India’s Garbhini-GA2 Model: Accurately Predicting Delivery Dates

Garbhini-GA2

Researchers from BRIC-THSTI Faridabad, IIT Madras, and other institutes have created a new method to figure out the age of a developing baby (foetus) in pregnant Indian women during the second and third trimesters (months 4-9 of pregnancy).

Currently, doctors use a formula made for Western populations to estimate the age of an Indian foetus. But this formula becomes less accurate later in pregnancy because Indian babies grow differently than Western babies.

The new formula, called Garbhini-GA2, can estimate an Indian foetus’s age much more precisely in months 4-9. It reduces error by 3 times compared to the Western formula.

Knowing the foetus’s exact age is important to give pregnant women the right care and to plan the delivery date. The Garbhini-GA2 formula will help Indian doctors estimate age better.

What is Garbhini-GA2?

Garbhini-GA2 is a new computer model that can estimate how far along an Indian woman’s pregnancy is. It was created using data from a study of pregnant women in North India called GARBH-Ini.

The model can calculate the age of the foetus (gestational age) during months 4-9 of pregnancy. It uses 5 common measurements from ultrasound scans:

1) Biparietal diameter (BPD): Baby’s head width
2) Occipitofrontal diameter (OFD): Baby’s head length
3) Head circumference (HC): Baby’s head size around
4) Abdominal circumference (AC): Baby’s belly size around
5) Femur length (FL): Baby’s thigh bone length

The model was trained using over 2000 ultrasound scans from Indian women whose exact pregnancy date was known from the 1st trimester. By looking at multiple body parts and using advanced machine learning, Garbhini-GA2 can estimate gestational age specifically for Indian babies more accurately than previous models.

Research shows it reduces error by 3 times compared to formulas based on Western populations. Knowing the exact age helps give pregnant women the right care.

Funding of Garbhini-GA2

The GARBH-Ini study to create the Garbhini-GA2 model was backed by the Indian government. Funding came from:

• Department of Biotechnology (DBT), Government of India
• Grand Challenges India program under the Biotechnology Industry Research Assistance Council (BIRAC)
  Additional money was given by:
• Robert Bosch Centre for Data Science and Artificial Intelligence (RBC-DSAI)
• Centre for Integrative Biology and Systems Medicine (IBSE) at IIT Madras

Why is Garbhini-GA2 Important for Gestational Age Estimation?

In India, many pregnant women see a doctor for the first time later in their pregnancy. So early ultrasounds to date the pregnancy do not happen.

Existing models like Hadlock and INTERGROWTH are used to estimate the age later in pregnancy. But these were created using Western data. Indian and Western babies grow differently during pregnancy.

So these models give inaccurate results for Indian women in the later months. They can predict the wrong due date, lead to poor care, or incorrectly estimate risks like premature birth across India.

That’s why a customized model like Garbhini-GA2 is important. It is specifically tailored to how Indian babies grow based on their ultrasound measurements. Using India-specific data gives much more precise age estimates in late pregnancy.

How Does Garbhini-GA2 Improve Gestational Age Estimation?

Most models just match the baby’s size to estimate its age. But Garbhini-GA2 is different. It looks at how multiple body parts change together over time to make personalized predictions.

Testing showed Garbhini-GA2 was much more accurate than older formulas:

• It cut the median error in age estimates by over 40% versus Hadlock’s formula.
• It cut the error by 25% compared to the INTERGROWTH-21st formula.
• It halved the chance of mixing up whether a birth would be premature or full term.

What are the Implications of Using Garbhini-GA2 in Clinical Practice?

The higher precision of Garbhini-GA2 in estimating gestational age and predicting delivery dates can enable obstetricians to better time interventions and plan neonatal care based on the expected date of delivery.

At the population health level, Garbhini-GA2 can significantly improve estimates of preterm birth and other gestation-dependent adverse outcomes that inform public health policies and resource allocation.

How was it Developed and Validated?

The data to create Garbhini-GA2 came from a large study of pregnant women at hospitals across North India called GARBH-Ini. Over 2000 women participated.

In the first 3 months, each woman got an ultrasound to accurately date their pregnancy. This gave the gold standard to compare against.

80% of the GARBH-Ini data was used to train the Garbhini-GA2 model to estimate age based on later ultrasounds. When tested on the remaining 20% of women, Garbhini-GA2 was much more accurate than older Hadlock and INTERGROWTH formulas.

More validation across India is still happening to check if Garbhini-GA2 works well for women from different regions. But initial testing indicates it could substantially improve gestational age and prematurity estimates compared to current Western-based models.

The large Indian-specific dataset was key to tailor the algorithm to local fetal development patterns. This population-customized approach helps provide pregnant women personalized and precise care.