A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 01 | Jan 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 662
A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC
KIDNEY DISEASE
PRASANTH.S1, PRATHAP.S2, HARISH KUMAR A.S3
1-3DEPARTMENT OF COMPUTER SCIENCE AND ENGINNERING, BANNARI AMMAN INSTITUTE OF TECHNOLOGY
ERODE, TAMILNADU, INDIA
-----------------------------------------------------------------------***--------------------------------------------------------------------
ABSTRACT: Persistent kidney infection (CKD) is a worldwide medical issue with high grimness and death rate, and it
initiates different illnesses. Since there are no prominent incidental effects during the starting periods of CKD, patients
consistently disregard to see the sickness. Early disclosure of CKD enables patients to seek ideal treatment to improve the
development of this contamination. AI models can effectively assist clinicians with achieving this target in view of their
speedy and exact affirmation execution. In this appraisal, we proposean KNN and Logistic relapse framework for
diagnosing CKD. The CKD informational index was got from the University of California Irvine (UCI) AI store, which has
countless missing attributes.
KNN attribution was used to in the missing characteristics, which picks a couple of complete models with the most
relative assessments to deal with the missing data for each divided model. Missing characteristics are by and large found,
in light of everything, clinical conditions since patients might miss a couple of assessments for various reasons. Later
enough balancing the divided instructive file, six AI estimations (key backslide, unpredictable woodlands, maintain vector
machine, k-nearest neighbor, unsuspecting Bayes classifier and feed forward neural association) were used to set up
models. Among these AI models, sporadic forest achieved the best execution with 99.75% end accuracy. By separating the
misjudgements delivered by the set up models, we proposed a joined model that solidifies determined backslide and
unpredictable woods by using perceptron, which could achieve a typical precision of 99.83% later on various occasions of
re-sanctioning. Thusly, we estimated that this way of thinking could be proper to more perplexed clinical data for ailment
finding.
INTRODUCTION
Their investigations have accomplished great outcomes
in the determination of CKD. In the above models, the
mean ascription is utilized to ll in the missing qualities
and it relies upon the demonstrative classifications of the
examples. Subsequently, their strategy couldn't be
utilized when the demonstrative consequences of the
examples are obscure. Actually, patients may miss a few
estimations for different reasons prior to diagnosing.
Furthermore, for missing qualities in clear cut factors,
information acquired utilizing mean ascription may have
an enormous deviation from the real qualities. For
instance, for factors with just two classes, we set the
classifications to 0 and 1, however the mean of the
factors may be somewhere in the range of 0 and 1.
fostered a dependent on highlight choice innovation, the
proposed models decreased the computational expense
through include determination, and the scope of
exactness in those model was from 97.75%-98.5%
CHRONIC KIDNEY DISEASE
Ongoing kidney illness (CKD) is a kind of kidney sickness
wherein there is continuous loss of kidney work over a
time of months to years. At first there are for the most
part no indications; later, manifestations might
incorporate leg enlarging, feeling tired, heaving, loss of
hunger, and disarray. Intricacies incorporate an
expanded danger of coronary illness, hypertension, bone
sickness, and anemia. Causes of persistent kidney
infection incorporate diabetes, hypertension,
glomerulonephritis, and polycystic kidney illness.
Hazard factors incorporate a family background of
persistent kidney illness. Determination is by blood tests
to quantify the assessed glomerular filtration rate
(eGFR), and a pee test to gauge egg whites .Ultrasound or
kidney biopsy might be performed to decide the hidden
reason. A few seriousness based arranging frameworks
are being used. Screening in danger individuals is
suggested.
MACHINE LEARNING
AI (ML) is the investigation of PC calculations that work
on naturally through experience. It is viewed as a subset
of man-made reasoning. AI calculations assemble a
model dependent on example information, known as
"preparing information", to settle on expectations or
choices without being unequivocally modified to do so.
Machine learning calculations are utilized in a wide
assortment of uses, for example, email separating and PC
vision, where it is troublesome or impractical to foster
regular calculations to play out the required tasks. A
subset of AI is firmly identified with computational
measurements, which centers around making forecasts
utilizing PCs; however not all AI is factual learning. The
investigation of numerical streamlining conveys
strategies, hypothesis and application spaces to the field
of AI. Information mining is a connected field of study,

