NIH Study Identifies Tools for Effective PCOS Diagnosis

Earlier this week, researchers at the National Institutes of Health published a systematic review in Frontiers Endocrinology, concluding that artificial intelligence (AI) and machine learning (ML) technology are highly accurate at diagnosing and classifying polycystic ovary syndrome (PCOS).

PCOS is one of the most common endocrine disorders that affect women of reproductive age. According to the Frontiers article, 4–20% of the global population is impacted by PCOS, with more than 66 million reported cases in 2019. The condition is characterized by hyperandrogenism, ovulatory dysregulation, and polycystic ovarian morphology.

PCOS has been linked to severe health conditions, including cardiovascular disease, infertility, and endometrial cancer. These comorbidities make diagnosing PCOS critical. With a proper diagnosis, providers can help patients manage the condition and its complications, reducing the disease burden.

“Given the large burden of under- and misdiagnosed PCOS in the community and its potentially serious outcomes, we wanted to identify the utility of AI/ML in the identification of patients that may be at risk for PCOS,” said Janet Hall, MD, senior investigator and endocrinologist at the National Institute of Environmental Health Sciences (NIEHS), part of NIH, and a study co-author, in the NIH press release. “The effectiveness of AI and machine learning in detecting PCOS was even more impressive than we had thought.”

Researchers estimate that, in 2020, under- and misdiagnosed PCOS contributed to $8 billion in healthcare spending.

PCOS diagnosis is typically based on clinical features, blood tests, and radiological findings. For example, providers may conduct a clinical      assessment to determine if patients have hormonal acne, excess hair growth, or irregular periods — trademark signs of PCOS. Additionally, blood tests may identify high testosterone levels, a symptom of the condition. Some patients may also undergo radiological testing that identifies cysts or increased ovarian volume linked to PCOS.

Despite the standard diagnostic protocols, PCOS can be difficult to diagnose as other conditions, including obesity, diabetes, and cardiometabolic disorders, may have similar features. However, this newest study reveals that AI and ML may ease the diagnostic burden and improve accuracy.

“PCOS can be challenging to diagnose given its overlap with other conditions,” said Skand Shekhar, MD, senior study author and assistant research physician and endocrinologist at the NIEHS. “These data reflect the untapped potential of incorporating AI/ML in electronic health records and other clinical settings to improve the diagnosis and care of women with PCOS.”

In the systematic review, NIH researchers reviewed studies across the past 25 years that used AI or ML to detect PCOS. Across 135 studies, the investigators focused on 31 of them.

“Across a range of diagnostic and classification modalities, there was an extremely high performance of AI/ML in detecting PCOS, which is the most important takeaway of our study,” said Shekhar. 

The study also found that the ten studies incorporating AI/ML with standardized diagnostic criteria from the NIH, Rotterdam criteria, or the Revised International PCOS classification had an 80–90% accuracy rate.

Leveraging AI/ML technology in PCOS diagnostics may reduce the time to treatment and disease burden.