According to a comprehensive study conducted by the National Institutes of Health (NIH), Artificial Intelligence (AI) and Machine Learning (ML) technologies have proven highly effective in identifying and diagnosing Polycystic Ovary Syndrome (PCOS), a widespread endocrine disorder in women.
The NIH study examined a quarter-century of data, revealing the potency of AI and ML in diagnosing this prevalent hormonal disorder.
The research by the National Institutes of Health (NIH) systematically evaluated published scientific literature that utilized AI and ML for the assessment and categorization of PCOS. It revealed that programs based on these technologies were highly successful in diagnosing the condition, commonly affecting women between the ages of 15 and 45.
Janet Hall, M.D., a senior investigator and endocrinologist at the National Institute of Environmental Health Sciences (NIEHS), part of NIH, and a co-author of the study, stated, “Given the widespread instances of misdiagnosis or under-diagnosis of PCOS and its potential severe consequences, the aim was to assess how effective AI and ML could be in pinpointing patients potentially at risk for PCOS. The performance of AI and ML in this regard exceeded our expectations.”
Table of Contents
The Complexities in Diagnosing PCOS
PCOS manifests when the ovaries malfunction, frequently leading to elevated testosterone levels. The condition can result in a variety of symptoms including irregular menstrual cycles, acne, excessive facial hair, or hair thinning. Additionally, women suffering from PCOS are more susceptible to Type 2 diabetes as well as sleep, psychological, cardiovascular, and other reproductive disorders like uterine cancer and infertility.
Skand Shekhar, M.D., the study’s senior author and assistant research physician and endocrinologist at the NIEHS, noted, “The diagnostic process for PCOS can be intricate due to its symptomatic similarity with other conditions. The data highlights the unrealized potential of incorporating AI and ML into electronic health records to enhance the diagnosis and treatment of women afflicted with PCOS.”
The authors recommend the amalgamation of broad population-based research with electronic health records, along with the analysis of standard laboratory tests to unearth sensitive diagnostic markers that could ease the identification of PCOS.
Standard Criteria and the Contribution of AI/ML in Diagnosis
PCOS diagnosis adheres to globally recognized criteria, which have evolved over time. These typically include clinical symptoms like acne, excessive hair growth, and irregular menstruation, supplemented by laboratory tests such as elevated blood testosterone and radiological evidence like multiple small cysts and an increased ovarian volume in ultrasound studies. However, due to symptom overlap with conditions like obesity, diabetes, and cardiometabolic disorders, the disorder often remains undiagnosed.
AI and ML technologies serve to mimic human cognitive processes to aid decision-making or forecasting. AI’s capacity for processing enormous and disparate data sets, such as those obtained from electronic health records, renders it a valuable tool in diagnosing disorders that are challenging to identify, like PCOS.
Synopsis of Study Findings
The research encompassed a systematic review of peer-reviewed studies spanning the past 25 years (1997-2022) that employed AI and ML for PCOS detection. With assistance from a seasoned NIH librarian, a total of 135 studies were screened, out of which 31 were included. The studies were observational in nature and evaluated the application of AI and ML technologies in patient diagnosis. Approximately half the studies incorporated ultrasound imagery. The mean age of study participants was 29 years.
Among studies utilizing standardized diagnostic criteria, the accuracy in detecting PCOS ranged from 80% to 90%.
Shekhar concluded, “Our study underscores the exceptional efficacy of AI and ML technologies in identifying PCOS. These programs hold significant promise in enhancing early diagnosis, thereby leading to cost reductions and decreasing the healthcare burden associated with PCOS.”
Subsequent studies incorporating rigorous validation procedures will facilitate the seamless integration of AI and ML technologies for managing chronic health conditions.
Reference
The study titled, “Application of machine learning and artificial intelligence in the diagnosis and classification of polycystic ovarian syndrome: a systematic review” was published on September 18, 2023, in Frontiers in Endocrinology. DOI: 10.3389/fendo.2023.1106625
This research was funded by the Intramural Research Program of the NIH/National Institute of Environmental Health Sciences (ZIDES102465 and ZIDES103323).
Frequently Asked Questions (FAQs) about Artificial Intelligence in PCOS Diagnosis
What is the main focus of the National Institutes of Health’s study?
The main focus of the study conducted by the National Institutes of Health (NIH) is to assess the efficacy of Artificial Intelligence (AI) and Machine Learning (ML) in diagnosing Polycystic Ovary Syndrome (PCOS), a prevalent endocrine disorder in women.
Who conducted this study and who are the key contributors?
The study was conducted by the National Institutes of Health (NIH). Key contributors include Janet Hall, M.D., a senior investigator and endocrinologist at the National Institute of Environmental Health Sciences (NIEHS), and Skand Shekhar, M.D., the study’s senior author and an endocrinologist at the NIEHS.
What are the challenges in diagnosing PCOS?
Diagnosing PCOS is complex due to its symptomatic overlap with other conditions such as obesity, diabetes, and cardiometabolic disorders. Additionally, the disorder often manifests through a variety of symptoms including irregular periods, acne, and elevated testosterone levels, further complicating diagnosis.
What role do AI and ML play in the diagnosis of PCOS?
AI and ML technologies have shown high efficacy in identifying and diagnosing PCOS. Their ability to process large and disparate data sets, like those derived from electronic health records, makes them valuable tools in diagnosing disorders that are often difficult to identify.
What is the suggested future direction for improving PCOS diagnosis using AI and ML?
The study authors recommend integrating broad, population-based research with electronic health records. They also suggest analyzing standard laboratory tests to identify sensitive diagnostic biomarkers for easier identification of PCOS.
How accurate have AI and ML proven to be in diagnosing PCOS?
Among studies that used standardized diagnostic criteria for PCOS, the accuracy of AI and ML technologies ranged from 80% to 90%.
What are the potential benefits of using AI and ML for diagnosing PCOS?
The use of AI and ML technologies has the potential to significantly improve early diagnosis, thereby reducing healthcare costs and the overall burden of PCOS on both patients and the healthcare system.
What types of studies were reviewed for this research?
The research encompassed a systematic review of peer-reviewed studies spanning the past 25 years that employed AI and ML for PCOS detection. In total, 135 studies were screened and 31 were included in the report.
Is there any follow-up research planned?
The article indicates that subsequent studies incorporating rigorous validation procedures will facilitate the seamless integration of AI and ML technologies for managing chronic health conditions, although specific follow-up research plans are not detailed.
More about Artificial Intelligence in PCOS Diagnosis
- National Institutes of Health Official Website
- Overview of Polycystic Ovary Syndrome by Mayo Clinic
- Introduction to Artificial Intelligence and Machine Learning
- Frontiers in Endocrinology Journal
- NIH/National Institute of Environmental Health Sciences Research Programs
