About the webinar
The entity that has caused a newfound global love of hand sanitizers and masks? The Coronavirus (known as ‘2019-nCov’ or ‘Covid-19), which has infected about 5,00,000 people globally within a few months!
According to the WHO: 'In the most severe cases, the infection can cause pneumonia, severe acute respiratory syndrome, and even death.' Statements like these beg the question: 'How accurate are the tests to spot the disease?' 'Can AI assist in giving a more accurate diagnosis?'
The AI Model generated via Skyl.ai’s deep learning platform can accurately detect COVID-19 through patterns in X-ray scans and differentiate it from community-acquired pneumonia and other lung diseases that may otherwise be overlooked by a doctor.
Through this webinar, we will demo how AI can be used to test the Covid19 infections, and help in treating patients with critical conditions quickly.
What you'll learn
- How healthcare institutions can leverage AI to detect COVID-19 and reduce the time taken to provide critical care to patients who are affected.
- Discuss the approach to automate the machine learning workflow, creating and deploying models in hours and not weeks or months.
- Demo: How to create an ML model that can detect COVID-19 from chest x-rays using Skyl.ai.
2. Technology leader with 20+ years expertise in Product
Development, Business strategy and Artificial Intelligence
acceleration. Active contributor in the New York AI
community.
Extensively worked with global organizations in BFSI,
Healthcare, Insurance, Manufacturing, Retail and
Ecommerce to define and implement AI strategies.
Nisha Shoukath
Co-founder,
People10 & Skyl.ai
The Speaker
3. Extensive experience building future tech products using
Machine Learning and Artificial Intelligence.
Areas of expertise includes Deep Learning, Data Analysis,
full stack development and building world class products
in ecommerce, travel and healthcare sector.
Shruti Tanwar
Lead - Data Science
The Speaker
4. Bikash Sharma
CTO and Co-founder
at Skyl.ai
CTO & Software Architect with 15 years of experience
working at the forefront of cutting-edge technology
leading innovative projects.
Areas of expertise include Architecture design, rapid
product development, Deep Learning and Data Analysis.
The Panelist
5. Getting familiar with ‘Zoom’
All dial-in participants will be muted to enable the presenters
to speak without interruption
Questions can be submitted via Zoom Questions chat
window and will be addressed at the end during Q&A
The recording will be emailed to you after the webinar
Please familiarize yourself with the Zoom ‘Control Panel’ on your screen
6. Live Demo of
detecting COVID-19
from chest X-rays using
AI and Computer Vision
Why AI &
Computer Vision
is important in
Covid-19 diagnosis
How to quickly
overcome the
challenges in building
ML models
1 2 3
In the next 45 minutes
7. Machine Learning automation platform for unstructured data
A quick intro about Skyl.ai
Guided Machine Learning Workflow
Build & deploy ML models faster on
unstructured data
Collaborative Data Collection & Labelling
Easy-to-use & scalable AI SaaS platform
8. POLL #1
At what stage of Machine learning adoption your
organization is at?
⊚ Exploring - Curious about it
⊚ Planning - Creating AI/ML strategy
⊚ Experimenting - Building proof of concepts
⊚ Scaling up - Some departments are using it
⊚ In production - Using it in product features
⊚ Transforming - AI/Ml driven business
10. > 2.6 Million
Confirmed cases of
COVID-19 worldwide
8/1000 tests as
Global average in a nation
60-80%
Perceptual error estimate
300,000+
People tested daily
using different
measures
Distinguishing COVID-
19 from similar conditions like
conventional pneumonia,
respiratory infections etc.
Large number of
affected
patients
Shortage of tests
and capacity
Complex and
error-prone data
Severe shortage of
reagents and chemicals
required for RT-PCR tests
Coronavirus Testing - Challenges now and ahead
11. Handle large volume of Data
● Learn automatically from previous diagnosis and continuously
improve
● Streamline and automate the repetitive tasks
Faster processing to aid human experts
● Identify features in images quickly and precisely
● Aid human experts to be more efficient and focus on value-add areas
Eliminate human errors
Machine Learning to the rescue!
12. Machine Learning and Computer vision
Machine learning is an application of artificial intelligence (AI) that provides
systems the ability to automatically learn and improve from experience
without being explicitly programmed.
Computer vision is the branch of machine learning specializing in how
computers “see” photos and videos.
