ElectrifAi systems diagnoses lung infections as COVID-19 positive (Credit: ElectrifAi)
1. CT scan of infected lungs; early symptoms (source). Credit: Bernheim A, Mei X, Huang M, et al. Chest CT findings in Coronavirus Disease-19 (COVID-19): Relationship to Duration of Infection. Radiology doi: 10.1148/radiol.2020200463. Published online February 20, 2020. © Radiological Society of North America.
2. Lung extraction using ElectrifAi’s I4-body suite product.
3. Symptoms of COVID-19 in the lungs of the patient detected using the I4COVID19 product.
4. Segmentation and extraction of symptoms as AI ready images patches.
This is an Exclusive, First Look into the story behind in ElectrifAi and their journey to open source AI for the fight against coronavirus, COVID-19.
An Interview with Edward Scott, CEO of ElectrifAi.
(1) Who is ElectrifAi?
ElectrifAI is the United States’ oldest machine learning company focused in the procurement industry from a closed proprietary platform to open source on the cloud with Spark, Kubernetes, Docker, and Notebooks. The company has been over 15 years in the Healthcare, manufacturing, Federal practice, and financial services industries.
ElectrifAi has re-architected and re-engineered their technology stack to be more modern, open, and agile.
Products focus on:
-ProcureAI (how companies spend money)
-ContractAI (what hidden risks exist)
-ImageAI (using data lakes with datasets such as the coronavirus)
-Customer retention and development with healthcare and customer steerage
-Revenue capture for hospitals to capture mischarges all done in machine learning
ElectrifAi has a team of 250 data scientists and full stack engineers between Shanghai, Delhi, Jersey City, and San Diego.
Today’s C-Suite and Executives need help to solve problems. The world does not need more bandwidth, storage, or compute.
Everyone’s data is disconnected and disparate in SAP systems, Oracle systems, IBM systems, EPIC systems, Allscript systems or even Cerner systems.
Without clean data there is no AI.
This is the state of data that nobody wants to talk about.
Data is stuck in all these data lakes.
ElectrifAi is good at unfreezing these data lakes quickly, to help the C-Suite with insights to drive business with regards cost reduction, risk reduction, and revenue improvement.
(2) How can companies drive insights for their data?
Over 50% of Machine learning and AI projects fail.
There has to be leadership and momentum from the C-Suite that this is incredibly important for our business.
Richard Fain at Royal Caribbean is doing good work here. His entire group got together as a massive transportation and hospitality cruise line, and decided they were going to drive AI and ML into every part of the business in 2019 and 2020 strategy.
A comprehensive ML program is needed by the C-Suite.
Companies must embrace digitization.
Most companies get the pressure from the board and the C-suite to turn to ML and AI.
You need leadership and clear scoping and project definition.
Everyone’s data is a mess
Assumption: Everyone has a data lake, and therefore they will be prepared for machine learning — this is not true.
Data has to be cleaned, normalized, and go through a DQM (data quality management) process.
The success of AI and ML is contingent upon your capability and competency in the data pipeline.
We have 15 years of experience in data pipeline experience for the world’s largest corporations and the world’s largest hospitals.
You cannot replicate that out in the market.
That is why these corporations and the cloud providers — Google, Databricks, Azure, Oracle, others are coming to us because of that competency in taking in massive amounts of data.
We process terabytes of data every day at ElectrifAi for all these companies.
The most successful companies of the next 10–20 years will harness the power of digitization.
Here is a challenge to Corporate America:
If you are a CEO and you do not and cannot press a return on investment or a return on investment capital from data lakes and datamarts, from all the tools and companies — all the tens of millions of dollars, you will be out of a job.
Going forward you have to say, I have spent this amount of money with this cloud provider, this tool provider, this ML company and this is my ROI. You need to be able to do this in the digital world.
(3) What has ElectrifAi done to prepare for Digital Transformation?
ElectrifAI is focused on the TMT sector, healthcare, financial services and federal space, to help companies to dial up revenue, dial down cost, and dial down risk.
We run procurement programs for large corporations such as CVS with machine learning to see where they are spending their money, and to connect that spend to the contract, so they can see with what vendors they are spending what money, and with the terms and conditions and relationships with those vendors were contractually.
Was there supposed to be a discount in the past?
In ContractAI, the power of ML is to use AI and NLP to extract key terms, words, and conditions from contracts.
What is your risk?
What is your opportunity?
How can you reduce the number of suppliers?
How can you reduce the suppliers who are not focused on social issues like human trafficking and proper growing of crops and goods?
These things matter today in the Board Room and C-suite.
This Coronavirus is not SARS.
Coronavirus is impacting the entire world.
