Hannah LaChapelle spent years suffering from recurring shoulder pain that doctors couldn’t explain. Artificial intelligence and doctors at Baylor St. Luke’s Medical Center helped her find an answer.
Brett Coomer/Houston Chronicle
Hannah LaChapelle poses for a portrait in Houston, Wednesday, May 20, 2026. She’s now living without pain for the first time in years.
Brett Coomer/Houston Chronicle
When doctors couldn’t identify the cause of Hannah LaChapelle’s intense shoulder pain, she made one last-ditch effort to find an answer.
The Houston resident typed her symptoms into ChatGPT last summer and asked the artificial intelligence chatbot for a possible diagnosis.
READ MORE: Sign up for alerts from the Medical Mystery Files
Article continues below this ad
LaChapelle, 29, had been dealing with pain in her right shoulder since she was a teenager, but it worsened after she moved to Texas in 2023. The pain typically lasted for a full week each month and felt like someone was stabbing her.
“It felt like someone was in there with a hot poker, all inside my shoulder and up into my neck,” she said.
By summer 2025, the pain was lingering for three weeks at a time, and LaChapelle spent the rest of the month living in fear of its inevitable return. She was taking the maximum recommended dosage of ibuprofen to control the pain during the day and relying on NyQuil to fall asleep at night.
CONTACT US: Tell us about a medical condition that was difficult to diagnose
Article continues below this ad
LaChapelle knew it was a longshot to ask AI for help, so she was shocked when ChatGPT offered a diagnosis that matched all her symptoms.
But she was also discouraged, because the explanation reminded her of a conversation she had with her former doctor when she was 13 years old. That doctor said LaChapelle might have a chronic medical condition that is usually associated with pelvic pain, but there was nothing to do except manage the symptoms.
“I was like, ‘Oh, great,’” LaChapelle said. “I was probably sassy to ChatGPT. Like ‘So what do I do about this? Where do I go?’”
Fortunately, the chatbot had additional advice to offer. It directed her to doctors at Baylor College of Medicine who specialized in treating the chronic medical condition.
Debilitating shoulder pain
LaChapelle, a Massachusetts native, first noticed the shoulder pain when she was a teenager living in the Boston area.
Article continues below this ad
It began as discomfort, as if someone was pressing on her shoulder. It got worse when she exercised, but it didn’t affect the range of motion in her right arm.
A spinal specialist prescribed her gabapentin, a medication primarily used to control seizures and relieve nerve pain after shingles, but LaChapelle decided not to take it after looking into possible side effects. A physical therapist suggested she might have thoracic outlet syndrome, caused by pressure on blood vessels or nerves between the neck and shoulder.
MORE COVERAGE: A Houston woman thought she had dementia. Then a doctor saw something new in a scan of her brain.
The pain became more severe after LaChapelle moved to Texas in the fall of 2023. It became unbearable by the time she attended the Houston Livestock Show and Rodeo six months later.
She called her father, a paramedic who is married to a nurse, and her mother, who works at a medical spa, but they didn’t know what was wrong.
Article continues below this ad
“I have a picture on my phone of me pointing to (my shoulder), and I’m like, ‘I’m dying right now. This is what hurts. What do you think it could be?’” she said. “From there it was just all downhill.”
Searching for an answer
LaChapelle’s primary care doctor and gynecologist couldn’t pinpoint the problem, either, so she tried to live with the pain.
At first, she couldn’t sleep on her right side, and then it became difficult to sleep at all without taking NyQuil. Her breathing became labored, especially when she took deep breaths when she exercised.
The pain initially lingered a full week, but over time it lasted closer to two or three. She stopped planning trips because she worried they’d be ruined by constant shoulder pain.
Article continues below this ad
The pain also took a toll on LaChapelle’s mental health. Even when she wasn’t hurting, she lived with the anxiety of knowing it would return. And the discomfort made her irritable, which she believes made her difficult to be around.
Her roommate, Bri Caron, helped by getting pain medication or warming up heating pads for LaChapelle. The two have been roommates since they lived together in Boston and it was difficult for Caron to see her close friend in so much pain.
“It was really hard because it was the exact opposite of who I know her to be,” Caron said. “I knew something was wrong.”
Spotting a pattern
LaChapelle refused to give up on finding a diagnosis, so she kept a log on her phone of her symptoms and the dates when she was hurting. She started to see a pattern while reviewing her notes in the spring of 2025.
The shoulder pain seemed to be occurring at the same time each month, overlapping with her menstrual cycle. It got worse about six months after she stopped taking birth control when she first moved to Texas, but she never suspected that shoulder pain could be connected to her period.
“It took months to put together that it was following such a strong pattern,” she said.
Her gynecologist brushed it off as a coincidence, but LaChapelle couldn’t shake the feeling her symptoms might be related to her menstrual cycle. She decided to ask ChatGPT.
The chatbot offered only one possible diagnosis: thoracic endometriosis, a rare condition where endometrial tissue that is typically found in the uterus instead grows in the chest cavity.