Hence, Medical imaging is treated as Computer Vision
problems in the AI world
Hence, Medical imaging is treated as Computer Vision problem in
the AI world
13. Analysing Brain
Anomalies
Identifying Fractures
& Musculoskeletal injury
Detecting Cardiovascular
Abnormalities
Machine Learning can
flag images that offer
risk ratios that the
images contain
evidence of ALS or PLS
etc.
AI can help in identifying
hard-to-see fractures,
dislocations, or soft
tissue injuries helping
doctors to avoid
misdiagnosis
Computer Vision can
help in Cardiac
segmentation that
helps heart surgeons
to be minimally
invasive in diagnosis
and treatments
15. Chest X-Ray (CXR)
● For preliminary classification, due to its prevalent usage as a primary diagnostic test
● Use of CXR for COVID-19 diagnosis, or triage for patient management has become an
important issue to preserve limited medical resources
Computed Tomography Scan (CT Scan)
● One of the most promising techniques that might lift some of the heavy weight of the
physicians’ shoulders
● Distinguishing COVID-19 from community acquired pneumonia based on chest CT
claims a sensitivity and specificity of 90% and 96% respectively[1]
[1] Li L, Qin L, Xu Z et al. Artificial Intelligence Distinguishes COVID-19 from Community Acquired Pneumonia on Chest CT. Radiology. Radiological Society of North
America; 2020;200905.
AI Diagnostic Pathways for Covid-19
16. Left : An example of a CXR with a label of COVID-19.
Right : An example of a CXR labeled as showing
evidence of pneumonia.
Using Deep Learning in CXR and CT scan imaging
Availability of CXRs, especially in
rural and isolated areas, including
the fact that mobile X-rays are
commonly used, makes using CXR
for a preliminary classification of
COVID-19 worth pursuing.
A deep learning model can
accurately detect COVID-19 from
CT scans. CT scans, however, are
expensive and inaccessible in
many areas.
Figure : CT scan showing Covid-19 infection, red color
highlights the activation region
17. Detecting COVID-19 from
CXR
Radiology images are used to segment COVID-19
from conditions like pneumonia.
X-ray readings may resemble community
acquired pneumonia (bacterial) or other
respiratory infections.
An X-ray will show formations in the lung that
are associated with a number of respiratory
conditions including pneumonia.
AI can detect patterns specific to COVID-19
to provide accurate diagnosis.
18. Live Demo of detecting COVID-
19 using Artificial Intelligence
and Computer Vision02
20. We would like to credit Joseph Paul Cohen for collecting this data and
making it available as open source.
Github Repo https://github.com/ieee8023/covid-chestxray-dataset
By: Joseph Paul Cohen
Github: https://github.com/ieee8023
Website: http://josephpcohen.com
Sources:
https://radiopaedia.org/ (license CC BY-NC-SA)
https://www.sirm.org/category/senza-categoria/covid-19/
https://www.eurorad.org/ (license CC BY-NC-SA)
Credits
21. Live Demo of detecting
COVID-19 from chest X-rays
using AI and Computer Vision
22. POLL #2
Some challenges that you are facing while
implementing AI & Machine Learning
⊚ Not started yet, so no challenges
⊚ Data collection
⊚ Data Labeling
⊚ Large volumes of data
⊚ Identifying the right data set to
train
⊚ Data Security
⊚ Lack of knowledge of ML tools
⊚ Lack of end to end platform
⊚ Lack of expertise
⊚ Choosing the right algorithms
23. Advantages of a unified
platform Speed, Visibility,
Quality, Collaboration,
Flexibility, Security
03
24. Data Collection - Flexible options
(CSV bulk upload, APIs, Mobile capture, Form based…)
26. Data Labeling - Simple 4 steps process
(collaboration jobs, guided workflow…)
27. Data Labeling - Real-time early visibility
(class balance, missing data…)
28. Data Labeling - Early Visibility
(data frequency, data intuition, outliers, trends, labeling accuracy…)
29. Data Labeling with Effective Collaboration
(Job allocation, trend, statistics, interactive messaging…)
Analyse trends and progress
of your data labeling job in
real time with statistics and
interactive visualizations
Manage collaborator
progress, activity, interactive
messaging
30. Data Visualization to build strong data intuition
( visuals for data composition, data adequacy)
31. One click training at scale
(Easy feature sets, out of the box algorithms, API integration, hyper
parameter tuning, auto scaling…)
● Train, Deploy and Version your models
by creating feature-sets in no time with
our easy feature selection provision.
● Choose from state-of-art neural
network algorithms, tune
hyperparameters and see logs for
your training in real time.
● Integrate our powerful inference API
with your application for AI-driven
actionable intelligence.
