Part 2: Our Golden Retriever Stared At The Baby Monitor For Three Hours Every Night — When The Pediatric Sleep Specialist Looked At Our Footage, She Said My Daughter Would Have Died Within Six Months Without That Dog
I want to walk you through the morning I noticed it, because the noticing was the entire turning point.
It was a Wednesday in late July. I had set the GoPro up the night before because I was making a small video montage for my mother’s birthday — a sweet little compilation of Sunny being a Good Big Sister. I had imagined I would clip together her best moments and add some music and send it as a card.
I poured my coffee at six-thirty a.m. while Iris was still sleeping. I plugged the GoPro into my laptop. I opened the footage on fast-forward.
The first thirty minutes were what I expected. Sunny walked into the bedroom at 8:34 p.m. She lay down on the rug in front of my nightstand. She put her chin on the floor. She locked her eyes on the screen.
I sped the footage up to four-times speed. Sunny did not move for the next forty-eight minutes of compressed video. The screen flickered with the small black-and-white image of Iris asleep in her crib. My husband walked in at one point, kissed me on the head while I was reading, walked out. I turned off the lamp. The room went dark except for the screen.
Sunny stayed.
At time stamp 10:47 p.m. — about two hours and twelve minutes after Iris had been put down — Sunny’s body language changed.
It was so small that on fast-forward I almost missed it. Her ears went up. Her front paws tucked tighter under her chest. Her tail, which had been resting flat behind her, gave one slow thump on the rug.
Then she stood up.
She walked to the bedroom door. She walked down the hallway. She nosed the nursery door open. She walked to the crib. She stood at the side of the crib for sixty-three seconds — I counted, on the playback — and then she walked back to our bedroom and lay down in front of the monitor again.
I rewound the footage. I went back to 10:47 p.m. I played it at normal speed.
I watched the small five-inch screen of the monitor on my nightstand, in the corner of the GoPro frame, where my daughter was visible on her back in the crib.
For thirty-one seconds, between 10:47:14 p.m. and 10:47:45 p.m., Iris’s chest did not rise.
It was so subtle that on first viewing I thought it was the limit of the monitor’s frame rate. I rewound it again. I watched it again. It was not a frame rate issue. The blanket on her chest, which had been moving in a small steady rhythm for the previous twenty minutes, was completely still.
For thirty-one seconds.
Then her chest jerked. She gave one big breath. She kept breathing.
Sunny had stood up at 10:47:50 p.m., five seconds after Iris started breathing again.
I sat at my kitchen counter and I watched the rest of the footage in real time. Six hours. From 8:30 p.m. to 2:30 a.m.
It happened three times in six hours.
Three times, Iris’s chest stopped moving for between fifteen and forty seconds. Three times, Sunny stood up — not before, exactly, but in a window between two seconds before and twenty seconds after Iris started breathing again. Three times, Sunny walked to the crib.
I watched the third event in slow motion three times in a row.
At 2:11 a.m., Iris’s chest stopped moving for forty-one seconds. The longest of the night. Her small body was completely still on the screen. The monitor did not alarm. The Owlet sock on her foot did not alarm. No part of the technology we had been told to trust as new parents made a single sound.
At 2:11:39 a.m. — two seconds before Iris took her next breath — Sunny had not just stood up. She had stood up and walked to the bedroom doorway and barked once, very quietly, almost a huff, in the direction of the hallway.
Iris took a breath. Sunny stopped. Sunny walked down the hallway to the crib. Sunny stood there for ninety seconds. Sunny walked back. Sunny lay down in front of the monitor again.
I sat at my kitchen counter and I started shaking.
I drank my cold coffee in two swallows. I put the laptop in my purse. I called my pediatrician’s office at 7:08 a.m. and I told the receptionist — a woman named Charlene who I had spoken to fifty times in five months and who, by then, knew my voice — that I needed an emergency appointment with Dr. Park that morning.
Charlene heard something in my voice. She got me in at nine.
Dr. Park is a kind, deliberate man in his fifties who has been practicing pediatrics in St. Paul for twenty-six years. He had been my pediatrician at our four-week, two-month, four-month, and six-month visits. He had told me, more than once, in the patient way he tells everything, that I was a typical anxious new mother and that Iris was thriving.
I sat in his office with Iris in her car seat at my feet and my laptop on his desk.
I said, “Dr. Park. I am going to ask you to watch a video. I do not want you to tell me I am being an anxious mom. I want you to watch the whole video before you say anything.”
I played him the four key clips.
He watched them. He watched the third one twice.
When he was done, he leaned back in his chair and he was very quiet for a moment.
Then he said, “Hannah. I do not think you are being an anxious mom. I think we need to do a sleep study. Today, if I can get one. I am going to make some calls.”
He made the calls in front of me. He got Iris onto the cancellation list at the University of Minnesota Pediatric Sleep Center. He told me to keep my phone on.
