HealthCPS: Healthcare cyberphysical system assisted by cloud and big data”, IEEE Syst. J, vol. 11,
no. 1, pp. 88-95, Mar. 2017.
[26] . Qiu, K. Gai, M. Qiu, ”Optimal big data sharing approach for telehealth in cloud computing”, Proc. IEEE Int. Conf. Smart Cloud (Smart Cloud), pp. 184-189,
Nov. 2016.
[27] S. Leoni Sharmila, C. Dharuman, P. Venkatesan, ”Disease Classification Using Machine Learning Algorithms - A Comparative Study”, Interna-
tional Journal of Pure and Applied Mathematics, vol. 114, no. 6, pp. 1-10, 2017.
[28] Allen Daniel Sunny, Sajal Kulshreshtha, Satyam Singh, Srinabh, Mohan Ba, H Sarojadevi, ”Disease Diagnosis System By Exploring Machine Learning
Algorithms”, International Journal of Innovations in Engineer- ing and Technology (IJIET), vol. 10, no. 2, May 2018.
[29] Shraddha Subhash Shirsath, ”Disease Prediction Using Machine Learn- ing Over Big Data”, International Journal of Innovative Research in Science, vol. 7, no.
6, June 2018.
[30] Shubham Rathi, Mahesh Motwani, Manish Ahirwar ”Data-Driven Clin- ical Decision Support System for Medical Diagnosis and Treatment Recommendation”
International Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-8 Issue-11, September 2019