zeroing in on exploratory information investigation
through solo learning. Machine learning includes PCs
finding how they can perform assignments without being
expressly modified to do as such. It includes PCs gaining
from information gave with the goal that they complete
specific errands.
KNN IMPUTATION
KNN Imputer by scikit-learn is a broadly utilized
technique to credit missing qualities. It is broadly being
seen as a swap for customary ascription methods. In this
day and age, information is being gathered from various
sources and is utilized for examining, creating
experiences, approving hypotheses, and so forth. This
information gathered from various assets may regularly
have some data missing. This might be because of an
issue in the information assortment or extraction
process that could be a human mistake. Managing these
missing qualities, in this way turns into a significant
stage in information pre-handling. The decision of
strategy for ascription is vital since it can fundamentally
affect one's work. A small bunch of writing in
measurements manages the wellspring of missing
qualities and ways of defeating the issue. The most ideal
way is to ascribe these missing perceptions with an
expected worth. In this article, we acquaint an aide with
attribute missing qualities in a dataset involving upsides
of perceptions for adjoining important informative
elements. For this, we utilize the extremely famous
KNNImputer by scikit-learn k-Nearest Neighbors
Algorithm. Missing esteems in a dataset can be a hornet's
home for any information researcher. Factors with
missing qualities can be a non-insignificant issue as there
is no path of least resistance to manage them. By and
large, assuming the extent of missing perceptions in
information is little comparative with the all out number
of perceptions, we can essentially eliminate those
perceptions. In any case, this isn't the regularly case.
Erasing the lines containing missing qualities might
prompt splitting away with helpful data or examples.
RELATED WORK
The information related to the endeavor what's more,
gains the characteristics of the relating plan . This
development can achieve definite and commonsense
examinations of afflictions; consequently, it might be a
promising method for diagnosing CKD.
The current framework predicts the persistent
sicknesses which are for a specific locale and for the
specific local area. Just specific illnesses are anticipated
by this framework. In this System, Big Data and CNN
Algorithm is utilized for Disease hazard forecast. For S
type information, the framework is utilizing Machine
Learning calculation i.e Decision Tree, Naïve Bayesian.
The precision of the current System is up to 94.8%.
In the current work, they smooth out AI calculations for
the viable forecast of ongoing illness episode in infection
regular networks. They try different things with the
changed forecast models over reallife emergency clinic
information gathered from focal China. They propose a
convolutional neural organization based multimodal
infection hazard forecast (CNN-MDRP) calculation
utilizing organized and unstructured information from
the clinic. It has gotten one more kind of clinical
instrument with the improvement of information
development what's more, has a far reaching application
prospect taking into account the quick improvement of
electronic prosperity record .
Md Murad Hossain, et al., has proposed in this work
kidney is an anisotropic organ, with higher versatility
along versus across nephrons. The level of mechanical
anisotropy in the kidney might be demonstratively
applicable if appropriately took advantage of;
nonetheless, if inappropriately controlled, anisotropy
might puzzle solidness estimations. The reason for this
review is to exhibit the clinical achievability of Acoustic
Radiation Force (ARF) instigated top relocation (PD)
measures for both taking advantage of and deterring
mechanical anisotropy in the cortex of human kidney
allografts, in vivo. Approval of the imaging techniques is
given by pre-clinical examinations in pig kidneys, in
which ARF-instigated PD esteems were measurably
altogether higher (p0.01). Comparable outcomes were
exhibited in vivo in the kidney allografts of 14 patients.
The symmetric ARF delivered PD measures with no
measurably critical contrast (p>0.01) between along
versus across arrangements, however the awry ARF
yielded PD proportions that stayed steady north of a six-
month perception period posttransplantation,
predictable with stable serum creatinine level and pee
protein to creatinine proportion in a similar patient
populace (p>0.01). The aftereffects of this pilot in vivo
clinical review recommend the plausibility of: 1)
carrying out even ARF to hinder mechanical anisotropy
in the kidney cortex when anisotropy is a frustrating
component, and 2) executing deviated ARF to take
advantage of mechanical anisotropy when mechanical
anisotropy is a possibly important biomarker. [1].
Erlend Hodneland, Eirik Keilegavlen et al., has proposed
in this work Chronic kidney sickness is a genuine ailment
described by continuous misfortune in kidney work.
Early discovery and conclusion is compulsory for
prognostic improvement. Thus, in the flow work we
investigate the utilization of picture enrollment
strategies for recognizing obsessive changes in patients
with persistent kidney sickness. Strategies: Ten sound
volunteers and nine patients with assumed persistent
kidney sickness went through powerful T1 weighted
imaging without contrast specialist. From genuine and
reenacted dynamic time series, kidney disfigurement
fields were assessed utilizing a poroelastic misshapening

