He told the Post that a Manhattan high school student has designed an AI algorithm that helps 911 callers get the help they actually need.
Pierce Wright, a 17-year-old, soft-spoken third-grader at Manhattan's Browning School, says his complex model can help respond to emergency calls by predicting, for example, when a caller is likely to have a mental health condition. He said that there is a possibility that he could help people.
“If the algorithm says, 'I believe it's a mental health call,' it will send a psychiatrist or mental health professional along with paramedics to assist the patient, rather than just rushing the police to the scene.” and provide better care,'' Wright said in an interview Wednesday.
“It's a time saver for both the patient and the city,” he said. “It also allows us to free up ambulances faster.”
To design the algorithm, Wright combined his experience as an EMS officer with his data science abilities.
He has spent the past year painstakingly coding the AI, training it with nearly two decades of statistics collected from New York City's massive online database containing approximately 24 million emergency calls.
His work has paid off, he says, because his model can predict the resources needed based solely on factors such as the incoming call's zip code, time of day, police precinct, and initial severity level. .
Kicker? The model's success rate is an astonishing 94.5%, he said, which is 2.2% higher than the human model.
“It requires almost no input from the actual caller,” Wright said, sitting at her family's Upper East Side dining room table overlooking Park Avenue.
“In reality, all you have to do is say, 'This is what it looks like.' And the 911 operator has the zip code, the police station, the time of day, etc. You put that all in there, and the model takes a few seconds to “And it prints out what it thinks the phone will be making within about 94% accuracy.”
“It essentially predicts what you believe the phone call will be,” the teen said. “Based on that, you can send an appropriate response.”
Wright said he was inspired to create the program after answering calls on his shift as a volunteer EMT in Westport, Conn.
In many cases, crews were dispatched to respond to what turned out to be mental health or substance abuse calls, rather than true medical emergencies for which they were trained, officials said.
It also doesn't help patients, who are often taken to local emergency rooms and left there.
“This is not providing care to patients and is wasting city resources,” he said, adding that his model would help eliminate patients “just sitting in the hospital waiting to be discharged.” Ta.
Wright's mother, Melanie, marveled at her son's ability to create the model.
“I was like, 'I hope it goes well,'” she said with a laugh. “Because I don't want people to think that he spent all this time on something and it didn't work out!”
“But it was so thrilling to witness his breakthrough,” she added. “And that's going to take him to the next level.”
Wright said his program can also be used for other types of emergency calls. For example, he said, if the model determines that a victim has sustained trauma, he could prepare to dispatch paramedics as well as a paramedic.
Still, the creators were quick to say that the program is meant to help, not replace, dispatchers.
But someday, this model could save cities millions of dollars and significantly reduce response times, once it is made more customizable and accessible to the general public. , he said.
This remarkable project, which took approximately 200 hours to complete, won first place at the TerraNYC STEM Fair held at NYU Tandon School of Engineering on April 7th and the New York State Science and Engineering Fair in Queens.
And that could lead to his future career — even though it remains to be seen.
Wright said he would eventually like to work in public health, computer science, or a combination of both.
“Whatever I do, I want to create something that can help people,” he said.
“That's definitely my goal.”