You are on page 1of 7

IOT HOSPITAL MANAGEMENT SYSTEM AND

ANALYSIS WITH ACCESSING DATA FROM CLOUD


USING MACHINE LEARNING
S.Vigneshwaran 18BEE0359 K.Balakrishnan 18BEE0365 Abhishek raj-18BEE0342
Abstract— The purpose of this project is to figures, examining and creating outcome and has
use machine learning methods on health records revealed it’s position in many sectors. It needs a
to improve distribution in hospitals and sizable amount of information to give precise
healthcare. The all inclusive method for results and output. The accuracy of machine
application of ML algorithm can be used in learning is very dependable on on information
number of ways comprising neural networks can sharing and credibility of that information. The
be operated in experimental and organisational implementation of these two methods will give
management. Machine learning has a lot of precise output in machine learning technology. This
usefulness in perceiving patterns in information paper will outline on how converging these two
and figures, examining and creating outcome methods can help in medical sector. Machine
and has revealed it’s position in many sectors. It learning algorithm helps in analysing and giving
needs a sizable amount of information to give some analysis of the disease with the data available
precise results and output. The accuracy of to it.
machine learning is very dependable on
information sharing and credibility of that Linear Regression
information. The implementation of these two
methods will give precise output in machine The fitting of correlation of two different
learning technology. This paper will outline on variables in the experimental data through a linear
how converging these two methods can help in equation is attempted by linear regression. There
the medical sector. are two different types of variable, the first one is
the explanatory variable and the second one is
dependant variable. For example, a modeller might
Keywords— machine learning, health sector want to fit the income of employees according to
Literature Survey their gender through a model of linear regression.
Machine learning is a rapidly emerging area in
the medical field that is linked to the digitization of A modeller can first decide whether or not it is
medical data. It's also employed in the corporate a relationship between the variables of significance
world. The numerous uses of machine learning before trying to integrate a linear model to actual
have the ability to improve healthcare delivery observations. This does not actually mean that one
while simultaneously dramatically lowering costs. variable affects the other (for example, higher SAT
A number of literature studies have confirmed this, scores do not always imply higher college grades),
and a significant number of algorithms with but rather that the two variables have a substantial
positive results have been created. It has resulted in relationship. When calculating the intensity of a
the identification of a number of diseases and the relationship between two variables, a scatterplot
forecasting of epidemics. Collaboration between may be useful. If the suggested predictive and
hospitals and academic institutions is critical dependent variables tend to have no relationship
because it bridges the divide between scientific (i.e., the scatterplot shows no growing or declining
findings and actual use. trends), then applying a linear regression model to
the information will not provide a useful model.
Introduction The correlation coefficient, which would be a figure
The purpose of this project is to use machine among -1 and 1 representing the frequency of the
learning methods on health records to improve relationship of the measured data for the two or
distribution in hospitals and healthcare. The all more variables, is a useful numerical indicator of
inclusive method for applicaton of ML algorithm in relationship between two variables.
number of ways comprising neural networks can be
operated in experimental and organisational The expression for just a linear regression line is
management. Machine learning has a lot of Y= a + bX, with X as that of the explanatory
usefulness in perceiving patterns in information and variable and Y as the dependent variable.
and False for non-missing cells. And the missing
values be replaced to No.

Calculating the percentage of the women and


men in the dataset and then plotting it. It is
observable that there are more number of men in
the data sample than the number of women by a
Model slight margin.
We have collected and made our own dataset for
our project for hospital database. There are 795
individuals of great diversity with various ailments
and diseases. The diseases in the dataset are:
Cough, fever, cold, headache, covid, Bp, sugar,
diabetes, history of cancer, death, pregnent,
Delivery, accident, smoking or tobacco, heart
disease, alcoholism, food poisoning, ulcer, kidney
stone, kidney failure, asthma, skin disease, dental,
abscess, thrombosis, coma, cirrhosis, syphilis,
rheumatoid, rabies, malaria, leukemia, appendicitis,
jaundice, herpes, genetical candidiasis, gonorrhoea,
diphtheria, Alzheimer, disease, cholera
chikungunya, anaemia, hepatitis, chronic diseases,
tuberculosis, lipid disorder, patch sneezing, hyper
tension, HIV, injury or violence, mental health, flu
or pneumonia, cataract. With the help of Python3
and linear regression we are applying in our model.