We were called at 1:14 p.m. There had been a cancellation. They could see Iris that afternoon for an evaluation, with an overnight sleep study scheduled for Friday night.
The pediatric sleep specialist was a woman named Dr. Aisha Henderson. She was about my age, maybe a little older — late thirties, dark hair pulled back in a low ponytail, a pair of cat-eye reading glasses on a beaded chain. She had warm eyes and a calm face, and she shook my hand when I walked in, and she introduced herself to Iris in the car seat as if Iris could understand her, which I appreciated more than I knew how to say.
I gave her my laptop. I cued up the same four clips I had shown Dr. Park.
She watched them.
When she got to the third clip — the forty-one-second event at 2:11 a.m. — she paused the video about halfway through. She watched Sunny on screen, frozen in the act of standing up. She watched Iris on the small monitor inside the larger frame, frozen with her small chest motionless.
Then she set the laptop down on her desk. She turned in her chair. She looked at me.
She said, “Hannah. Your dog has been keeping your daughter alive.”
I started crying. I could not help it.
Dr. Henderson did not interrupt me. She handed me a tissue. She waited until I could breathe again.
Then she explained, very carefully, what she was looking at.
Iris, she said, almost certainly had a form of central sleep apnea — a condition in which the brain temporarily fails to send the signal to breathe during sleep. It is rare in infants who are otherwise healthy. It is, critically, the kind of apnea that the standard at-home monitors — the Owlet sock, the Wi-Fi baby monitor, even most hospital-grade pulse oximeters set at standard sensitivity — often miss, because the events are short, because they self-resolve, and because the baby’s oxygen saturation does not drop low enough or long enough to trigger the alarm thresholds.
She said, “Most children grow out of it. Some don’t. Some, in very rare cases, have an event that does not self-resolve, and we lose them. The medical community is still working out which kids are at high risk and which kids are not. The honest answer is that we don’t have a perfect test. The current best practice is in-lab polysomnography — a sleep study, which is what we’re going to do for Iris on Friday.”
She paused.
She said, “Hannah. I have been a pediatric sleep specialist for thirteen years. I have never — never — seen a dog do what your dog is doing on this footage. Goldens are, as a breed, sensitive to subtle changes in human respiration and body language. There is a small but real body of research on dogs alerting to seizure events in epileptic children. There is a much smaller body of research on dogs detecting medical events in infants. Most of what is in that research is anecdotal.”
She tapped the laptop.
She said, “This footage is not anecdotal. Your dog is responding to specific apneic events with consistent timing. Three out of three. Each event your dog responded to is a real event. I am going to take this footage, with your permission, to my colleagues. I would like to write about it, eventually, with your permission. But that is for later. Right now, we need to focus on Iris.”
I said, “Dr. Henderson. What were you going to say at the end. You stopped halfway through.”
She looked at me for a long moment.
She said, “Hannah. Statistically — and I want to be careful with this number, because it is not a number we use in clinical decision-making — but statistically, infants with untreated central apnea events of the duration we are seeing in your daughter, occurring three times in a six-hour window, have a non-trivial risk of a fatal event within the following six to twelve months. The odds are not a coin flip. They are far smaller than that. But they are not zero. And without treatment, every night is a roll of the dice.”
She paused.
She said, “Your dog has been rolling the dice for you. Three times a night. For five and a half months. She has not lost yet.”
Iris had her sleep study on Friday night.
She was wired up to seventeen sensors. She slept in a small white crib in a small white room while a technician watched her in real time on a screen down the hall. Daniel and I slept in a parents’ suite next door, on a fold-out couch, holding hands across the space between us in the dark.
Sunny was at home with my mother, who had flown in from Wisconsin for the week.
The study confirmed what Dr. Henderson had already known from my footage. Iris had nine apneic events over the course of an eight-hour sleep, ranging from sixteen seconds to forty-three seconds. Her oxygen saturation dropped to an average of 91 percent during events, with one event briefly hitting 86 percent. None of the events alone were severe enough to trigger most consumer-grade monitors. The cumulative pattern, however, was clinically significant.
Iris was started on a low-dose stimulant therapy and was scheduled for follow-up sleep studies every six weeks for the next year. We were given a hospital-grade pulse oximeter calibrated to a much lower threshold than the consumer ones we had been using. We were taught how to perform infant CPR, and we both took a certification class.
We were also told, very gently, by Dr. Henderson, that there was a real possibility Iris would grow out of the condition by age two, but that there was also a non-zero possibility she would not, and we would need to be vigilant.
She said, “Hannah. Daniel. The technology you have now is much better than what you had a week ago. But your dog is still going to be more sensitive than the technology. I would, with respect, listen to her.”
We listen to her. We have listened to her every night since.
I want to tell you what changed in our house after the diagnosis, because the changes matter.
