John Kahan, GM of Customer Data and Analytics at Microsoft
Every year more than 1,000 children die in the United States from sudden infant death syndrome (SIDS).
It’s a frustrating diagnosis, characterized as the sudden death of an infant under 1 year of age that cannot be explained even after a full medical investigation.
For years, the medical community has been trying to understand why some children are at risk for SIDS.
It’s recommended that babies sleep on their backs and items that can cause suffocation be removed from infants’ sleeping areas. Public health campaigns encouraging parents to take these precautions have reduced the risk of SIDS in the United States.
Still, every year children die from SIDS despite no discernible cause.
In an effort to find answers, medical researchers in Seattle have gotten a boost from an unexpected source: a team of data scientists from Microsoft.
One family’s story
John Kahan, the general manager of customer data and analytics at Microsoft, works with his team members to sift through massive amounts of data every day, looking for patterns and trends to better understand the attitudes of their customers.
In their off hours, the team is using these same skills to help medical researchers parse through reams of medical data to uncover which children are at increased risk for SIDS.
The work for Kahan is personal. Almost 14 years ago, his son, Aaron, died due to SIDS.
“He cried and pooped and did the things that a baby does,” Kahan recalled of the day Aaron was born.
“The first several hours were wonderful. This was my first son,” Kahan said.
Later in the day after Kahan left the hospital, he received a call that something was wrong.
“I got this horrendous call, Aaron had stopped breathing and they put him on a respirator,” Kahan said.
Despite the medical team’s efforts, Aaron died a few days after he initially stopped breathing. Even after an autopsy, nothing was found to explain why Aaron was at risk for SIDS.
“You do as most humans do,” Kahan said. “You pick up the pieces and you hug your family and you try to figure out why.”
Last year in honor of what would have been Aaron’s 13th birthday, Kahan decided to fundraise for Seattle Children’s Hospital by scaling Mount Kilimanjaro. He told his team, including Juan Miguel Lavista, the senior data science director of Microsoft data and analytics, about the fundraiser.
Lavista said the team knew they wanted to help Kahan.
“But we thought, well, in order for us to make an impact … we wanted to donate our skills,” Lavista said. When Kahan returned from his fundraising trip, his team had a surprise for him. They wanted to see if they could use their skills as data scientists to help researchers better understand the mysterious condition that caused Aaron’s death.
Today the team is working with Dr. Nino Ramirez, the director for the Center for Integrative Brain Research at Seattle Children’s Hospital, to apply their data science techniques to information released by the U.S. Centers for Disease Control and Prevention (CDC).
Ramirez has been studying SIDS for years, and was intrigued after Kahan approached him.
“A lot of the traditional clinical works is done by pediatricians and not really by using the most modern advancements of data managing,” Ramirez said.
While there are known risk factors for SIDS including exposure to cigarette smoke, Ramirez said doctors are still trying to understand the condition. Most infants will change position, cry, or wake up in response to diminished oxygen. However, some children at risk for SIDS do not have that response.
“The prevailing hypothesis in SIDS is that these children don’t arouse when they don't have enough oxygen and then basically stop breathing,” said Ramirez. In 2015, there were 1,600 infant deaths in the United States due to SIDS.
Ramirez explained that with the help from Kahan’s team, they have been able to approach the problem in a new manner.
Kahan and Lavista’s team is focused on using the cloud computing platform Microsoft Azure, and Power BI, a visual information display program, to help researchers make sense of the data.
Hosted on Microsoft servers, the goal is to make sense of the huge amount of health data so that researchers, who are not trained in data science, can more easily study the information.
“The thing is if you look at the CDC data it’s like bewildering,” said Ramirez.
A quick search
By applying the tools Kahan’s team built to their research, Ramirez said finding useful information in the massive amount of data has become much easier. For example, Ramirez said he can quickly find out how much the risk for SIDS is increased if a woman smokes 200 cigarettes vs 100 cigarettes during the pregnancy.
“It would have taken me a lot of time and then it takes me two seconds,” Ramirez explained. The team has already found signs that every hour of prenatal care is associated with a drop in SIDS risk, while every 10 cigarettes is associated with a significant increase in risk.
“The more risk factors we understand the more we can give parents a warning to prevent this,” he said.
Ramirez’s team is hoping to publish some of their early findings in the next year. The tools they’re using will also be opened up to other researchers if they send in a request explaining what they want to study. Microsoft Philanthropies has agreed to host the cloud services for free for two years.
In addition to SIDS, Ramirez expects that the tools could be used to search for the risk factors of other rare conditions like mitochondrial disease or sudden cardiac death.
For Kahan, the work has become his “personal mission” to help doctors find a way to keep children from dying due to SIDS. He also keeps a memento of the day his son was born close by as he works.
“I keep a picture of him on my desk,” Kahan said of Aaron. “It was a wonderful time and it was very, very short.”