model. From the misshapening fields a few quantitative
boundaries reflecting strain inclinations, and volumetric
and shear distortions were registered. Eight of the
patients additionally went through biopsy as a best
quality level. Results: We observed that the outright
twisting, standardized volume changes, just as strain
inclinations corresponded altogether with
arteriosclerosis from biopsy appraisals. Besides, our
outcomes show that current picture enlistment
approaches are inadequate with regards to affectability
to recuperate gentle changes in tissue firmness. End:
Image enrollment applied to dynamic time series ought
to be additionally investigated as an instrument for
intrusive estimations of arteriosclerosis.[2].
Gabriel R. Vásquez-Morales, Sergio M. Martínez-
Monterrubio et al., has proposed in this work presents a
neural organization based classifier to foresee whether
an individual is in danger of creating persistent kidney
infection (CKD). The model is prepared with the segment
information and clinical consideration data of two
populace gatherings: from one perspective, individuals
determined to have CKD in Colombia during 2018, and
on the other, an example of individuals without an
analysis of this infection. When the model is prepared
and assessment measurements for characterization
calculations are applied, the model accomplishes 95%
precision in the test informational index, making its
application for infection guess doable. In any case, in
spite of the exhibited productivity of the neural
organizations to anticipate CKD, this AI worldview is
hazy to the master with respect to the clarification of the
result. Ebb and flow research on eXplainable AI proposes
the utilization of twin frameworks, where a discovery AI
strategy is supplemented by another white-box
technique that gives clarifications about the anticipated
qualities. Case-Based Reasoning (CBR) has ended up
being an ideal supplement as this worldview can track
down logical cases for a clarification as a visual
demonstration avocation of a neural organization's
forecast. [3].
Njoud Abdullah Almansour, Hajra Fahim Syed et al., has
proposed in this work plans to aid the avoidance of
Chronic Kidney Disease (CKD) by using AI methods to
analyze CKD at a beginning phase. Kidney illnesses are
messes that disturb the typical capacity of the kidney. As
the level of patients impacted by CKD is essentially
expanding, compelling forecast techniques ought to be
thought of. In this work, we center around applying
diverse AI arrangement calculations to a dataset of 400
patients and 24 ascribes identified with analysis of
ongoing kidney infection. The grouping methods utilized
in this review incorporate Artificial Neural Network
(ANN) and Support Vector Machine (SVM). To perform
tests, all missing qualities in the dataset were supplanted
by the mean of the comparing credits. Then, at that point,
the streamlined boundaries for the Artificial Neural
Network (ANN) and Support Vector Machine (SVM)
methods were dictated by tuning the boundaries and
playing out a few analyses. The last models of the two
proposed procedures were created utilizing the best-got
boundaries and elements. [4].
Diego Buenaño-Fernández , David Gil et al., has proposed
in this work present work proposes the use of AI
methods to anticipate the last grades (FGs) of
understudies dependent on their authentic execution of
grades. The proposition was applied to the authentic
scholarly data accessible for understudies selected the
PC science certification at an Ecuadorian college. One of
the points of the college's essential arrangement is the
advancement of quality schooling that is personally
connected with manageable improvement objectives
(SDGs). The utilization of innovation in instructing
learning processes (Technology-improved learning)
should turn into a vital component to accomplish the
target of scholarly quality and, as an outcome, upgrade
or advantage the benefit of all. Today, both virtual and
eye to eye instructive models advance the use of data
and correspondence innovations (ICT) in both educating
learning cycles and scholastic administration processes.
This execution has created an over-burden of
information that should be handled appropriately to
change it into important data valuable for every one of
those engaged with the field of instruction. Foreseeing an
understudy's presentation from their verifiable grades is
quite possibly the most famous utilizations of instructive
datum mining and, accordingly, it has turned into a
significant wellspring of data that has been utilized for
various purposes. [5].
PROPOSED METHODOLOGY
The ckd dataset is given as information which comprise
of various attributes.Removal of undesirable information
and unkown credits are done in preprocessing
Featrueseleciton/trait determination is finished.
Grouping execution is done in calculations like NB, DT,
KSTAR, LOGISTIC, SVM. Accuracy, review, f-measure,
exactness will be classified. Those boundaries will be
displayed in type of graphical portrayal.
They utilized picture enlistment to perceive renal
morphologic changes and set up a classifier reliant upon
neural association using huge extension CKD data, and
the precision of the model on their test data.
Additionally, most of the past looks at utilized the CKD
educational record that was obtained from the UCI AI
store. This work explores how CKD can be analyzed by
utilizing AI (ML) methods. ML calculations have been a
main thrust in identification of irregularities in various
physiological information, and are, with an incredible
achievement, utilized in various arrangement
undertakings. In the current review, various diverse ML
classifiers are tentatively approved to a genuine
informational collection, taken from the UCI Machine