Applying the function to get missing values and


know the number of missing values from each
variable. It will return True for missing components
heatmap for the dataset

The various number of pair plots for heart


disease, heart disease vs accident, age vs heart
disease, accident, accident vs heart disease, accident
vs age, Delivery, delivery vs death, delivery vs age,
death, death vs delivery, death vs age, age vs Correlation matrix in accordance with heatmap
delivery, age vs death.

Boxplot of pregnancy vs. weight

scatterplot for features and dependent variables


Matplotlib is a library for plotting data. To map
graphs of the results, I've imported pyplot. To
separate the data, we can use the train test Then, through using KNN algorithm,
split feature. The test size=() parameter in the determine an illness and how certain it would be to
occur in a given individual. Both classification and
method specifies how many data can be held
regression issues can be overcome using the KNN
for processing. Now we'll use the training data algorithm. The KNN algorithm predicts the values
to match the model. As you'll see, we're trying of new data points based on ‘feature similarities.'
to anticipate the test data by training a model This implies that a value is given to the new point
to the training samples. Let's look at (some of) depending on how similar it is to the points in the
the forecasts before plotting the training set. Since ID11's height and age are
identical to that of ID1 and ID5, we can assume that
his weight is also similar.

If it had been a classification problem, we


might would use the same approach as the final
prediction. In this case, we have two weight values:
72 and 77. Any guesses of how the final number
will be calculated? The number of the values is
used to make the final forecast.

The distance between the new point and


model: each training point should be calculated first. There
are many methods for measuring this distance, the
most well-known of which are: – Euclidian,
Manhattan and Hamming distance (for categorical).
Result
We are successfully able to implement our
model to predict the occurrence of a particular
disease with a fairly high accuracy. Similarly, we
can implement this model for prediction of different
type of diseases in the dataset.

Conclusion
The website and prediction model will be used to
apply ML processes in a comprehensive manner,
allowing a health system to not only leverage
current study outcomes but also do its own research
at the lowest possible cost. It aspires to be an
appealing option in terms of both quality and cost
effectiveness, given the simplicity and cost benefit
of cloud computing offered as a subscription-based
service.

References
The next move is to choose the nearest points after [1] N. Satoshi (2008), Bitcoin: A Peer-to-Peer
measuring the difference between a new Electronic Cash System, [Online]. Available:
observation and the points in our training range. https://bitcoin.org/bitcoin.pdf
The value of k determines the number of points to
[2] Matthias Mettler, “Blockchain Technology in
be considered.
Healthcare”, 2016 IEEE 18th International
Conference on e-Health Networking, Applications
and Services (Healthcom)
[3] Freya Sheer Hardwick, Apostolos Gioulis, Raja
Naeem Akram, and Konstantinos Markantonakis,
E-Voting with Blockchain: An E-Voting Protocol
with Decentralisation and Voter Privacy

[4] Nakamoto, S., 2012. Bitcoin: A peer-to-peer


electronic cash system, Oct,2008

[5] Pinyaphat Tasatanattakool and Chian


Techapanupreeda, “Blockchain: Challenges and
Applications”

[6] Francisca Adoma Acheampong, Big Data,


Machine Learning and the BlockChain Technology:
An Overview, International Journal of Computer
Applications (0975 - 8887) Volume 180 - No.20,
March 2018.

[7] G. Dwyer. “The economics of Bitcoin and


similar private digital currencies,” Journal of
Financial Stability, vol. 17, pp. 81-91, April 2015.

You might also like