The tech community has come a long way since I was 7 years old, discovering technology through a computer that ran on tape drives. That computer changed my life, which is why I’ve been so excited about the Lumiata COVID-19 Global AI Hackathon. Our event attracted people from 239 cities across 70 countries (one of our participants was only 10 years old!) — all to help improve the world in the face of a global pandemic.
I’ve been reflecting on the incredible results and wanted to share a few of the most inspiring themes as well as how we plan to explore and incorporate them into Lumiata’s future.
Uncertainty is opportunity.
COVID has challenged the tech community on so many fronts. We’ve done exceedingly well with what we have, but this novel virus moves fast and data collection has been disjointed. This has made it increasingly difficult to build meaningfully accurate predictions. If the pandemic has taught us anything, it’s the value of being able to react rapidly in situations where data is changing quickly.
One of the central challenges in healthcare and public health has been achieving consistent and comprehensive data collection in real time. Consumer-facing organizations like Facebook and Google have used real-time, online training of machine learning models for AI applications for a while now, but it’s new to healthcare — it’s an area of immense potential that I have a fresh perspective on after our hackathon. Internally, we’re building an infrastructure that allows our customers to leverage that same real-time online training, and we’re also digging deeper into unstructured data (like audio, video, and chatbot logs).
At the same time, technology hasn’t been a neutral player in the COVID-19 response. AI has been used to generate false information, increasing uncertainty around the world. This illuminates the importance of building systems that can vet systems — essentially using AI to combat AI. We’re planting these seeds at Lumiata, exploring how we can get involved in solving defensive problems, at least starting at the thought-exercise level.
Efficiency yields answers.
We’re facing a wealth of opportunities to streamline the way we think about ML modeling in healthcare. Here’s an example.
Generally, for COVID-type predictions, modeling is siloed off into three categories: statistical, simulation, and machine learning. Most COVID models have been simulations applied to solutions like capacity planning for hospitals and predictions of economic impact. Unfortunately, while the simulation models are very accurate, it’s hard to find the right inputs for those models — a challenge that machine learning could address by feeding data into simulations; in essence, combining statistics-based simulation models with machine learning models works best.
This hybrid-model approach could streamline longstanding modeling issues and improve efficiency in uncertain and increasingly complex environments. Ultimately, this application of simulations points to reinforcement learning instead of hard coding simulations, allowing us to generate millions of simulations simultaneously. Look at what Salesforce is doing in Texas with their AI Economist, working to figure out how to reopen the economy, among other things.
This is something that the hackathon clarified for me, and we’re especially excited about it at Lumiata. Just this year, we started to think about how we could use reinforcement learning to replace static simulated environments. We’re confident that this shift will enable our customers to not only solve new problems but empower them to do so more efficiently.
Imagination is everywhere.
Technology is a tool of creativity, and this hackathon proved that. With everyone practicing social distancing and rethinking the way we live and connect, it might not feel like a time of innovation; but this event demonstrated that we’re sitting on enormous potential to tap into the global community and drive capabilities that don’t exist today.
For example, one of our submissions involved a machine learning model that detects COVID using an inexpensive microphone. It’s a unique approach that addresses the challenge of physical distance and highlights the potential of using unstructured audio data to answer complex public health problems in creative ways.
Lumiata is paving the road forward.
One of my favorite lessons from this event was understanding the scale at which technology can bring us together. Looking into Lumiata’s future, we could do the same for research projects, supporting researchers in areas of the world that struggle with limited resources and infrastructure challenges. The opportunity to help people realize their potential and solve big problems looks almost limitless. The hackathon shrunk the size of the planet in my head and opened ideas on how to source talent for specific projects on a global basis. It opened my mind.
This vision of harnessing creativity in technology is part of what inspired us to launch our Innovation Lab — a long-term initiative that will harvest internal creativity, gathering solutions into a shared platform for our customers. We see it as an emerging opportunity to support growth for our customers, employees, and the tech community as a whole, and a way to prioritize our efforts that are most closely aligned with needs created by COVID-19 — initiatives like telemedicine and applying AI to personalized medicine. Ultimately, it’s about shaping our Innovation Lab so that Lumiata customers have the tools they need to solve their most pressing problems.
I want to thank the Lumiata team for putting on the hackathon while also supporting our customers. The event is something that we’re incredibly proud to have hosted and are excited about the changes it will generate for us, the people we serve, and the global community.
I’m incredibly excited about what the future of AI will bring to healthcare and so many other businesses. Some say that we’re going through a transformative period with AI that could result in a bigger impact than the first Industrial Revolution. I deeply believe that.