FORM 2
THE PATENT ACT 1970&
The Patents Rules, 2003
COMPLETE SPECIFICATION
(See section 10 and rule 13)
TITLE OF THE INVENTION:
Intelligent Detection and Notification of an Autoimmune Disease using Machine
Learning at Covid-19 and Oral Cancer Patient’s (Including Children and Adults)
Name Nationality Address
Applicants
People’s University
AN INDIAN
NATIONAL People’s University, Bhopal (M.P) 462037, India.
Dr. Parimala Kulkarni
AN INDIAN
NATIONAL
Dean, PCD & RC, People’s University, Bhopal (M.P)
462037
padutt@gmail.com,
akumarrmuit@gmail.com
Ankit Kumar Dwivedi AN INDIAN
NATIONAL
Assistant Professor, ECE,SORT, People’s University,
Bhopal (M.P) 462037
Inventors
Dr. Parimala Kulkarni AN INDIAN
NATIONAL
Dean, PCD & RC, People’s University, Bhopal (M.P)
462037
padutt@gmail.com,
akumarrmuit@gmail.com
Ankit Kumar Dwivedi AN INDIAN
NATIONAL
Assistant Professor, ECE,SORT, People’s University,
Bhopal (M.P) 462037
REAMBLE TO THE DESCRIPTION
PROVISIONAL COMPLETE
The following specification describes the The followings pacification Invention.
Particularly describes the invention
and the manner in which it is to be
performed.
2
FIELD OF THE INVENTION
Our Invention is related to an Intelligent Detection and Notification of an
Autoimmune Disease using Machine Learning at Covid-19 and Oral Cancer
Patient’s (Including Children and Adults)
BACKGROUND OF THE INVENTION
Autoimmune disease
Three components add to immune system infection advancement: hereditary
inclination, ecological elements and insusceptible framework dysregulation (Fig. 1).
Because of the heterogeneity of beginning and movement, analysis and anticipation
for immune system infection is flighty. An inclination to autoimmunity is
emphatically connected to hereditary qualities, and brought about by surrenders in
components that outcome in loss of self-resilience.
1. Autoimmune sickness creates after additional invulnerable framework
dysregulation, in both the intrinsic and versatile insusceptible framework.
2. Microbial antigens, unfamiliar antigens and cytokine dysregulation, can cause
enlistment of self-receptive lymphocytes.
3. Moreover, hyper-initiation of T and B cells might happen, alongside an
adjustment of the span and nature of their reaction, which further disturbs the
homeostasis of the invulnerable framework.
4. The predominance of immune system sickness is hard to gauge; illnesses are
dynamically addressed across various examinations and no conclusive rundown
exists.
5. 4–6 There is a detailed commonness pace of somewhere in the range of 4.5%5
and 9.4%,4 across every single immune system illness.
Past investigations have shown that liquor utilization is an autonomous danger
factor for the advancement of malignancy in the portion subordinate way. Liquor is
first oxidized to acetaldehyde by liquor dehydrogenase (ADH). Acetaldehyde is
viewed as a gathering I cancer-causing agent as indicated by the International
Agency for Research on Cancer (IARC). Acetaldehyde is additionally processed to
acetic acid derivation by aldehyde dehydrogenase (ALDH).
Any deformity in these proteins (ADH and ALDH) may impact the carcinogenesis by
liquor (12). Liquor likewise instigates basal cell multiplication and creates free
revolutionaries which have the pernicious impacts on DNA. Furthermore, liquor
related disability of the body's capacity to breakdown and assimilate supplements
and resistant concealment might additionally advance carcinogenesis.
Aside from tobacco use and liquor misuse, human papilloma infection has as of late
got uncommon consideration. Human papilloma infection, HPV-16 specifically, has
been demonstrated as an etiological specialist for the improvement of a subset of
squamous cell carcinoma, particularly at the foundation of the tongue and the
tonsillar region in the more youthful people contrasted with the HPV-antagonistic
3
partner .The extent of HPV-positive oropharyngeal malignant growth was 56% in
North America, 52% in Japan, 45% in Australia, 39% in Northern and Western
Europe, 38% in Eastern Europe, 17% in Southern Europe and 13% in the remainder
of the world.
