About half the babies born too small or too large are missed at the 30-week scan.
When and how a baby is delivered rests on its estimated weight. Growth Intelligence reads the scans you already take and says which babies to scan again, and when.
Where care followed Growth Intelligence
Where its rescan was done, Growth Intelligence caught 58% of the babies the 30-week scan missed.1
Across the whole cohort
Across every missed baby, it caught 24%.
Counting only rescans clinicians ordered on their own.1
Fewer scans, earlier warning
Twenty-two rescans do the work of 340.
Two pregnancies
Same scans. Two ways to read them.
Two real pregnancies, scan by scan. The standard estimate gives a weight for that day. Growth Intelligence gives the weight at birth, how sure it is, and whether to act.3
How it speaks
It names the next scan.
When Growth Intelligence is sure, it says so with a number. When one more scan would settle the call, it names the week. When it would not, it says that too. At each scan, its flag raises no more false alarms than the standard estimate's own call.4
The clinician decides
Every call shows its reasoning. Disagree in one tap.
Say what you are doing instead, and why. Both calls are scored at birth.
Same scans, same workflow.
Read against pregnancies with known birth weights.
The decision stays yours.
Know before delivery.
Send 200 recent births. See which babies Growth Intelligence would have caught, and when.
- Validation cohort of 814 pregnancies with recorded birth outcomes, enriched for growth abnormality. Among the 499 with a full scan at 30 to 34 weeks, the standard estimate at that scan flagged 52 percent of babies born small and 43 percent of babies born large; it missed 122. Growth Intelligence projected each pregnancy to delivery, read growth rate, and asked for a rescan when the projected centile fell below the 25th or above the 75th. Of the 122 missed babies, 50 were in pregnancies it flagged that were rescanned at 35 to 38 weeks: 29 caught (58 percent). 18 were flagged and not rescanned: none caught. 54 were not flagged. Across all 122: 29 caught (24 percent), raising detection to 63 percent of small and 57 percent of large babies, for 1.5 percentage points more false alarms among babies not small and 2.2 among babies not large. Rescans were ordered by clinicians, not by Growth Intelligence, and rescanned pregnancies were small 1.71 times as often. Cross-validated. Internal analysis, not yet externally validated. Data on file, October 2026.
- Small babies only. 340 pregnancies scanned at 30 to 34 and again at 35 to 38 weeks. Rescanning all 340 found 41 of 71 small babies (58 percent), a median 1.3 weeks before birth, with 13 false alarms among 269. Growth Intelligence's rescan rule asked for 22 rescans and found 42 (59 percent), a median 4.4 weeks before birth, with 18 false alarms, the same as with no rescan. For large babies the rule asks for few rescans and adds one. Cross-validated; on the 19 held-out small babies the policies did not differ.
- Two real pregnancies from the cohort; names changed for privacy. Standard estimate: Hadlock (HC, AC, FL) weight and centile on the Hadlock 1991 chart, with the ±15 percent in grams that the ultrasound machine displays beside it, at each scan. Growth Intelligence: each pregnancy scored by a model fitted without it; 80 percent range for weight at birth. Size at birth on the INTERGROWTH-21st newborn standard.
- At each scan, the flag threshold is set to the false-alarm rate of the standard estimate's own 10th and 90th centile call. The rescans it asks for add the false alarms counted in note 1.