Cars that recognize hypoglycemia? Maybe soon

 

SAN DIEGO – When researchers at the University of Nebraska placed sensors in the cars of patients with type 1 diabetes, they found something interesting: About 3.4% of the time, the patients were driving with a blood glucose below 70 mg/dL.

Almost 10% of the time, it was above 300 mg/dL, and both hyper and hypoglycemia, but especially hypoglycemia, corresponded with erratic driving, especially at highway speeds.

The finding explains why patients taking insulin for type 1 diabetes have a 12%-19% higher risk of crashing their cars, compared with the general population. But in a larger sense, the study speaks to a new possibility as cars become smarter: monitoring drivers’ mental states and pulling over to the side of the road or otherwise taking control if there’s a problem.

The “results show that vehicle sensor and physiologic data can be successfully linked to quantify individual driver performance and behavior in drivers with metabolic disorders that affect brain function. The work we are doing could be used to tune the algorithm that drive these automated vehicles. I think this is a very important area of study,” said senior investigator Matthew Rizzo, MD, chair of the university’s department of neurological sciences in Omaha.

With funding from Toyota, his team placed a kind of black box inside the cars of 19 patients with type 1 diabetes and 16 diabetes-free controls who were of similar age and educational background. The box had a GPS and an accelerometer to detect and record hard turns, sudden stops, swerves, and other signs that something dangerous had happened. The cars were also rigged with video cameras that recorded both the driver and the view out the windshield.

Participants had the devices in their cars for a month, during which time the diabetes patients were also on continuous, 24-hour blood glucose monitoring. The investigators then synched the car data with the glucose readings, and compared it with the data from the controls’ cars. In all, the system recorded more than 1,000 hours of road time across 3,687 drives and 21,232 miles.

“What we found was that the drivers with diabetes had trouble,” Dr. Rizzo said at the American Neurological Association annual meeting.

Glucose was dangerously high or low about 13% of the time when people with diabetes were behind the wheel. Their accelerometer profiles revealed more risky maneuvering and variability in pedal control even during periods of euglycemia and moderate hyperglycemia, but particularly when hypoglycemia occurred at highway speeds.

One driver almost blacked out behind the wheel when his blood glucose fell below 40 mg/dL. “He might have been driving because he was not aware he had a problem,” Dr. Rizzo said. He is now; he was shown the video.

The team reviewed their subjects’ department of motor vehicles records for the 2 years before the study. All three car crashes in the study population were among drivers with diabetes, and they received 11 of the 13 citations (85%).

The technology has many implications. In the short term, it’s a feedback tool to help people with diabetes stay safer on the road. But the work is also “a model for us to be able to approach all kinds of medical disorders in the real world. We want generalizable models that go beyond type 1 diabetes to type 2 diabetes and other forms of encephalopathy, of which there are many in neurology.” Those models could one day lead to “automated in-vehicle technology responsive to driver’s momentary neurocognitive state. You could have [systems] that alert the car that the driver is in no state to drive; the car could even take over. We are very excited about” the possibilities, Dr. Rizzo said.

Meanwhile, “just the diagnosis of diabetes itself is not enough to restrict a person from driving. But if you record their sugars over long periods of time, and you see the kind of changes we saw in some of the drivers, it means the license might need to be adjusted slightly,” he said.

Dr. Rizzo had no relevant disclosures. One of the investigators was an employee of the Toyota Collaborative Safety Research Center.

[email protected]

SOURCE: Rizzo M, et al., ANA 2017 abstract number S131

 

SAN DIEGO – When researchers at the University of Nebraska placed sensors in the cars of patients with type 1 diabetes, they found something interesting: About 3.4% of the time, the patients were driving with a blood glucose below 70 mg/dL.

Almost 10% of the time, it was above 300 mg/dL, and both hyper and hypoglycemia, but especially hypoglycemia, corresponded with erratic driving, especially at highway speeds.

The finding explains why patients taking insulin for type 1 diabetes have a 12%-19% higher risk of crashing their cars, compared with the general population. But in a larger sense, the study speaks to a new possibility as cars become smarter: monitoring drivers’ mental states and pulling over to the side of the road or otherwise taking control if there’s a problem.

The “results show that vehicle sensor and physiologic data can be successfully linked to quantify individual driver performance and behavior in drivers with metabolic disorders that affect brain function. The work we are doing could be used to tune the algorithm that drive these automated vehicles. I think this is a very important area of study,” said senior investigator Matthew Rizzo, MD, chair of the university’s department of neurological sciences in Omaha.

With funding from Toyota, his team placed a kind of black box inside the cars of 19 patients with type 1 diabetes and 16 diabetes-free controls who were of similar age and educational background. The box had a GPS and an accelerometer to detect and record hard turns, sudden stops, swerves, and other signs that something dangerous had happened. The cars were also rigged with video cameras that recorded both the driver and the view out the windshield.

Participants had the devices in their cars for a month, during which time the diabetes patients were also on continuous, 24-hour blood glucose monitoring. The investigators then synched the car data with the glucose readings, and compared it with the data from the controls’ cars. In all, the system recorded more than 1,000 hours of road time across 3,687 drives and 21,232 miles.

