I recently spoke with Anasse Bari, a professor of computer science at NYU, about his efforts to use predictive analytics and AI to predict outcomes in COVID-19 patients. He shared a bit of background about the project, as well as how Design Thinking Executive Education at Stanford has influenced his work.

Predicting COVID-19 outcomes

Anasse is leading a team — with his colleague Megan Coffee of NYU Grossman School of Medicine and collaborators in China — to help doctors predict the outcomes of patients admitted with COVID-19. They are doing so by applying predictive analytics to demographic data, clinical lab data, and radiological data to predict whether a patient will be released, develop severe respiratory problems, or fall somewhere in between.

We want to create intelligent tools to help doctors in their day-to-day work make data-driven decisions.

By developing a tool with explainable AI that can identify who is at higher risk for developing a COVID-19 related respiratory problem, doctors can augment their decision with supporting data. In turn, doctors can better allocate limited hospital resources like respirators to the most critical patients.

Where design thinking comes in

Anasse uses the principals of design thinking to understand what a business he is working with needs. In this case, before jumping into AI and data-science he’ll use design thinking methods to capture:

• What is it that the doctors and hospital need, and why?

• What are the data sources available to predict COVID-19 outcomes?

• What are the impacts of doing this?

There are natural parallels between design thinking and data science, in that both are trying to make surprisingly insightful connections between two factors that may be traditionally overlooked. In design thinking, we have methods to brainstorm, ideate, narrow down the solution space. In data science, there are algorithmic and statistical parallels to draw connections between seemly distant datasets. For example, Anasse’s work from earlier this year connects restaurant health inspections to the real estate market.

Design thinking in the modern AI classroom

Design thinking principals even permeate down to the way Anasse runs his research lab. He doesn’t use chairs and gives everyone sticky notes, which changes the traditional meeting dynamic and gives everyone an equal opportunity to contribute ideas. These methods are foundational to d.school teaching.

Anasse also teaches the next generations of data scientists at NYU, where he employs design thinking in his classrooms.

I learned at Stanford d. school the importance of design thinking in everything we do in education, product design and AI research…

We learned how design thinking uses processes from engineering, the arts, social sciences and business for the sole purpose of solving problems… and in computer science classrooms we are constantly solving problems…

The design thinking approach has been allowing me to create an efficient space for rapid AI prototyping…for instance in 2018 we did a project that aimed at answering the question: What the World Wants? From analyzing over 600 million tweets using natural language processing, the ideas of this project emerged from a design thinking meeting with sticky notes with ideas on the wall… 

From research to practice

Anasse’s initial work has been featured on CNBC, France24, and the LA Times, yet his team is still expanding their work to a larger network of hospitals. More data and more collaboration are two elements that they believe are needed for the research to have a broader impact.

My general theory about AI and Predictive Analytics is that they exist to serve as an extension to human intelligence and not as a replacement. That is the same vision we have for the applicability of AI in healthcare, where human creativity is needed more than ever in the process of building tools with AI. And marrying design thinking with AI makes it easier to solve problems in healthcare and in many other fields.

Reach out to Anasse Bari at abari@nyu.edu if you have questions about his research at the Courant Institute of Mathematical Sciences at NYU, his book “Predictive Analytics for Dummies”, his prior work at the World Bank, or his role as an advisor on AI for the United Nations.

As for design thinking, as many of us are still sheltering in place, Stanford and the d.school are transitioning our content online. Check out d.school News and Events for the latest offerings.