Learning Repository, and our discoveries are contrasted
and the discoveries revealed in the new writing. The
outcomes are quantitatively and subjectively examined
and our discoveries uncover that the Logistic relapse
(LR) classifier accomplishes the close ideal exhibitions
on the recognizable proof of CKD subjects. Thus, we
show that ML calculations serve significant capacity in
determination of CKD, with acceptable heartiness, and
our discoveries recommend that LR can likewise be used
for the analysis of comparable diseases. Their
assessments have achieved incredible results in the
finding of CKD. In the above models, the mean
attribution is used to fill in the missing characteristics
and it depends upon the illustrative groupings of the
models. Subsequently, their procedure couldn't be used
exactly when the decisive outcomes of the models are
dark. In actuality, patients might miss a couple of
assessments for various reasons preceding diagnosing.
DATA PROCESSING
Information handling, control of information by a PC. It
incorporates the change of crude information to
machine-decipherable structure, stream of information
through the CPU and memory to yield gadgets, and
organizing or change of result. Standardization of
undesirable information is done in this process. Each
downright (ostensible) variable was coded to work with
the handling in a PC. For the upsides of rbc and pc,
ordinary and unusual were coded as 1 and 0, separately.
For the upsides of pcc and ba, present and not present
were coded as 1 and 0, separately. For the upsides of htn,
dm, lowlife, pe and ane, yes and no were coded as 1 and
0, individually. For the worth of appet, great and poor
were coded as 1 and 0, separately. Albeit the first
information portrayal denes three factors sg, al and su as
all out kinds, the upsides of these three factors are as yet
numeric based, hence these factors were treated as
numeric factors. Every one of the downright factors were
changed into factors. Each example was given an
autonomous number that went from 1 to 400. There is
an enormous number of missing qualities in the
informational collection, and the quantity of complete
occurrences is 158. As a general rule, the patients may
miss a few estimations for different reasons prior to
making a determination. Consequently, missing qualities
will show up in the information when the indicative
classes of tests are obscure, and a relating ascription
strategy is required.
FEATURE SELECTION
Include determination dependent on credits (age , sexual
orientation., and so forth,) . Feature choice is the
method involved with decreasing the quantity of
information factors when fostering a prescient model. It
is attractive to lessen the quantity of information factors
to both diminish the computational expense of
displaying and, sometimes, to work on the presentation
of the model. Separating highlight vectors or indicators
could eliminate factors that are neither valuable for
expectation nor identified with reaction factors and in
this way forestall these irrelevant factors the models to
make an exact forecast. Here in, we utilized ideal subset
relapse and LR to extricate the factors that are generally
significant to the forecast. Ideal subset relapse identifies
the model presentation of all potential blends of
indicators and chooses the best mix of factors. LR
distinguishes the commitment of every factor to the
decrease in the Gini list. The bigger the Gini file, the
higher the vulnerability in grouping the examples.
Subsequently, the factors with commitment of 0 are
treated as repetitive factors. The progression of element
extraction was run on each total informational index The
mixes are positioned from left to right by the degree The
upward pivot addresses factors. The even pivot is the
changed r-squared which addresses how much the blend
of factors clarifies the reaction variable.
Classification performance
We utilize the AI calculation like Naïve Bayes , Decision
Tree, Kstar , Logistic Regression, Svmshow the order
performance. Logistic shows the most elevated
conceivable exactness alongside the accuracy , review , f-
measure.
PERFORMANCE INDICATORS
In this review, ckd was set to be positive and
notckd was set to be negative. The disarray lattice was
utilized to show the particular outcomes and assess the
presentation of the AI models.True positive (TP)
demonstrates the ckd tests were accurately analyzed.
Bogus negative (FN) demonstrates the ckd tests were
mistakenly analyzed. Bogus positive (FP) demonstrates
the notckd tests were mistakenly analyzed.
ESTABLISHING AND EVALUATING INDIVIDUAL
MODELS
The accompanying AI models have been acquired by
utilizing the comparing subset of highlights or indicators
on the total CKD informational collections for diagnosing
CKD.
Distance-based model: KNN
By and large, in sickness finding, symptomatic examples
are dispersed in a multi-faceted space. This space
includes indicators that are utilized for information
classi_cation (ckd or notckd). Tests of information in the
space are grouped in various districts because of their
various classifications. In this manner, there is a limit
between the two classifications, and the distances
between tests in a similar class are more modest. As
indicated by the viability of order, we pick the previously
mentioned techniques for infection analysis. LR is based