OBJECTIVES OF THE INVENTION
1) The objective of the invention is to provide a “Intelligent Detection and
Notification of an Autoimmune Disease using Machine Learning at Covid-
19 and Oral Cancer Patient’s (Including Children and Adults)” is anImmune
system illnesses are ongoing, multifactorial conditions.
2) The other objective of the invention is to provide a through AI (ML), a part
of the more extensive field of man-made reasoning, it is feasible to remove
designs inside understanding information, and take advantage of these
examples to anticipate patient results for worked on clinical administration.
3) The other objective of the invention is to provide a assess potential changes
in Chest CT, through a score, that propose a more terrible forecast in patients
with COVID-19, and to distinguish designs related with more awful clinical
turns of events, to direct, in the imminent unfurling of the review, the
assessment of prognostic markers emerging robotized investigations of
Chest CT and add to focusing on treatment as per seriousness (orotracheal
intubation, hospitalization).
4) The other objective of the invention is to provide a Organize an information
base with clinical pictures and their individual anonymized reports for CT
methodology, in various transform utilitarian changes, in patients with
intense respiratory disorders.
5) The other objective of the invention is to provide anEvaluate the exhibition
of AI calculations in this information for undertakings like arrangement,
division, picture enrollment and translation of reports.
6) The other objective of the invention is to provide a Evaluate the effect of the
utilization of these models on clinical act of imaging experts.
SUMMARY OF THE INVENTION
The new Covid, which started to be called SARS-CoV-2, is a solitary abandoned RNA
beta Covid, at first recognized in Wuhan (Hubei territory, China) and at present
spreading across six mainlands making a significant mischief patient, with no
particular devices as of recently to give prognostic results. Accordingly, the point of
this review is to assess potential discoveries on chest CT of patients with signs and
manifestations of respiratory disorders and positive epidemiological components
for COVID-19 contamination and to associate them with the course of the infection.
In this sense, it is likewise expected to foster explicit AI calculation for this reason,
through aspiratory division, which can foresee conceivable prognostic elements,
through more exact outcomes.
4
Our elective theory is that the AI model dependent on clinical, radiological and
epidemiological information will actually want to anticipate the seriousness
forecast of patients contaminated with COVID-19. We will play out a multicenter
review longitudinal review to acquire an enormous number of cases in a brief
timeframe, for better review approval. Our accommodation test (no less than 20
cases for every result) will be gathered in each middle thinking about the
consideration and avoidance standards. We will assess patients who enter the
medical clinic with clinical signs and side effects of intense respiratory disorder,
from March to May 2020.
We will incorporate people with signs and manifestations of intense respiratory
condition, with positive epidemiological history for COVID-19, who have played out
a chest registered tomography. We will evaluate chest CT of these patients and to
correspond them with the course of the illness.
Essential outcomes:
1) Time to emergency clinic release;
2) Length of stay in the ICU;
3) orotracheal intubation;
4) Development of Acute Respiratory Discomfort Syndrome. Auxiliary outcomes:
1) Sepsis;
2) Hypotension or cardiocirculatory brokenness requiring the solution of
vasopressors or inotropes;
3) Coagulopathy;
4) Acute Myocardial Infarction;
5) Acute Renal Insufficiency;
6) Death.
We will utilize the AUC and F1-score of these calculations as the fundamental
measurements, and we desire to distinguish calculations equipped for summing up
their outcomes for each predetermined essential and optional result.
Supervised and unsupervised machine learning
Two kinds of ML are talked about here: directed and unaided learning. During
regulated learning, a calculation is prepared on a "preparation dataset" to perceive
the examples that are related with explicit "names" (for instance, solid or ailing).
When prescient examples have been gained from preparing information, the ML
calculation is then ready to appoint marks to concealed "test information". In a very
much prepared model, the examples recognized in the preparation information will
sum up to the test information.