● Auto scaling of model training based on
data and hyperparameters.
32. Model Monitoring of metrics in real-time
(inference count, execution time, accuracy…)
● Monitor your deployed
models and analyse
inference count, accuracy
and execution time.
● See how your models are
performing in real-time.
No black boxes here.
33. ● Monitor your deployed
models and analyse
inference count, accuracy
and execution time.
● See how your models are
performing in real-time. No
black boxes here.
Model Evaluation - Release Confidently
(Accuracy, Precision, Recall, F1 Score)
34. No upfront cost in Infrastructure set up
(no DevOps needed, auto-deploy, SaaS & On-prem models…)
No DevOps
required
01
Latest tech
stack
02
On premise
and saas
models
03
Scalable
On
demand
04
36. Offers for you!
1. Personalised demo
2. 15 days free trial with data credits
3. Complimentary consultation on pilot project
4. AI Implementation Playbook
www.skyl.ai contact@skyl.ai
38. 85 Broad Street, New York, NY, 10004
+1 718 300 2104, +1 646 202 9343
contact@skyl.ai
Thank you for joining!
We hope to hear from you soon
Editor's Notes
Hello everyone and welcome. Thank you for joining, my name’s Ethan and I’ll be hosting today’s webinar on Can AI help us in diagnosing Coronavirus?. First off, I’d like to introduce our 3 speakers and panelist for today’s webinar
First we have Nisha Shoukath - Nisha is a technology entrepreneur with background in investment banking.
She’s co-founded two successful technology startups and has worked with wide variety of global organizations from different industries.
She helps enterprises with defining AI strategy, and AI adoption roadmaps. Welcome, Nisha!
Next we have Shruti Tanwar - Shruti is an expert in data science who is a veteran in building SaaS products using Machine Learning and AI.
Her expertise includes Deep Learning and Data Analysis, as well as full stack development and building tech products in various different fields such as ecommerce, travel, and healthcare. Welcome, Shruti!
Finally, we have Bikash Sharma joining today.
Bikash is CTO and Software Architect with 15 years of experience in leading innovative software projects and solutions.
He’s co-founded Skyl with his expert knowledge in AI and Machine Learning. Welcome, Bikash!
Now before we begin, I’d like to briefly talk about Zoom features.
All participants in the webinar will be muted to avoid any interruptions during the session.
Any questions you might have can be submitted to the Zoom Questions chat window in the control panel which is located on the bottom of the screen and we’ll make sure to address them towards the end during the Q&A session.
Also, the recording of the webinar will be emailed to you afterwards, so don’t worry if you’ve missed any points during the session or wish to view it again
So that’s all for the introduction - now, we’ll get started with the webinar and I’ll hand over the session to Nisha
Exploring - Curious about it
Planning - Creating AI/ML strategy
Experimenting - Building proof of concepts
Scaling up - Some departments are using it
In production - Using it in product features
Transforming - AI/Ml driven business
AI tools can augment the workflow of radiologists and pathologists, acting as clinical decision support and enhancing care delivery.
Sources: https://en.wikipedia.org/wiki/2019–20_coronavirus_pandemic_by_country_and_territoryhttps://ourworldindata.org/covid-testinghttps://www.statista.com/statistics/1104645/covid19-testing-rate-select-countries-worldwide/
https://emea.gehealthcarepartners.com/images/pdfs/Rapid-Review--Radiology-Workforce-Review-FINAL.pdf
As per ACR DSI - Manual segmentation and quantitative susceptibility mapping (QSM) assessments of the motor cortex are necessary, difficult, and time consuming. Machine Learning can flag images that offer risk ratios that the images contain evidence of ALS or PLS, populate reports and streamline the process
Fractures contribute to long-term, chronic pain if not treated quickly and correctly. Fractures contribute to long-term, chronic pain if not treated quickly and correctly.
Measuring various structures of the heart to reveal an individual’s risk for cardiovascular abnormality. Computer Vision can help in Cardiac segmentation ( coronary CT angiography, artery diameter segmentations, Ventricle segmentation etc) that helps heart surgeons to be minimally invasive in diagnosis and treatments.
https://pubs.rsna.org/doi/10.1148/radiol.2020200905
https://pubs.rsna.org/doi/pdf/10.1148/radiol.2020200905
test sensitivity is the ability of a test to correctly identify those with the disease (true positive rate), whereas test specificity is the ability of the test to correctly identify those without the disease (true negative rate )
In case of viral pneumonia, chest radiographs demonstrate normal findings or unilateral or patchy bilateral areas of consolidation, nodular opacities, bronchial wall thickening, and small pleural effusions; lobar consolidation is uncommon in patients with viral pneumonia. Patients may develop acute pneumonia with rapid progression to acute respiratory distress syndrome.
