For the 10% of women of reproductive age affected by endometriosis, the lack of advancement in diagnosis and treatment can feel frustrating. It is what prompted Noémie Elhadad, Ph.D., chair of the Department of Biomedical Informatics at Columbia, to launch a research initiative founded on the principles of citizen science known as Citizen Endo.
A collaboration between Columbia scientists and clinical experts, including Arnold Advincula, M.D., chief of gynecologic specialty surgery at NewYork-Presbyterian and Columbia, the project leverages digital health tools to collect patient-reported data to better understand the disease and inform new treatment approaches.
Dr. Noémie Elhadad
“I’ve been an endometriosis patient since age 13, so this research project is very dear to me,” Dr. Elhadad says. “We built this with patients. We involved them as partners in deciding which instruments to use for data collection and the questions to ask to obtain the right information.”
The heterogeneous nature of endometriosis means each patient can experience the disease in vastly different ways. There are also currently no known biomarkers associated with endometriosis; a diagnosis is typically confirmed after laparoscopic excision surgery to explore the pelvic region and remove any lesions or cysts. All this means that the average time to diagnosis from symptom onset can range from six to 15 years.
“A patient could have debilitating symptoms day-to-day but minimal endometriosis found in their pelvis. Conversely, a patient can be asymptomatic, but upon examination, you could find a significant amount of endometriosis,” says Dr. Elhadad. “That’s why I felt it was important to pay attention to the symptomatology, because we know that everything that has been described surgically about the disease was not always correlating perfectly with the patient experience.”
It was important to pay attention to the symptomatology, because we know that everything that has been described surgically about the disease was not always correlating perfectly with the patient experience.
— Dr. Noémie Elhadad
Understand Endometriosis Symptoms Through a Patient’s POV
At the heart of Citizen Endo is Phendo, a research app launched in 2016 that allows users to report and track their symptoms and what actions they take to manage their disease. It was developed with input from over 1,000 women who participated in focus groups, interviews, and surveys, as well as from information gathered by artificial intelligence from publicly available online endometriosis communities.
Currently, Phendo has over 21,000 users, and those who sign up understand it is for the purpose of advancing research on endometriosis and consent to their data being used. They complete daily questionnaires about their overall health status, symptoms, menstruation, sleep, diet, mood, exercise, and other reproductive and chronic conditions, along with any treatments they have tried. “When asking about daily experiences of disease, the questions need to resonate with patients while being formatted for scientific analysis,” says Dr. Elhadad. “That’s the balance we aimed for in the app.”
The Phendo app allows endometriosis patients to track and report their symptoms and self-management strategies.
The data collected through Phendo has informed 11 published papers that have offered new insights into the disease and its management. For instance, in a study published in npj Digital Medicine, the research team employed machine learning to analyze data collected from more than 776,000 self-tracked observations from 4,368 participants on factors that included pain, gastrointestinal and genitourinary symptoms, menstruation and bleeding patterns, daily living difficulties, sexual activity, medication usage, and quality of life, and they identified four endometriosis phenotypes characterized by the severity of signs and symptoms and the burden these had on participants’ lives.
The findings suggest that by identifying a patient's phenotype, clinicians can tailor treatments — for example, focusing on managing inflammation instead of pain — rather than relying on a trial-and-error approach.
This type of digital phenotyping more accurately reflects the complex pathophysiology of endometriosis, Dr. Elhadad says. “This showed us that it’s about more than just where the lesions are and which organs are affected. It’s about the severity and degree of inflammation. Endometriosis is systemic — it’s a whole-body disease.”
The Future of Endometriosis Research
Beyond the Phendo app, The Citizen Endo research team is leveraging AI to analyze electronic health records and claims data to help detect endometriosis earlier in the patient journey. “Our preliminary results indicate that, based on patterns of visits by a patient to a hospital or doctor, we can actually detect the disease two to three years in advance with impressive accuracy,” Dr. Elhadad says. “This could provide patients with validation of their diagnosis.”
They are also designing a new research project, dubbed Phendo Village, to understand the biology behind the disease. Previous twin studies suggest that approximately half of endometriosis patients have the disease due to genetic factors. “If your mother or sister experiences endometriosis, there’s a high likelihood that you will, too,” she adds.
If we understand the genetic variants involved, we could eventually make more accurate predictions about which treatment would be most effective.
— Dr. Noémie Elhadad
Dr. Elhadad and colleagues want to use trio sequencing, a form of genetic testing, to identify genetic variants associated with endometriosis. Under this model, researchers would analyze the DNA of a person with endometriosis alongside two biological relatives to identify rare, inherited variants or de novo genetic mutations that may contribute to the disease. “By examining the family structure, we can see what genetic traits are transmitted from parents to children and which arise as new mutations,” she says.
She believes this research could lead to more precise therapeutics. “We know that a one-size-fits-all approach doesn’t work for endometriosis,” Dr. Elhadad says. “Currently, first-line treatment is hormonal birth control, but we can’t predict who it will work for. If we understand the genetic variants involved, we could eventually make more accurate predictions about which treatment would be most effective.”
The research team is currently seeking funding for the trio sequencing study. In the meantime, they will continue analyzing patient-reported data from the Phendo app to improve understanding of the patient journey and enhance the patient-provider relationship.
“The biggest lesson we’ve learned so far is that providers and patients want to collaborate but often misalign on self-management goals,” says Dr. Elhadad. “This presents an opportunity to use technology to improve communication and recommend self-management strategies.”