Brief depictions of the absolute most normal directed ML procedures alluded to in
this audit are summed up in Box 1. For unaided getting the hang of, preparing
information are unlabelled, and the calculation rather endeavors to discover and
address designs inside the information, for instance by distinguishing bunches
dependent on the closeness of the models. Different sorts of ML exist, yet are
evaluated elsewhere.15 Some of the more normal unaided techniques examined in
this audit incorporate progressive grouping and self-arranging maps.
5
BRIEF DESCRIPTION OF THE DIAGRAM
FIG.1: Intelligent Detection and Notification of an Autoimmune Disease using
Machine Learning at Covid-19 and Oral Cancer Patient’s (Including Children and
Adults) Flow Chart.
FIG.2: Intelligent Detection and Notification of an Autoimmune Disease using
Machine Learning at Covid-19 and Oral Cancer Patient’s (Including Children and
Adults) Block Diagram.
FIG.3: Intelligent Detection and Notification of an Autoimmune Disease using
Machine Learning at Covid-19 and Oral Cancer Patient’s (Including Children and
Adults)
DESCRIPTION OF THE INVENTION
Algorithm development and AI data evaluation
Informational index: The informational index will comprise of Chest CT acted in
research facilities having a place with the Coordinator Center organization and
accomplice establishments. Radiologists will comment on piece of this arrangement
of tests, and every explanation will be made out of a polygon that fragments the lung
in each cut of the figured tomography. Each preparation example will then, at that
point, comprise of a cut of a processed tomography test (section to the model) and
the polygon that isolates the lung (anticipated reaction from the model). It ought to
be noticed that the development of the calculation will be performed at the
Coordinator Center.
The preparation of the division model will happen in a managed way, utilizing the
commented on set of information, where each occasion of this set should have a
polygon that delimits the area alluding to the lung in the test. We will utilize the
division model to compute the lung volumes and rates of GGO and solidifications in
the non-clarified Chest CT. This data, alongside clinical and laboratorial
information, will give information to foster an AI model that predicts the essential
and optional results.
For all models, we will just consider patients that either passed on or were released.
For essential and auxiliary results examination, we will prohibit the obscure cases.
We will analyze distinctive arrangement models, similar to Light GBM and Catboost.
CatBoost is a model from the group of slope boosting trees. Albeit incredible, slope
boosting trees tend to overfit the preparation information when taken care of with
such a large number of components, that is the reason we limited the quantity of
provisions to be concentrated in this convention. We will likewise limit the
profundity of the trees to three and fix the learning rate at 0.01 and the persistence
for early halting at 50 ages.
The essential measurement is the region under the recipient working trademark
bend (AUC—ROC). For each model, we will isolate 20% of the considered patients
to fabricate our test set. On the other 80%, we will run a 5 overlap cross-approval,
6
and every one of our outcomes will have a mean and a standard deviation on the
holdout patients. These calculations will be assessed utilizing different execution
measurements (affectability, particularity, positive and negative prescient
qualities, and F1-Score), to decide the speculation of results in our populace, and
the exhibition according to adequacy, wellbeing and quality. We will ascertain F1
and precision for all edges, and select the best limit.
Information Availability:
The information can't be made freely accessible in light of the fact that only one
organization had an IRB endorsement to share information openly (Universidade
Federal de São Paulo - Unifesp), as this foundation has specialized means currently
set up to adequately redact patient information to permit public sharing. The IRBs
in different organizations (Universidade Federal do Rio de Janeiro, Universidade do
Estado do Rio de Janeiro, Hospital 9 de julho, Hospital São Lucas, Hospital Santa
Paula, and Hospital Alemão Oswaldo Cruz) permit studies to be directed inside
foundations, however not to share information openly due to conceivably
recognizing or delicate patient data, despite the fact that the information is dedistinguished.
Subsidizing:
This review was subsidized by Diagnósticos da América (DASA). The funder offered
help as pay rates for creators FPPLL, FCK, PEAK, MRTG, VPSR, MRFM, RZP, RASO,
LGPD, PRTPD, and RVO, yet didn't play any extra part in the review plan,
information assortment and investigation, choice to distribute, or arrangement of
the original copy. The particular jobs of these creators are verbalized in the 'creator
commitments' segment.
