Ahead of the Curve

Michele Samorani and Haibing Lu were walking back from lunch one day in spring 2019 when a chance encounter in Lucas Hall nudged their research in an unexpected direction.

When fellow Santa Clara University Leavey School of Business professor Michael Santoro bumped into the two data scientists, they were clearly excited. They told their colleague they had developed a new machine-learning algorithm that would better optimize medical scheduling for health care organizations. Even in a field brimming with AI tech, this could be a leap forward.

They had a problem, though. In testing the algorithm, Black patients invariably got the least desirable appointment slots with doctors and other caregivers. Bias reared its ugly head in the dataset.

That chance encounter led to a long summer spent scribbling on whiteboards and locked in deep conversation. Lead researcher Samorani, an associate professor and the director of the master of science and information systems program, and Lu, professor and department chair of information systems and analytics, patiently explained the data science behind the algorithm to Santoro. In turn, the management professor dug deeply into the ethical implications of bias at work in AI.

“We’re used to technology being ahead of ethics,” Santoro says. “The Silicon Valley mantra has been to break things first and figure out what’s happening after. We did not want that to be the case here.”

SOLVE FIRST, PUBLISH SECOND

In many ways, the work of Samorani, Lu, Santoro, and their partners on the project is ideal for Leavey School of Business research in 2023.

It blends technical business analysis with the Jesuit socially minded ethics that have been present since the school’s founding 100 years ago. It crosses disciplines. It is rooted in the advanced technology that drives Silicon Valley’s economy and name recognition.

Perhaps most important, though, the team was more interested in developing a solution than just pointing out a problem.

To accomplish that, the three researchers realized they needed more diverse perspectives on the subject matter. They had neither the personal experience of being a Black American navigating the health care system nor the public health policy expertise to understand systemic issues.

So they expanded the team. They added Shannon Harris, an associate professor at Virginia Commonwealth University and a fellow data scientist. They also invited Linda Goler Blount, president and CEO of the Black Women’s Health Imperative and a public health policy expert with years of experience studying inequities in health care.

Blount in particular sought to dispel the myth that machines are fair by their very nature. “The development and application of technology are not without bias and not only perpetuate but can, in fact, cause negative patient experiences leading to worsening patient outcomes and disparities,” she said upon the research’s release.

Together, the team worked to identify the reasons Black patients received worse appointment time slots. For example, AI algorithms often reward patients who have a history of attending all of their previous appointments. Those patients get the best time slots—the ones less likely to be overbooked and have long wait times. On the flip side, patients who have missed past appointments get penalized with bad appointment time slots.

In theory, this helps health care providers prioritize compliant patients over those likely not to show up. However, it fails to account for the systemic factors at play. For instance, Black patients have been more likely to miss an appointment if they lack access to transportation. Is a system that then penalizes them with increasingly worse health care outcomes in the future just?

Over time, the team refined the algorithm to make it race-aware, ensuring the model would still optimize scheduling for providers while not penalizing any particular group of patients. The resulting paper, “Overbooked and Overlooked: Machine Learning and Racial Bias in Medical Appointment Scheduling,” was published in August 2021.

It is already widely cited and has received awards and recognition from the AACSB (business schools’ primary accrediting organization) and other groups. Perhaps most impressively, said Santoro, this year the research was recognized with the Manufacturing and Service Operations Management “Best Paper Award,” a prestigious distinction most recently earned by researchers from Stanford, UC Berkeley, and Carnegie Mellon.

He attributes the recognition to both the practical nature of the outcome and the interdisciplinary nature of the team.

“We’re a diverse team of ethicists, technologists, and public policy people,” Santoro says. “And we were way ahead of the curve in terms of wanting to come up with solutions and not just point out a problem.”

Given Santa Clara’s location in the heart of Silicon Valley and the Leavey School of Business’ connections to industry, a research focus on particularly modern technological and societal problems is hardly surprising.

“We are squarely focused on helping our Leavey community—students, faculty, and organizations— to address 21st-century business challenges, conduct impactful research, and navigate technological advances and key societal issues,” says Leavey Dean Ed Grier. “It’s always been part of our history and, I hope, always will be part of our core.”

That statement rings true for Esther Sackett, who started at Leavey in 2019. The assistant professor of management—one of Poets & Quants Top 50 Undergraduate Business Professors—studies the communication and motivations of teams.

Such a topic was always of interest in the field of management, but the shifting nature of workplace models and the very nature of teams since the pandemic have brought it into sharp relief. For Sackett, bringing approaches from other disciplines, such as anthropology and psychology, into play makes sense. So does melding traditional discussions of what motivates workers with more modern concepts about social justice and diversity, equity, and inclusion. Leavey has the foundation for such discussions.

“There’s this kind of emphasis on ethics and understanding different cultures and religions here,” Sackett says of Leavey. “And we have this backdrop of interdisciplinarity that you don’t always get elsewhere.”

Sackett also appreciates that even as the school has built its research reputation, it lacks any sense of publish-or-perish grind. Quality matters most.

“The expectation is that you will publish high-quality research, but it’s a little more free and less intensive than it might be elsewhere. And that freedom leads to interesting work,” she says.

Location matters, too. And given the school’s presence in Silicon Valley, Santoro and company are far from the only researchers tackling AI and machine learning.

Longtime finance professor Hersh Shefrin was teaching how computer neural networks could be used in finance in MBA classes as early as the 1990s. And his colleague Sanjiv Das now counts Amazon’s AI labs as his research stomping grounds.

Das had held faculty appointments at both Harvard and UC Berkeley before coming to Leavey in 2000. In addition to his long and fruitful career, and because of connections he made at Santa Clara, he has also been an Amazon Scholar, with access to Amazon Web Services’ expansive resources, for nearly four years.

“When I started there, the first thing I worked on was actually algorithmic fairness,” Das says. “I was looking at all these algorithms that make decisions: whether to give somebody a loan, whether to give them parole in a court system, even something like whether to send someone an advertisement or not. All these algorithms learn these biases because they’re trained on data that was originally biased. So the work I did there was really to measure the bias, and then look for ways to mitigate it.”

Das’ work is rooted in the idea of multimodality. The gist: Humans use many modes to assess information, influencing financial information. They might look at “tabular data” such as spreadsheets, keep up with news, and are influenced by opinions. Data does not exist in a vacuum.

His attitude and ambitious approach to research fits well with the history of scholarship at Leavey, where self-starters can thrive. When a research question requires him to learn something new, he just does it. During Das’ career, he expanded his mathematics repertoire and even studied physics so he could be better at data analysis. He has learned new coding languages when the need has arisen. He has been willing to step far outside the typical business realm because the research warrants it.

In turn, he’s flourished in what he sees as an environ- ment conducive to academic freedom and innovation. “Some of the research I’ve done here, I’m not sure I could have done anywhere else.”

Much like his colleagues, he’s also most interested in practical outcomes.

“The things that interest me are the things that are coming next,” Das says. “They’re going to change the world. If we’re early to the party, we can have an effect on them.”