How a machine learning tool could predict postpartum depression risk sooner


Postpartum depression is a mood disorder that can occur in the weeks and months after giving birth. It goes beyond the baby blues, experts say. — Pixabay

After giving birth, new mothers typically return for a follow-up appointment six to eight weeks later.

But if postpartum depression symptoms emerge, “that can be a really long time for someone who is suffering,” said Dr Mark Clapp, a maternal-fetal medicine provider at Massachusetts General Hospital.

New research pioneered by Clapp and other Mass General Brigham researchers aims to detect the risk of postpartum depression earlier, starting in the delivery room after the baby is born.

They’ve developed a simple machine learning model to be used before a new mother is even discharged from the hospital.

Postpartum depression is a mood disorder that can occur in the weeks and months after giving birth. It goes beyond the “baby blues,” experts say.

Symptoms can include depressed mood or severe mood swings; excessive crying; difficulty bonding with a baby; feelings of hopelessness or worthlessness; reduced ability to think clearly; loss of appetite and interest in activities; and thoughts of death or suicide, among others.

Perinatal and postpartum mood and anxiety disorders have historically been stigmatised, and those suffering from them often say it can be difficult to talk about.

Refrigeration redefined

In Massachusetts, the high-profile tragedy of Lindsay Clancy, a Duxbury mother who is charged with killing her three children in 2023, has elevated the topic in recent years.

An attorney representing Clancy, who attempted to take her own life after her children’s deaths, has said that she suffered from severe mental illness at the time, was overmedicated and grappled with postpartum depression – and potentially postpartum psychosis.

Clancy worked as a labour and delivery nurse at Massachusetts General Hospital.

“We want to be able to identify and treat depression as early as possible,” said Dr Andrea Edlow, also a maternal-fetal medicine specialist at Massachusetts General Hospital and a researcher on the study. “It’s like many other kinds of illness – by treating earlier, we can prevent it from getting worse, and we can try to minimise the impact it has on people’s lives."

The research is also in line with state objectives. Last summer, Gov. Maura Healey signed a major maternal health bill, which, among other things, established a task force on maternal health access and birthing patient safety.

State law also requires that the Department of Public Health collect annual data on screening for postpartum depression from providers.

‘If we can identify it, we can treat it’

Last month, the Mass General Brigham researchers published their study in the American Journal of Psychiatry, demonstrating how the machine learning detection tool can help identify patients at the highest risk for postpartum depression.

By doing so, the hope is that they can facilitate prevention, ongoing connection with health care providers, and, if necessary, treatment.

And it is treatable, Edlow emphasised.

“If we can identify it, we can treat it,” she said. “There are multiple well-established and safe treatments for depression.”

Providers typically administer the Edinburgh Postnatal Depression Scale, a 10-item questionnaire, during a postpartum follow-up visit. But the growing concern, Clapp said, is that a weeks-later appointment is too late. And statistically, the visit return rate is between 60% and 70%.

“By default, if we’re using screening at a postpartum visit as our primary means of detecting postpartum depression, we’re already missing 30-40% of people,” Clapp said.

The hospital after delivery is a “unique capture point” to do so, he added, given that the vast majority of pregnant people give birth in hospitals today.

In testing the prediction model, researchers used electronic health record data, sociodemographic factors and prenatal depression screening information to establish which patients could be at risk for ultimately developing postpartum depression.

The model weighs and integrates the complex variables in order to establish risk.

According to the study’s results, the model was effective in ruling out postpartum depression in 90% of cases, while also showing promise in predicting it. Nearly 30% of those predicted to be high risk developed postpartum depression within six months of delivery.

The study excluded patients with a previous psychiatric diagnosis to determine if the model could predict depression in patients who – at face value – are considered low-risk.

Currently in the test phase, researchers will now be tasked with figuring out how a machine learning tool as such can be implemented in real-time clinical practice.

“Many preventable deaths occur by suicide in the postpartum period,” Clapp said. “I think us as a specialty is really working on making this a priority.” – masslive.com/Tribune News Service

 

Follow us on our official WhatsApp channel for breaking news alerts and key updates!

Others Also Read


Want to listen to full audio?

Unlock unlimited access to enjoy personalise features on the TheStar.com.my

Already a subscriber? Log In