Material and Methods
Biopsy records of the taking part establishments were explored for oral disease
cases analyzed from 2005 to 2014. Segment information and site of the sores were
gathered. Destinations of the injury were partitioned into lip, tongue, floor of the
mouth, gingiva, alveolar mucosa, sense of taste, buccal/labial mucosa, maxilla and
mandible. Oral disease was partitioned into 7 classifications: epithelial cancers,
salivary organ growths, hematologic cancers, bone growths, mesenchymal growths,
odontogenic growths, and others. Information were broke down by unmistakable
measurements utilizing SPSS programming form 17.0.
The biopsy records of the Department of Oral Pathology, Chulalongkorn University,
Gangneung-Wonju National University and Kyungpook National University,
Department of Oral Biology and Diagnostic Sciences, Chiangmai University,
Department of Oral Diagnosis, Khon Kaen University, Department of Stomatology,
Prince of Songkla University, Department of Oral and Maxillofacial Pathology,
Tehran University of Medical Sciences, Department of Pathology and Laboratory
Medicine, Western University and Department of Oral Pathology and Oral
7
Diagnosis, School of Dentistry, National Taiwan University were inspected for oral
malignant growth cases analyzed from 2005 to 2014.
Segment information and site of the sore were likewise gathered. Destinations of
the injury were partitioned into lip, tongue, floor of the mouth, gingiva, alveolar
mucosa, sense of taste, buccal/labial mucosa, maxilla and mandible. Oral disease
was partitioned into 7 classifications: epithelial cancers, salivary organ growths,
hematologic cancers, bone growths, mesenchymal growths, odontogenic growths,
and others. Information gathered were broke down by fitting measurements
utilizing SPSS Statistics for Windows, Version 17.0. Chicago: SPSS Inc. A P esteem
under .05 was viewed as measurably critical. This exploration was endorsed by the
moral advisory group of the Faculty of Dentistry, Chulalongkorn University (no.
90/2015).
Identification of patients
Studies used ML strategies to distinguish patients with immune system illnesses
from electronic clinical records,19–25 and utilized regular language handling.
Gronsbell et al. attempted to work on the proficiency of calculations for this
purpose.26,27 These calculations are expected to supplant International
Classification of Diseases charging codes, which have mistake paces of between
17.1–76.9% due to conflicting terminology.19 Electronic clinical records likewise
recognized comorbidities related with alopecia and vitiligo utilizing normal
language preparing.
This recognized comparative immune system comorbidities for both diseases.28,29
Identifying and evaluating immune system sickness hazard Prediction of infection
risk30–39 and ID of novel danger factors through include selection40–44 was
archived for IBD, type 1 diabetes (T1D), RA, fundamental lupus erythematosus
(SLE) and MS. Fifteen investigations utilized hereditary information, utilizing either
sequencing exhibits (GWAS) or exome information (nine examinations), individual
SNPs38 inside in the HLA regions37,45 or pre-chosen genes,41 or quality
articulation data.30,43 Only one review utilized clinical data,31 and two others
joined clinical and genomic data.30,45 Popular models included arbitrary
timberland, support vector machine and strategic relapse.
Checking and Notification:
Ten distinct investigations of type 1 diabetes (T1D) utilized ML for observing and
the executives: four anticipated blood glucose level, four recognized or anticipated
hypoglycemic occasions, and two upheld dynamic utilizing case-based thinking or
choice emotionally supportive networks. Most of models utilized clinical
information. Three models were created utilizing action estimations for checking
development in MS, and one in RA. Backing vector relapse was utilized most
frequently.100
Results
8
Of the 474,851 accessioned cases, 6,151 cases (1.30%) were analyzed in the class
of oral disease. The mean age of the patients was 58.37±15.77 years. A sum of 4,238
cases (68.90%) were analyzed in guys, though 1911 cases (31.07%) were analyzed
in females. The male-to-female proportion was 2.22:1. The destinations of
inclination for oral disease were tongue, labial/buccal mucosa, gingiva, sense of
taste, and alveolar mucosa, individually. The three most normal oral disease in the
dropping request of recurrence were squamous cell carcinoma, non-Hodgkin
lymphoma and mucoepidermoid carcinoma.
