This week, we hosted our second virtual panel focused on how AI can empower healthcare organizations to make better decisions and save lives. Improved forecasting and predictions lead to higher chances in managing and mitigating adverse events, such as the COVID-19 pandemic. I’m proud to acknowledge that H2O.ai is committed to helping customers and researchers with various use-cases in the healthcare industry and that we are also looking ahead. How does a post-COVID-19 world look like? What can we learn from this pandemic? What are the main changes society will face once the imminent threat has passed? 

These are some of the questions that set the tone of the conversation during the panel hosted by Ian Gomez with Sri Ambati, CEO and Founder, H2O.ai and healthcare industry experts:  

• Nicholas P. Jewell, Professor of Biostatistics and Epidemiology at London School of Hygiene & Tropical Medicine
• Balaji Apparsamy, VP, Data & Analytics at Unum
• Niki Athanasiadou, Customer Data Scientist at H2O.ai
• David Engler, Senior Data Scientist & Director of Customer Success at H2O.ai
• Marios Michailidis, Kaggle Grandmaster & Research Data Scientist at H2O.ai 

The virtual event gathered more than 800 attendees of our global community represented by over 40 countries. 

“It’s great to speak to the community in times of a global pandemic. Covid-19 made the world face unprecedented challenges and there’s an urgent need to help our citizens and, ultimately, our nation and global economy. Our goal is to use the latest developments in data AI, ML to help reduce and, ultimately, eliminate risk”, said Sri Ambati when opening the panel discussion.  

Check below the highlights from the answers given by some of our panelists: 

Sri Ambati: What is the foremost thought process in the mind of the policymakers today? 

Nicholas Jewell: “I think the big question of the day is the intense pressure to bring economies back and into action”. He then completed: “In many parts of the world, the issue is how to bring the economy back up and what sometimes becomes lost in that conversation is the need to still focus intensively on the public health impact to try and reduce new infections and deaths. So the real question is how do you reverse shelter in place or shut down or lockdown policies. How do you do that without increasing the reproductive number? How easily the virus is transmitted so that it doesn’t go above? We’re getting slight evidence of that as economies start to slowly edge back to coming, reverting to something approaching normal. I think that’s the big issue at the moment. For most policymakers, I should say that it’s been in some sense to be positive. I tend to be a pessimist, so I’m going to say that up front, but to be positive has been a remarkable achievement around the world and many places even with great costs both in terms of lives and in terms of the economy, but the world as a whole has slowed and stopped to some extent the transmission of the virus by collective community participation. And that’s really a remarkable achievement that we shouldn’t lose sight of,  he analyzed.  

Niki Athanasiadou: Do you think there could be more efficient alternatives to the social distancing in the future moving forward? 

Nicholas Jewell: “I think technology clearly has a role to play in contact tracing. And there’s also a role for technology and artificial intelligence in surveillance that we’re not yet taking advantage of. Most people are fixating on the number of confirmed cases a day or the number of tests that come out positive, but no one is really monitoring the population taking advantage of modern technology. I’m thinking of things like mobility, taking temperatures of people automatically, and assessing risk because of what I see in the future. Maybe not tomorrow, but in the future, I see an opportunity for local 

communities to have a crowdsource monitoring surveillance system that will indicate when the risk of transmission starts to get uncomfortably high and interventions are required. That would allow a lot more interaction of the community than we do with complete shelter in place.  

Sri Ambati: How do you think insurance companies and businesses, such as hospitals, should cope with this situation? 

Balaji Apparsamy: “This is a great question that everybody is looking for the silver bullet, but it’s a collection of things that are driving us towards where we want to be. Like you said, this is an unprecedented event for all of us in the generation, not only in insurance or healthcare but across the globe, which impacted the economy and various businesses. But focusing specifically on the insurance side, there is a tremendous increase in short-term disability claims as you can imagine. That is also an unprecedented impact on business models. So we are readjusting our scale in many different ways using our analytics, ML and AI to figure out where it’s going to be the sweet spot” (12’07”-13’20”). Balaji then concluded his thought by adding: “We are privileged to be at this time. That’s the way I want to see it. With all the challenges, we have the opportunity to make an impact”. (full answer 12’08” – 15’39”) 

Sri Ambati: Marios, you’ve been playing on the datasets in Kaggle. Would you like to talk about data quality in the age of predictions for Covid 19 and how that’s impacting both the predictions as well as how we can go back and improve some of those models that we started off?  

Marios Michailidis:  “Data quality, preprocessing, making certain that you don’t have abnormalities in the data and also always be prepared that things might change. This is something you need to have in the back of your mind in order to be able to make better, more robust predictions to the extent that this is possible.” 

David Engler: “I’d like to just comment on data quality. There are a lot of comments and justifiably so about the questions around data quality right now with Covid 19, there’s incomplete reporting, inaccurate reporting. But I think that what is going to happen and granted those are all challenges for predicting COVID 19 today. It’s a very challenging problem as Marios mentioned, especially when you’re looking at 30 days out. But as we progress beyond this, one of the roles of AI will be to utilize the current data that we’re gathering for Covid 19 and use that to assist us in future pandemics. When you look at the countries right now that have managed this well, this particular pandemic, well it’s those countries who have prior experience. Taiwan, South Korea, experienced SARS in 2003, swine flu in 2009. And while you know those weren’t perfect matches to this outbreak, obviously they were to obtain lessons from that. So as the data continue to become accumulated, as the data become more complete, as we get a fuller picture of what actually happened during this paramedic operate, it will be useful in the future.”  

While the conversation went on, the community had the opportunity to see demos of H2O Q , our new and innovative AI platform that delivers instantaneous insights and predictions for the “in the moment” business questions. And as the questions from the audience started coming in, the panelists we able to exchange valuable insights on how has Covid-19 become a case study of how different countries and local geographies reacted differently to the pandemic; how some of the retailers are beginning to assess how their sales are correlating with the rise of cases (and how some of the modelings can help understand sales of critical supplies); as well as the economic effects and challenges businesses are facing.  

You can watch the full recording of the panel here