We are seeing the stress of the financial markets in the United States.
We are seeing it in the equity markets, the debt markets, the CDS spreads of these companies, which is a reflection of the risk.
Airlines are facing an existential moment right now.
As a fixed cost business, they cannot survive with the loads and yields that they are currently experiencing.
You are starting to see airlines in Europe go bankrupt, and you will see that soon in the United States.
The same with hotel companies and cruise line companies.
Anybody who has fixed expenses, transportation companies — when the yields and the loads drop, businesses cannot respond quickly enough.
The real issue is that executives do not understand the risk they are exposed to in their contractual requirements.
ElectrifAi’s ContractAI lifts out those key terms, words, phrase, and clauses that show the risk that they are under.
By the same token, what we saw in 2008 with the insurance companies was a massive dislocation.
Insurance companies and reinsurers are often times the last holdout for risk.
Today, we are talking with insurance companies about what is in their contracts.
People don’t know what is in their contracts, and money is moving too quickly.
ContractAI shows the risk.
We have built a suite of practical ML products that lets the C-Suite to drive down the cost, drive down the risk, and help drive revenue.
It has got to be more than bandwidth, storage and feeds and speeds.
To be a fully world class machine learning products company, we have:
A culture of urgency, transparency, re-invention and self-examination and teamwork.
And team members including:
LuMing Wang CTO, former Head of Deep Learning at Uber and Microsoft
Gregg McNulty, CIO, former CTO of CVS Health
Steve Holodak, SVP Healthcare, the former Data Officer at CVS Health
Jim McGowan, SVP Cloud Partners, former Senior data scientist at Bell Labs.
Michael Fox, SVP, Delivery & Operations, from VMWare
Nancy Hornberger, EVP Healthcare, from IBM Watson
We have 65 people in China, the same in India, and the balance of our company here in the United States, and every office, 24 hours a day is sharing technology, is sharing product information, and is sharing knowledge.
We are developing, delivering and selling ML products for our customers in each of the markets, 24 hours a day.
(4) How big of a threat is COVID-19 to the economy?
Could it be the biggest startup bust since 1999?
ElectrifAi can help better dial down that risk to hedge bets.
Sequoia, Upfront Ventures, and Andreessen Horowitz all say:
Startup capital may be challenging to raise through 2021.
“I think we are in a once in 100 year Black Swan event.” — Edward Scott, CEO of ElectrifAi
Our Head of China and VP of Analytics, Xian Sun adds input:
China could control where 750 million people did or did not move.
The United States cannot do that.
The network effect of how this virus spreads will benefit from that inability to control it.
It is going to hurt in technology businesses, small businesses and businesses with tremendous fixed costs and they cannot adjust those fixed costs or that risk quick enough.
You have got to see:
(1) What am I spending my money on that I can cut? We have that with our ProcurementAI.
(2) What’s in my contracts — rights and obligations. That’s ContractAI.
We are talking to manufacturing companies, banking companies, hospitals, and payeers.
This is a real-time event.
Nobody has ever had to think about staying in place for 2 weeks.
Most companies and families are not set up for this.
I think you are going to see pressure in the airline business.
We have a long-term relationship with United Airlines to help them get out of this mess, and we have been partners with them for 5 years.
You have to dial down cost and risk.
This is not about speeds feeds instances, GPUs and TPUs.
(5) How are you helping companies manage their risk?
What are airline’ fixed costs ? Labor and fuel.
You have to know what is in your contracts on the fuel side.
Most airlines have forward bought purchases on fuel and there may or may not be hedges on that.
Airlines are going to seek to get out of those fuel contracts with firms like traders and refineries like Marathon, Valero, and Exxon.
It’s all going to be about the contracts.
As a CEO or CFO you cannot go in front of your board unless you truly understand your risk.
You need to use AI and ML to get a view of all those words, terms and conditions to act.
(6) What are you doing in the race for COVID-19?
We are doing our part for COVID-19.
We are working with cancer institutes and health institutes to facilitate the detection of COVID-19.
We have an advanced image analytics department and expertise to automate the annotation of libraries of images for some of the worlds leading hospitals.
Imagine you are Sloane Kettering with 50 years of cancer images — livers, breast tumors and brain tumors.
That is unstructured data and difficult to make sense of for research or the data.
We have come up with a way to annotate the automation, and then turn those pixels into 1s and 0s and mimic SQL to search a database to say over the last 50 years, “give me all the liver tumors.”
That is real power for ML.
For COVID — we have had institutions across the world send us images of infected lungs.
Our system does not require 100,000 images of COVID-infected lungs.
We require 10.