Albeit the predominance of oral malignancy isn't high contrasted with different
elements, oral malignancy presents huge mortality and dismalness in the patients,
particularly when found late over the span of the illness. This review features some
physical areas where oral diseases are often experienced. Thus, clinicians should
focus on teeth, however oral mucosa particularly in the high predominance region
also since early discovery of precancerous injuries or diseases in the beginning
phase increment the shot at patient being relieved and significantly lessen the
mortality and dismalness. This concentrate additionally shows a few contrasts
among pediatric and older oral malignancy patients just as among Asian and non-
Asian oral disease patients.
9
WE CLAIMS
1. Our Invention “Intelligent Detection and Notification of an Autoimmune
Disease using Machine Learning at Covid-19 and Oral Cancer Patient’s
(Including Children and Adults)” is a Immune system illnesses are ongoing,
multifactorial conditions. Through AI (ML), a part of the more extensive field
of man-made reasoning, it is feasible to remove designs inside
understanding information, and take advantage of these examples to
anticipate patient results for worked on clinical administration. Here, we
reviewed the utilization of ML strategies to resolve clinical issues in immune
system sickness. An orderly survey was led utilizing MEDLINE, embase and
PCs and applied sciences complete information bases. Important papers
included "AI" or "man-made consciousness" and the immune system
infections search term(s) in their title, dynamic or watchwords. Rejection
rules: concentrates on not written in English, no genuine human patient
information included, distribution preceding 2001, concentrates on that
were not peer assessed, non-immune system illness comorbidity
exploration and audit papers. 169 (of 702) reads met the models for
incorporation. Backing vector machines and irregular woodlands were the
most well-known ML techniques utilized. ML models utilizing information
on numerous sclerosis, rheumatoid joint pain and incendiary inside
infection were generally normal. A little extent of studies (7.7% or 13/169)
consolidated various information types in the demonstrating system. Crossapproval,
joined with a different testing set for more vigorous model
assessment happened in 8.3% of papers (14/169). The field might profit
from taking on a best act of approval, cross-approval and free testing of ML
models. Many models accomplished great prescient outcomes in
straightforward situations (for example order of cases and controls).
Movement to more intricate prescient models might be reachable in future
through reconciliation of numerous information types.
2. According to claim1# the invention is to a “Intelligent Detection and
Notification of an Autoimmune Disease using Machine Learning at Covid-
19 and Oral Cancer Patient’s (Including Children and Adults)” is a Immune
system illnesses are ongoing, multifactorial conditions.
3. According to claim1,2# the invention is to a Through AI (ML), a part of the
more extensive field of man-made reasoning, it is feasible to remove designs
inside understanding information, and take advantage of these examples to
anticipate patient results for worked on clinical administration.
4. According to claim1,2,3# the invention is to a we reviewed the utilization of
ML strategies to resolve clinical issues in immune system sickness and also
10
an orderly survey was led utilizing MEDLINE, embase and PCs and applied
sciences complete information bases.
5. According to claim1,2,4# the invention is to a Important papers included
"AI" or "man-made consciousness" and the immune system infections search
term(s) in their title, dynamic or watchwords. Rejection rules concentrates
on not written in English, no genuine human patient information included,
distribution preceding 2001, concentrates on that were not peer assessed,
non-immune system illness comorbidity exploration and audit papers. 169
(of 702) reads met the models for incorporation.
6. According to claim1,2,4,5# the invention is to a Backing vector machines and
irregular woodlands were the most well-known ML techniques utilized and
also ML models utilizing information on numerous sclerosis, rheumatoid
joint pain and incendiary inside infection were generally normal.
7. According to claim1,2,6# the invention is to a little extent of studies (7.7%
or 13/169) consolidated various information types in the demonstrating
system. Cross-approval, joined with a different testing set for more vigorous
model assessment happened in 8.3% of papers (14/169) and also The field
might profit from taking on a best act of approval, cross-approval and free
testing of ML models.
8. According to claim1,2,5,6# the invention is to a Many models accomplished
great prescient outcomes in straightforward situations (for example order
of cases and controls) and also Movement to more intricate prescient models
might be reachable in future through reconciliation of numerous
information types.