Our Take
The real story is how basic lab improvements delivered measurable gains while expanding IVF from infertility treatment to elective fertility preservation.
Why it matters
Healthcare AI teams building reproductive health tools need to understand these established baselines before claiming algorithmic improvements. The shift to preservation-focused services also opens new market segments beyond traditional infertility.
Do this week
Healthcare AI teams: audit your fertility prediction models against these documented success rate improvements before claiming novel performance gains.
Culture medium advances doubled IVF success rates
IVF success rates climbed from 12-15% in the early 1990s to 25% by the 2000s through extended embryo culturing (per Boston IVF's Alan Penzias). Early labs could only culture embryos for two days before transferring all of them to the uterus. Improved culture medium allowed embryos to survive five to six days outside the body, reaching 80-100 cells versus the original 2-4 cells.
Vitrification freezing replaced slower cooling methods about a decade ago, allowing embryos to survive freezing and thawing more reliably. This eliminated the need to transfer multiple embryos simultaneously, reducing twin and triplet pregnancies that increase complication risks.
Extended culturing also enabled genetic testing of embryos before implantation. Labs can now extract cells from 100-cell embryos and test them genetically before freezing, giving patients genetic readouts of all embryos before choosing which to implant.
Treatment became prevention and preservation
These lab improvements changed IVF's function beyond infertility treatment. People can now freeze eggs or embryos to delay parenthood, and cancer patients preserve fertility before treatments that might damage reproductive organs. Single IVF cycles can produce embryos used years apart for multiple children.
The technology also reduced medical risks. Vitrification gives patients recovery time between hormonal treatments, reducing ovarian hyperstimulation syndrome risk. Single embryo transfers became standard practice, cutting pregnancy complications from multiple births.
Scientists have preserved and successfully reimplanted ovarian and testicular tissue, enabling healthy births from recipients. The field now serves fertility preservation as much as infertility treatment.
Document baselines before building prediction tools
Healthcare AI teams working on fertility applications should establish these documented success rate improvements as baselines. Any algorithmic prediction model needs to demonstrate gains beyond the 25%+ rates achieved through improved lab techniques alone.
The expansion into fertility preservation creates new data sets and use cases beyond traditional infertility treatment. Genetic testing of embryos generates structured datasets that could support AI applications, but the testing technologies remain imperfect according to the source.
Teams building reproductive health tools should also account for the timing flexibility that vitrification introduced. Patient decision-making now spans multiple years rather than single treatment cycles, changing how prediction models should account for time-dependent variables.