With that 10, we can pick out all the features of COVID as it relates to all the other lung and pulmonary related maladies, and detect and tell someone whether they have COVID straight away.
Whether you at JFK, LAX, or Singapore, or Beijing, or Toronto or Heathrow — everyone is getting wanded.
Well, if we can see who has a very high temperature, we can get that person into an area, we can get a quick CT scan, we can take the data from that CT scan and run it directly through our model, and within minutes tell someone whether you have COVID or not.
If that is true, we can get that person segregated quickly into care, versus having them go into the city and spreading it more as they go into the buses and subways, the hotels, the restaurants etc before they are really sick.
That is a game changer.
(7) What is enabling this technology transformation to detect COVID-19 so quickly?
We have told the Trump and Pence administration that we will give this technology away to the United States and to the world.
That is AI for Good.
That is ElectrifAi.
David — I don’t think in the last 100 years have we seen researchers collaborate to this extent in modern history. Google and Deepmind just decentralized the protein folds for COVID as well.
People’s data is either a mess or it is frozen and locked and they cannot get to it.
That is the principal problem in the world.
These data lakes of images are a perfect example of this.
What we are doing with the great hospitals of the world is to take all of their scans and to automatically annotate those and make them searchable in a database, not with inference, but with precision.
To do that with precision and to make it searchable — that is practical machine learning.
Ed — We have not seen since 1940s or 1950s the collaboration of the world together to attack this issue.
We are not Americans. We are not Chinese. We are not French. We are not Indians. We are not Moroccans.
We are citizens of the world, and we have to solve this problem together and now.
We are glad to provide our small part in the fight against COVID-19.
(8) How are you adapting to the new world defined by COVID-19?
We have procedures and policies put together on travel since January 2019.
We began to talk about economic impacts when it was called coronavirus 60 days ago.
We have over 65 data scientist professionals in China and we heard from them about their fear in January and February.
We saw what occurred in Shanghai, when people were required to stay in their apartments, when food was delivered to people, when only one person could go out every 2 or 3 days.
To the Chinese people’s credit, they soldiered through it.
COVID-19 occurred in China in the Chinese New Year.
We knew it was only a matter of time before it is here in the United States and across Europe.
Sadly, in this case we were right.
Today, we could shut our offices in Jersey City and San Diego, and continue to deliver, develop and sell to clients each and every day with no interruption.
This is a credit to our staff such as Gregg McNulty who thinks about risk.
We needed to put these policies and procedures well ahead — and we are ready.
I think businesses are ready to adapt and will be willing to adjust to the new world.
Those that are adaptable, customer-driven, value-driven and flexible will figure out ways to organize themselves remotely with Zoom and Slack to provide continuous delivery of service to their clients.
COVID-19 can go on for 1 or 2 quarters, and many companies do not have the financial wherewithal to survive one or two quarters.
(9) How bad will it get for the economy?
I don’t think we know the answer to your question.
Our Head of China, Ms. Sun shared that from her perspective that controlling a population of 750 million people for a period of time did control the spread of coronavirus.
Whether it arrests the second wave spread due to potential COVID-19 mutations remains to be seen.
In the United States, we don’t have those same luxuries as the Chinese government to control how people control their lives or move.
The United States is going to be faced with that pretty quickly.
I don’t see COVID-19 going away in a quarter or two. I see this as a meaningful 2020 event, and it will impact the valuation of companies and survivability.
I think Sequoia and Andreessen Horowitz are right that this is an existential moment.
There are many differences between 2008 and today.
Then was a leveraging and a housing and financial services driven problem, and largely a problem in the credit markets with discipline, and the fall of AIG and the assumptions of all the banks in the fed system.
This is a black swan event that came out of nowhere.
Unlike 2008, the Fed and the ECB had the ability to grease the system and restart it and reliquidify to drop those rates to near 0.
You remember TARP and HARP and all those programs?
We don’t have those right now.
The 30 year and the 10 year is around the lowest it has been ever.
We have never seen spreads or rates like this.
The bond market is telling you that there is a real crisis out there.
The markets on the equity side do not know what to do.
The smartest funds in the world do not know what is going on.
The credit markets are much smarter than the equity markets.
Michael Milken had told me, “When you see spreads in the high yield market blow out, watch out.”
It’s beginning to percolate.
Fixed income markets are telling us there has been trouble for a while.
(10) Who will be the Time Person of the year in 2019?
Ed Scott — Humanity.
David Yakobovitch — COVID-19
Ed believes we will come together as a family.
AI for the Good.
“Corporations, Hospitals, Payors, and Providers, we are here to help and solve problems to drive revenue, reduce costs, reduce risk.”
Our people are here standing by 24 hours a day.
We are here to help.