“What we found was that the drivers with diabetes had trouble,” Dr. Rizzo said at the American Neurological Association annual meeting.

Glucose was dangerously high or low about 13% of the time when people with diabetes were behind the wheel. Their accelerometer profiles revealed more risky maneuvering and variability in pedal control even during periods of euglycemia and moderate hyperglycemia, but particularly when hypoglycemia occurred at highway speeds.

One driver almost blacked out behind the wheel when his blood glucose fell below 40 mg/dL. “He might have been driving because he was not aware he had a problem,” Dr. Rizzo said. He is now; he was shown the video.

The team reviewed their subjects’ department of motor vehicles records for the 2 years before the study. All three car crashes in the study population were among drivers with diabetes, and they received 11 of the 13 citations (85%).

The technology has many implications. In the short term, it’s a feedback tool to help people with diabetes stay safer on the road. But the work is also “a model for us to be able to approach all kinds of medical disorders in the real world. We want generalizable models that go beyond type 1 diabetes to type 2 diabetes and other forms of encephalopathy, of which there are many in neurology.” Those models could one day lead to “automated in-vehicle technology responsive to driver’s momentary neurocognitive state. You could have [systems] that alert the car that the driver is in no state to drive; the car could even take over. We are very excited about” the possibilities, Dr. Rizzo said.

Meanwhile, “just the diagnosis of diabetes itself is not enough to restrict a person from driving. But if you record their sugars over long periods of time, and you see the kind of changes we saw in some of the drivers, it means the license might need to be adjusted slightly,” he said.

Dr. Rizzo had no relevant disclosures. One of the investigators was an employee of the Toyota Collaborative Safety Research Center.

[email protected]

SOURCE: Rizzo M, et al., ANA 2017 abstract number S131

 

SAN DIEGO – When researchers at the University of Nebraska placed sensors in the cars of patients with type 1 diabetes, they found something interesting: About 3.4% of the time, the patients were driving with a blood glucose below 70 mg/dL.

Almost 10% of the time, it was above 300 mg/dL, and both hyper and hypoglycemia, but especially hypoglycemia, corresponded with erratic driving, especially at highway speeds.

The finding explains why patients taking insulin for type 1 diabetes have a 12%-19% higher risk of crashing their cars, compared with the general population. But in a larger sense, the study speaks to a new possibility as cars become smarter: monitoring drivers’ mental states and pulling over to the side of the road or otherwise taking control if there’s a problem.

The “results show that vehicle sensor and physiologic data can be successfully linked to quantify individual driver performance and behavior in drivers with metabolic disorders that affect brain function. The work we are doing could be used to tune the algorithm that drive these automated vehicles. I think this is a very important area of study,” said senior investigator Matthew Rizzo, MD, chair of the university’s department of neurological sciences in Omaha.

With funding from Toyota, his team placed a kind of black box inside the cars of 19 patients with type 1 diabetes and 16 diabetes-free controls who were of similar age and educational background. The box had a GPS and an accelerometer to detect and record hard turns, sudden stops, swerves, and other signs that something dangerous had happened. The cars were also rigged with video cameras that recorded both the driver and the view out the windshield.

Participants had the devices in their cars for a month, during which time the diabetes patients were also on continuous, 24-hour blood glucose monitoring. The investigators then synched the car data with the glucose readings, and compared it with the data from the controls’ cars. In all, the system recorded more than 1,000 hours of road time across 3,687 drives and 21,232 miles.

“What we found was that the drivers with diabetes had trouble,” Dr. Rizzo said at the American Neurological Association annual meeting.

Glucose was dangerously high or low about 13% of the time when people with diabetes were behind the wheel. Their accelerometer profiles revealed more risky maneuvering and variability in pedal control even during periods of euglycemia and moderate hyperglycemia, but particularly when hypoglycemia occurred at highway speeds.

One driver almost blacked out behind the wheel when his blood glucose fell below 40 mg/dL. “He might have been driving because he was not aware he had a problem,” Dr. Rizzo said. He is now; he was shown the video.

The team reviewed their subjects’ department of motor vehicles records for the 2 years before the study. All three car crashes in the study population were among drivers with diabetes, and they received 11 of the 13 citations (85%).

The technology has many implications. In the short term, it’s a feedback tool to help people with diabetes stay safer on the road. But the work is also “a model for us to be able to approach all kinds of medical disorders in the real world. We want generalizable models that go beyond type 1 diabetes to type 2 diabetes and other forms of encephalopathy, of which there are many in neurology.” Those models could one day lead to “automated in-vehicle technology responsive to driver’s momentary neurocognitive state. You could have [systems] that alert the car that the driver is in no state to drive; the car could even take over. We are very excited about” the possibilities, Dr. Rizzo said.

Meanwhile, “just the diagnosis of diabetes itself is not enough to restrict a person from driving. But if you record their sugars over long periods of time, and you see the kind of changes we saw in some of the drivers, it means the license might need to be adjusted slightly,” he said.

Dr. Rizzo had no relevant disclosures. One of the investigators was an employee of the Toyota Collaborative Safety Research Center.

[email protected]

SOURCE: Rizzo M, et al., ANA 2017 abstract number S131