Usually bacterial pneumonia is confined to one lobe that is one section.
Cases of coronavirus pneumonia tend to affect all of the lungs, instead of just small parts.
Automating the detection of abnormalities in commonly-ordered imaging tests, such as chest x-rays, could lead to quicker decision-making and fewer diagnostic errors.
For example, when a patient enters the emergency department with a complaint such as shortness of breath, “the chest radiograph is often the first imaging study that is available,” ACR DSI says.
“It can be used as a quick initial screening tool for cardiomegaly, which in and of itself can be used as a marker for heart disease. A quick visual assessment by a radiologist is sometimes inaccurate.”
3. Pneumonia and pneumothorax are two conditions that require quick reactions from providers. Both may also be prime targets for artificial intelligence algorithms.
Pneumonia, either acquired in the community or after a medical procedure, can be life threatening if left untreated. Radiology images are often used to diagnose pneumonia and distinguish the condition from other lung conditions, such as bronchitis.
Yet radiologists may not always be available to read images – and even if radiologists are present, they may have difficulty identifying pneumonia if the patient has pre-existing lung conditions, such as malignancies or cystic fibrosis.
In addition, “subtle pneumonias, such as those projecting below the dome of the diaphragms on front chest radiographs, can easily be overlooked and lead to unnecessary CT scans, which AI could help reduce,” ACR DSI says.
An AI algorithm could assess x-rays and other images for evidence of opacities that indicate pneumonia, then alert providers to the potential diagnoses to allow for speedier treatment.
Images obtained by MRI machines, CT scanners, and x-rays, as well as biopsy samples, allow clinicians to see the inner workings of the human body. However, these images often contain large amounts of complex data that can be difficult and time consuming for human providers to evaluate.
AI tools can augment the workflow of radiologists and pathologists, acting as clinical decision support and enhancing care delivery.
5 minutes intro - 10 industry awareness - 15 min demo - 20 minutes QnA
Define problem - Features model - How this model is built using skyl.ai
Add slide of Pneumonia detection
Not started yet, so no challenges
Data collection
Data Labeling
Data Bias
Large volumes of data
Identifying the right data set to train
Lack of knowledge of ML tools
Lack of end to end platform
Lack of expertise
Choosing the right algorithms
Monitoring the model performance
Benefit
For data security concerns, on-prem solution comes to the rescue.
Data stays inside your firewall on your own servers.
Data stores are encrypted.
User based access control for data labellers, collaborators and project leads.
Now, we
Thank you Nisha and Shruti, for the wonderful presentation and demo.
We had some people asking if the recording will be shared during the webinar.
As mentioned earlier, the recording of the webinar will be sent to you by email afterwards.
Before we get to the Q&A, I want to mention some of the offers Skyl has for those of you that are interested in learning more about machine learning and those of you who want to incorporate ML to your business,
Skyl offers a personalized demo as well as a free trial for 15 days.
You’ll be able to use the platform to see actual numbers on the UI like the workbench shruti’s demoed earlier and see how you can go from collection, labeling all the way to model deployment
Skyl also offers a complimentary consultation on a pilot project of your choice and an AI implementation playbook.
This is a great opportunity to get to know exactly how Skyl can achieve machine learning solutions to personal challenges you might have.
You’ll be able to see how Skyl handles your machine learning challenges, whether your challenge is in data collection and labeling or all the way to deploying and maintaining machine learning models.
If you’re interested in finding out more about how Skyl can help your machine learning needs, please visit Skyl at skyl.ai or you can send an email directly to contact@skyl.ai as you see on the screen
Now we will go ahead and take some time for questions.
Once again as a reminder, if you have any questions, you can type your questions in the question box in your control panel - located on the bottom of your Zoom screen and I’ll try to address them as many as possible if we have enough time.
Sample questions:
Jonas
Can Skyl help me in figuring out if my data needs re-labelling?
Jessica
If I have a lot of models built, how do I handle model deployments?
John
All right, so that’s it for today’s webinar, I hope you enjoyed it
We have a lot more webinars coming up on different machine learning topics and how they can be implemented into different businesses and industries,
So don’t miss out and make sure you sign up for upcoming ones as well
Thank you for joining and I hope you have a wonderful day.