onLogistic Regression, and it acquires the heaviness of
every indicator and an inclination. On the off chance that
the amount of the impacts of all indicators surpasses an
edge, the classification of the example will be named ckd
or notckd. RF creates an enormous number of choice
trees by haphazardly inspecting preparing tests and
indicators. Every choice tree is prepared to observe a
limit that boosts the distinction among ckd and notckd. A
ultimate conclusion is dictated by the expectations of all
trees in the sickness determination. Isolates various
types of tests by setting up a choice surface in a complex
space that contains the indicators of the examples. KNN
finds the closest preparing tests by ascertaining the
distances between the test and the preparation tests and
afterward decides the symptomatic classification by
casting a ballot. Gullible Bayes classifier ascertains the
restrictive probabilities of the example under the span
by the quantity of ckd and notckd tests in each unique
estimation stretch. KNN can dissect non-direct
connections in the informational collections because of
its complicated construction, and the sigmoid actuation
work was utilized in the secret layer and the result layer.
MISJUDGMENT ANALYSIS AND SELECTING
COMPONENT MODELS
In the wake of assessing the above models, the potential
part models were removed for misjudgement
examination to figure out which would be utilized as the
parts. The misjudgement examination here alludes to nd
out and analyze the examples misconstrued by various
models, and afterward figure out which model is
reasonable to build up the nal coordinated model. The
misjudgement examination was performed on the
separated models. The essential for creating a
coordinated model is that the misinterpreted tests from
every part model are unique. Assuming every part model
misconstrues similar examples, the created incorporated
model would not make a right judgment for the
examples all things considered. At the point when the
information were perused, each example was given an
exceptional number going from 1 to 400.
The quantities of misjudgements for the separated
models on each total information and the dark part
shows that the examples were misinterpreted by
different models with the exception of KNN.LR), when K
equalling to 7, just a single misjudgement is at the same
time misconceived by the LR. In different cases, every
one of the examples that are misinterpreted by LR can be
accurately decided by the remainder of the models.
Subsequently, the mixes of the LR with the remainder of
the models could be utilized to set up an incorporated
model. Then, we research which explicit model mix could
produce the best incorporated model for diagnosing
CKD.
EXPERIMENTAL SETUP
To assess model execution exhaustively, on account of
holding the example dispersion in the first information, a
total informational index was separated into four
subsets equally. For all of the above models, every subset
was used once for testing, and different subsets were
used for preparing, the general outcome was taken as the
last presentation.
To check whether the coordinated model can work on
the presentation of the part models, Our outcomes show
the possibility of the proposed strategy. By the
utilization of LR, accomplish preferable execution over
the attribution was utilized. KNN attribution could fill in
the missing qualities in the informational collection for
the cases wherein the symptomatic classifications are
obscure, which is nearer to the genuine clinical
circumstance. Through the misconceptions investigation,
LR were chosen as the part models. The LR accomplished
an exactness of around 86.45 which shows most
examples in the informational collection are straightly
distinct.
ALGOR
ITHM
PRECISION
(decimal)
RECALL(
decimal)
FMEASURE
(decimal)
ACCURA
CY(%)
Naïve
Bayes
84 84.5 84 84.5
Decisi
on
Tree
80 81.9 80 81.9
Kstar 83 83 83 83.8
Logisti
c
86 86 86 86.45
SVM 80 82 80 82.9
CONCLUSION
The proposed CKD indicative system is plausible as far as
information ascription and tests finding. Later unaided
ascription of missing qualities in the informational index
by utilizing calculated attribution, the incorporated
model could accomplish a palatable precision. In this
evaluation, we propose a Logistic relapse, framework for
diagnosing CKD Hence, we theorize that applying this
76
78
80
82
84
86
88
PRECISION(de
cimal)
RECALL(deci
mal)
FMEASURE(d
ecimal)
ACCURACY(%
)

