a day ago
How AI is revolutionising modern obstetrics and gynaecology
The use of AI in medicine has increasingly been heralded as promising change, one that is expected to make diagnosis faster, seamlessly integrate a patient's medical records across systems, read and decode images and lab results, help in medical modelling and even perhaps help surgeons.
Of the many fields in medicine that AI is already beginning to help in, obstetrics and gynaecology is one that it seems to be on the cusp of a revolution.
How is this happening?
AI is increasingly being used in gynaecology for various applications such as diagnosis, treatment planning, and patient care, says Abha Majumdar, director and head of Centre of IVF and Human Reproduction at Sir Ganga Ram Hospital, New Delhi. 'Currently, AI-powered tools are being used for tasks such aa image analysis, predictive modelling, and personalised medicine. In the next 5-10 years, we can expect AI to become even more integral to gynaecology, with potential applications in areas including robotic-assisted surgery, fertility treatment optimisation, and menopause management.'
Sunita Tendulwadkar, president of the Federation of Obstetric and Gynaecological Societies of India, explains that AI has become like a trusty assistant. 'AI is the stethoscope of the 21st century,' she says. Used wisely, she adds, it can allow doctors to deliver safer, more personalised women's healthcare, from city hospitals to the last tribal hamlet. She notes however, that 'while it amplifies our senses, it can never replace the mind or the heart of the physician'.
AI's applications
In one generation, says Dr. Tendulwadkar, doctors have moved from paper charts to algorithms that predict complications before they happen. Today, an AI model built from routine antenatal data can warn doctors of the risk of pre-eclampsia or postpartum haemorrhage weeks earlier than the old scoring systems, she explains.
Analysing data apart, ultrasound machines now come with 'auto-measure' buttons that capture standard foetal planes in seconds. She points out that systems such as SonoCNS can automatically segment the foetal heart or brain, label every chamber, and 'hand me precise biometrics while I'm still holding the probe'.
'Of course, I still review every image, but the heavy lifting is already done when I sit down to report,' she says.
AI is also useful in interpreting ultrasound images to check for abnormalities and assess risks when dealing with high-risk pregnancies. AI can also help predict the chances of a preterm birth, and assess the risk of complications such as high blood pressure and organ damage.
It can also help build personalised treatment planning for conditions such as PCOS and menopausal symptoms, says Dr. Majumdar.
Helai Gupta, head, department, obstetrics and gynaecology, Artemis Daffodils, Delhi, says AI is definitely a 'welcome tool'' in gynaecological practice, as it is far easier to rely on an AI bot to sift through huge amounts of literature while arriving at a diagnosis. Similarly, if you want to compute or collate large amounts of data, AI is a godsend, she adds. However, 'In practical clinical settings, AI has limited utility at least in my field,' she says.
Bandana Sodhi, director, obstetrics & gynaecology, Fortis La Femme, Delhi, adds that AI can help in reducing maternal/neonatal morbidity. 'Real-time AI-powered foetal monitoring continuously analyses foetal heart rate during labour and detects abnormalities promptly – leading to an 82 per cent reduction in stillbirths,' she says.
Is AI better at diagnosing cancer?
When it comes to mammograms, AI may be faster, but not necessarily better, says Dr. Helai. She explains that mammograms are designed to be sensitive and can give false positives with borderline cases. 'While AI may diagnose or 'undiagnose', and may do it faster, it may lack the necessary empathy and human skill required to ask for a second scan or put together patient-specific personal information that may alter simple reporting,' she says.
In comparison to an ultrasound screening test, which detected abnormal results less than 5% of the time, an AI neural network accurately recognises nearly 100% of anomalies associated with ovarian cancer, according to a paper published in the online journal Cancers published by MDPI. Large, randomised trials from Sweden and the UK show AI catching up to 20% more cancers without increasing false alarms, says Dr. Tendulwadkar.
'The direction is clear: very soon the standard will be one radiologist plus an AI safety net instead of two humans double-reading every film,' she says
AI in IVF
A couple in the United States who had an 'AI-assisted pregnancy' recently made headlines, but what is the science behind this? According to media reports, the couple who had been trying to conceive for 18 years, underwent several rounds of in vitro fertilization, or IVF. In the IVF process a woman's egg is removed and combined with sperm in a laboratory to create an embryo, which is then implanted in the womb. Their IVF attempts were unsuccessful however, due to azoospermia, a rare condition wherein instead of hundreds of millions of sperm in a sample there were no measurable sperms at all. Finally, at the Columbia University Fertility Center in USA, after hours of fruitless meticulous searching under a microscope for sperms in the husband's sample, AI was used, and it helped identify and recover three sperms which were then used to fertilise the wife's eggs. The women became the first successful pregnancy enabled under this novel 'STAR' method. The baby is due in December.
AI is being used to identify the most viable oocytes and embryos -- those with a high chance of leading to a pregnancy -- as well in selecting the correct timing of embryo transfer to the uterus.
Nandita Palshetkar, medical director and IVF specialist Bloom IVF, Lilavati Hospital, Mumbai, says, 'AI is giving us incredible new tools in IVF – helping us choose the best embryos, tailor treatments and improve success rates. It does not replace the doctor's judgement but supports it with sharper insights.'
Fertility consultant Ashwani Kale, director Asha Kiran Hospital and Asha IVF Centre, Pune, adds that better embryo selection with AI may soon replace the need for invasive testing.'
Risks and privacy concerns
Privacy and data concerns have also arisen when it comes to the use of AI. There is a critical need for responsible AI in gynaecology, says Dr. Majumdar, particularly when it comes to confidentiality. 'AI systems require access to sensitive patient data, which must be protected from unauthorised access and misuse. Patient consent is crucial for building trust in AI-powered healthcare solutions.'
Data shared on the internet lives forever, warns Dr. Hellai Gupta, reiterating the need to remove patient identifiers and private information to maintain confidentiality.
Dr. Tendulwadkar emphasises that AI tools have to be fine-tuned to Indian conditions and Indian women so that they respect 'our diversity in body habitus and disease patterns'.
Another potential threat Dr. Majumdar points to is the risk of bias in decision-making. 'If AI systems are trained on biased data, they may lead to unequal treatment and their outcomes for certain patient populations,' she says.
(Satyen Mohapatra is a senior journalist based in New Delhi. satyenbabu@