We moved Sunny’s bed. We bought a soft, oversized memory-foam dog bed — the nice kind, the kind that costs more than I am comfortable admitting we paid for it — and we put it on the floor of the nursery, next to Iris’s crib. We did not banish her from our bedroom. We just gave her a softer post.
She uses it. Every night.
She still does her rounds. She still walks down the hallway to check on Iris in the night, even though now she only has to walk three feet from her own bed to do it. She still stands at the side of the crib. She still looks down. She still — sometimes, but less often now — gets up onto her hind legs to put her front paws on the crib rail and look in.
She has saved Iris’s life on at least six occasions that we can document with footage and pulse-oximeter data. There are probably more we cannot document.
She is six years old now. She has gone from sixty-five pounds to sixty-eight pounds because we feed her too well. She has gray hair coming in around her muzzle. She still carries the stuffed lamb in the morning, although the lamb now lives in Iris’s room and Iris carries it around in the afternoon while Sunny watches her with the patient attention of a much older dog.
There is one more thing I want to write down, because it has changed the way I think about my life and our family in a way I am still working out.
Dr. Henderson called me three months after Iris’s diagnosis to tell me she had presented an anonymized version of our case at a small pediatric sleep medicine conference in Boston. The response, she said, had been a mix of skepticism and intense interest. Several other clinicians had reached out afterward to share similar anecdotal cases — dogs in family homes who had appeared to alert to nocturnal events in infants and small children. None of them had had the clean documentation we had, because Iris was the rare case in which a parent had filmed the dog directly during the relevant time window.
A study is now being developed at the University of Minnesota, in collaboration with two other institutions, to look at canine sensitivity to subtle respiratory changes in pediatric patients. Sunny is one of the subjects. The principal investigator, who is a friend of Dr. Henderson’s, has a Lab named Cooper who does something similar with the investigator’s own asthmatic son.
Whatever the study finds, the science is going to take a long time. The clinical community is rightly skeptical. Anecdote is not data. A few dogs sitting in front of baby monitors in Minnesota and Massachusetts is not going to rewrite pediatric sleep medicine.
But Dr. Henderson said something to me on that phone call that I have been turning over in my head for months.
She said, “Hannah. I have been doing this for thirteen years. I trust technology. I trust science. I trust evidence. But I also know that there are things our species has co-evolved with dogs to detect that we are not going to have a sensor for in our lifetimes. Some of those things are about our breath and our blood and the way our bodies hum when they are healthy or quiet down when they are not. Dogs hear that hum. We do not. The question is not whether your Golden is special. The question is whether you were paying attention. Most parents would not have filmed her. Most parents would have thought her behavior was sweet and posted pictures to Instagram. You filmed her. That is the part that saved your daughter.”
I have thought about that a lot.
I do not know if it is true. I think I would have figured it out eventually. I think Sunny would have stood up enough times that one night I would have woken up and watched her and put it together. I think Iris was probably going to be okay either way, with or without the GoPro, with or without my anxiety, with or without the fact that I happened to choose Sunny over her birthday montage on a Wednesday morning in late July.
But I do not actually know that. None of us do. The honest truth is that we do not get to know what would have happened if we had not done the thing we did. We only get to know that we did it, and that on the other side of having done it, we are still here, all four of us, in our small house in Highland Park in St. Paul, on a Wednesday night in November of 2025.
Iris is twenty-one months old now.
She is talking. She is running. She climbs onto the couch and falls off the couch. She has a vocabulary of about sixty words. The two she uses the most are Mama and Sunny. Sunny came before Daddy, which Daniel pretends to be hurt about.
Her last sleep study, in October, showed continued mild central apnea events but with significant improvement in duration and frequency. Dr. Henderson is cautiously optimistic that Iris will outgrow the condition by age two, possibly age two and a half. We will keep doing studies every six months for the foreseeable future.
She still has the medical-grade pulse oximeter on her foot every night.
She also still has a Golden Retriever sleeping three feet from her crib.
The pulse oximeter is the technology. The Golden Retriever is the senior member of the security team.
I want to end this with one more image, because it is the image that keeps coming back to me.
Last Sunday night, I woke up at three in the morning to use the bathroom. The hallway was dark. I padded past the nursery door, which is left ajar at all times now, and I stopped and looked in.
Sunny was on her bed. Her eyes were open. She was watching the crib.
Iris was asleep on her back, breathing the slow even breath of a healthy toddler.
Sunny did not look at me when I stopped at the door. She did not need to. I was not the threat. I was the other backup. She and I were working the same shift, on the same flock, and we both knew it.
I went to bed. I slept until six.
I trust her with my daughter’s life.
I trust her with my daughter’s life because she has earned it. Three apneic events a night. For five and a half months. Until I finally caught up.
If this story moved you, follow the page — there are more like Sunny and Iris I haven’t told yet.