technique to the useful finding of CKD would accomplish
an advantageous impact. What's more, this philosophy
may be appropriate to the clinical information of
different illnesses in real clinical finding. In any case,
during the time spent setting up the model, because of
the limits of the conditions, the accessible information
tests are moderately little, including just 400 examples.
Accordingly, the speculation execution of the model may
be restricted. What's more, due to there are just two
classifications (ckd and notckd) of information tests in
the informational index, the model can't analyze the
seriousness of CKD. In the future, an enormous number
of more intricate and agent information will be gathered
to prepare the model to further develop the speculation
execution while empowering it to distinguish the
seriousness of the infection. We accept that this model
will be increasingly more wonderful by the expansion of
size and nature of the information.
REFERENCES
1. M. M. Hossain et al., "Mechanical anisotropy appraisal
in kidney cortex utilizing ARFI top relocation: Preclinical
approval and pilot in vivo clinical outcomes in kidney
allografts," IEEE Trans. Ultrason. Ferr., vol. 66, no. 3, pp.
551-562, Mar. 2019.
2. E. Hodneland et al., "In vivo identification of constant
kidney illness utilizing tissue misshapening fields from
dynamic MR imaging," IEEE Trans. BioMed. Eng., vol. 66,
no. 6, pp. 1779-1790, Jun. 2019.
3. G. R. Vasquez-Morales et al., "Logical expectation of
constant renal illness in the colombian populace utilizing
neural organizations and case-based thinking," IEEE
Access, vol. 7, pp. 152900-152910, Oct. 2019.
4. N. Almansour et al., "Neural organization and backing
vector machine for the expectation of constant kidney
illness: A similar report," Comput. Biol. Medications., vol.
109, pp. 101-111, Jun. 2019
5. M. Alloghani et al., "Uses of AI methods for computer
programming learning and early forecast of
understudies' presentation," in Proc. Int. Conf. Delicate
Computing in Data Science, Dec. 2018, pp. 246-258.
6. L. Du et al., "An AI based way to deal with recognize
secured wellbeing data in Chinese clinical text," Int. J.
Medications. Illuminate., vol. 116, pp. 24-32, Aug. 2018
7. R. Abbas et al., "Arrangement of fetal trouble and
hypoxia utilizing AI draws near," in Proc. Int. Conf. Smart
Computing, Jul. 2018, pp. 767-776
8. M. Mahyoub, M. Randles, T. Cook and P. Yang,
"Correlation examination of AI calculations to rank
alzheimer's illness hazard factors by significance," in
Proc. eleventh Int. Conf. Improvements in eSystems
Engineering, Sep. 2018.
9. Q. Zou et al., "Foreseeing diabetes mellitus with AI
strategies," Front. Genet., vol. 9, Nov. 2018
10. Z. Gao et al., "Finding of diabetic retinopathy utilizing
profound neural organizations," IEEE Access, vol. 7, pp.
3360-3370, Dec. 2018.

A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE

Recommended

Recommended

More Related Content

Similar to A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE

Similar to A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE (20)

More from IRJET Journal

More from IRJET Journal (20)

Recently uploaded

Recently uploaded (20)

A MACHINE LEARNING METHODOLOGY FOR DIAGNOSING CHRONIC KIDNEY DISEASE