When LaChapelle was 13, her doctor suggested the heavy bleeding she experienced during her period might be a sign of endometriosis. But the doctor told her there was nothing to be done except put her on birth control to manage the symptoms.
Things had changed over the 15 years since she’d spoken to the doctor, though. ChatGPT suggested that LaChapelle reach out to doctors at Baylor College of Medicine who specialized in treating endometriosis.
Hannah LaChapelle poses for a portrait in Houston, Wednesday, May 20, 2026. LaChapelle noticed the recurring shoulder pain that ailed her for years seemed to follow a pattern, so she asked artificial intelligence to help her find a cause that doctors struggled to identify.
Brett Coomer/Houston Chronicle
An underdiagnosed condition
LaChapelle’s story is common for women with thoracic endometriosis, who typically experience symptoms for five to seven years before they’re diagnosed, said Dr. Shanda Blackmon, a thoracic surgeon and director of the Lung Institute at Baylor.
Endometriosis is relatively common; the condition is a leading cause of infertility that affects roughly 10% of women during their reproductive years. But just one in eight women with endometriosis see it spread to their chest cavity, Blackmon said.
Blackmon suspects its true prevalence is higher, though, because many doctors are not aware of the disease or mistake it for other lung or breathing conditions. It’s also tricky to identify because the only definitive way to diagnose is by threading a camera into a patient’s chest.
READ MORE: A Spring teen was told months of severe pain was all in her head. Then a doctor took a closer look.
“It’s probably one of the most underdiagnosed diseases that I can think of,” Blackmon said.
Thoracic endometriosis occurs when cells similar to those that line the uterus grow in the chest, typically on the diaphragm or lungs. It’s unclear what causes the condition, but if left untreated it can lead to severe complications like breathing problems or a collapsed lung, Blackmon said.
LaChapelle felt pain in her shoulder due to what is known as referred pain, Blackmon said. The brain may interpret pain in one part of the body as coming from somewhere else that is connected to the same branch of nerves. The endometriosis irritated LaChapelle’s phrenic nerve, which runs from the neck to the diaphragm.
Thoracic endometriosis is often managed with hormone suppression to prevent the cells from growing. But LaChapelle was sick of living with pain, so she opted for surgery to remove the abnormal cells.
“It just felt more finite,” she said.
A novel surgery
Surgery for thoracic endometriosis has commonly involved the use of a laser to vaporize or cut away the abnormal cells, similar to the way surgeons treat many types of cancer. But that procedure risks leaving behind some cells that could grow back.
Blackmon instead proposed a novel surgery where she used a robot to cut out the abnormal cells. Robots are increasingly being used in a variety of surgeries because experts say they offer benefits such as improved precision and faster recovery times.
LaChapelle’s surgery is believed to be the first single-port robotic surgery for thoracic endometriosis performed in Texas, Blackmon said.
The surgery took place March 17. Blackmon cut small holes below LaChapelle’s ribcage and inserted a small robot into her chest. Blackmon maneuvered the robot upwards to cut out the abnormal cells and part of LaChapelle’s diaphragm.
MORE COVERAGE: A rare condition left a Houston student with seizures and psychosis. Why was her ‘brain on fire?’
Making the incisions below the ribcage instead of maneuvering through the ribs also minimized LaChapelle’s pain during her recovery. She was walking the halls of her apartment complex within 24 hours of the procedure and taking a Pilates class within one week.
“It was definitely much better than I anticipated,” she said.
Hannah LaChapelle poses for a portrait in Houston, Wednesday, May 20, 2026. Doctors at Baylor St. Luke’s Medical Center used a novel robotic surgery to help alleviate the recurring shoulder pain that ailed LaChapelle for years until artificial intelligence helped her identify its cause.
Brett Coomer/Houston Chronicle
A new tool for solving medical mysteries
LaChapelle is still pain-free nearly four months after her surgery. She’ll eventually need another procedure to remove the endometriosis in her pelvic area, but that procedure has not been scheduled.
Living without pain has also improved LaChapelle’s mood, said Caron, who has noticed the difference in her roommate since the surgery.
“She’s just back to her normal self,” Caron said. “She’s not always being consumed by this worry of having to go through pain soon.”
LaChapelle feels lucky that ChatGPT pointed her to the right diagnosis, because she knows it was probably a longshot that it would provide the right answer. She still prefers a doctor’s opinion, but she believes AI and the internet could be a useful starting point for those who are suffering from unexplained symptoms.
“I certainly wouldn’t want ChatGPT to do my surgery, that’s for sure,” she said.
Blackmon believes LaChapelle’s story is an example of how AI and the internet could help to identify rare conditions that are often relatively unknown, even within the medical community.
Chatbots are not a replacement for medical expertise, but their ability to scour the internet for information and offer a variety of potential solutions could be an invaluable tool for doctors who are struggling to diagnose their patients, she said.
“I know a lot of physicians complain that their patients get on ChatGPT and come in with a list of diagnoses that are wrong. But how frequently are they right?” she said. “We as physicians need to change our minds and start to listen.”