By Aayush Agarwal, Ph.D., Senior Clinical Embryologist Medically reviewed by Dr. Shweta Agarwal, MBBS, DGO Last updated: June 2026
Information on this page is educational and does not replace a medical consultation. Outcomes depend on individual clinical factors.
Aansh Hospital & IVF Center is a government-registered Level-2 ART clinic (Reg. No. MH/AC/2024/15441/L2/Chandrapur/132), with our headquarters and in-house embryology lab in Chandrapur. Our ART registration can be verified at the National ART & Surrogacy Registry. Embryology is led by Aayush Agarwal, Ph.D., Senior Clinical Embryologist. Clinical consultations are led by Dr. Shweta Agarwal, MBBS, DGO. This page explains how embryo selection works — both conventional and AI-assisted — so that patients can understand what they are being offered and ask the right questions.
I wrote this guide because I regularly field questions from patients who have read or heard the phrase "AI embryo selection" and are unsure whether it means the embryologist is no longer involved, or whether it guarantees a better outcome. Neither is true. What follows is a direct explanation of both approaches, what they can and cannot do, and how they fit together in a well-run IVF lab.
What is conventional embryo grading, and how does it work?
Conventional morphological grading is the long-established method by which a clinical embryologist evaluates embryos for transfer selection. It relies on direct microscopic observation at defined time points during embryo development — typically at Day 3 (cleavage stage) and Day 5 or 6 (blastocyst stage).
At Day 3, the embryologist assesses:
- Number of cells (ideally 6–8 for Day 3)
- Cell size regularity (even or uneven)
- Percentage of fragmentation (cellular debris — lower is better)
- Presence of multinucleation
At the blastocyst stage (Day 5–6), the embryologist uses a standardised scoring system — typically the Gardner/Schoolcraft classification — assessing:
- Expansion stage (1 through 6, with 6 fully hatched)
- Inner cell mass (ICM) quality: A (prominent, tightly packed), B (fewer, loosely grouped), C (very few cells)
- Trophectoderm quality: A (many cells forming a cohesive layer), B (few cells), C (very few, large, loose cells)
A blastocyst graded 4AA, for example, is an expanded blastocyst with excellent ICM and trophectoderm — typically the first choice for transfer if available. A 3BB is an expanding blastocyst with good but not top-tier quality in both cell masses.
The embryologist makes a judgment call based on the full picture of all available embryos, the patient's clinical history, and the transfer plan — fresh transfer or freeze-all. Morphological grading is well-validated: it correlates meaningfully with implantation outcomes and is the foundation of embryo selection worldwide.
Its limitation is that it is a snapshot assessment, made at specific time points, by visual inspection. It does not capture the developmental kinetics — the speed and pattern by which the embryo divided from fertilisation to blastocyst — unless the embryologist was at the incubator at every division point, which is impractical in a busy lab.
What is AI-assisted embryo selection, and what does it actually do?
AI-assisted embryo selection tools add a computational layer to the grading process. Most current AI embryo selection systems work in conjunction with time-lapse incubation systems — incubators that contain a built-in camera and capture images of embryos at regular intervals (sometimes every five to fifteen minutes) throughout culture.
The AI system analyses the time-lapse image sequence and scores each embryo based on parameters that correlate with developmental quality and implantation likelihood in the training dataset. These parameters typically include:
- Timing of cell divisions (e.g., time from fertilisation to 2-cell, 4-cell, 8-cell, and blastocyst stages)
- Division patterns (direct cleavage, reverse cleavage, multinucleation events)
- Morphological measurements derived from images (cell symmetry, ICM compactness, trophectoderm pattern)
The AI generates a score — sometimes called a viability score or ranking score — for each embryo. The embryologist then reviews the AI-generated rankings alongside their own morphological assessment and makes the final selection decision.
The key word is "alongside." AI is a decision-support tool. It gives the embryologist an additional layer of quantitative information — particularly about kinetic parameters that conventional observation at static time points would miss — and it may help differentiate between embryos that appear morphologically similar under conventional grading. The embryologist retains clinical judgment over the final decision.
Aansh has adopted AI-assisted embryology as part of its laboratory workflow.
An AI-assisted embryology system supports the embryologist's assessment by providing an AI-derived score based on time-lapse and morphological data. Our embryology lab is led by Aayush Agarwal, Ph.D., who integrates these scores into the selection decision alongside conventional grading for each cycle.
What can AI-assisted selection do, and what can it not do?
Understanding both sides is important for informed consent and realistic expectations.
What AI-assisted selection can do:
- Analyse continuous time-lapse data across the full culture period — capturing kinetic parameters that static observation misses
- Generate reproducible, quantified scores based on a consistent model — reducing the inter-observer variability that can occur between embryologists at different time points
- Help rank embryos when several appear morphologically similar under conventional grading
- Provide the embryologist with an additional data point for the transfer decision
What AI-assisted selection cannot do:
- Guarantee implantation or pregnancy — embryo selection is one of many factors in an IVF outcome
- Assess chromosomal status — AI morphological tools do not replace preimplantation genetic testing (PGT-A) when chromosomal screening is clinically indicated
- Compensate for poor embryo quality — if all available embryos score poorly, AI cannot create a better outcome than the biology allows
- Function without an experienced embryologist — the AI system requires interpretation within the clinical context of the patient's cycle, history, and lab conditions
The fundamental biological reality is this: even morphologically excellent embryos — graded highly by both conventional and AI methods — do not implant in every cycle. Implantation is the result of embryo quality, endometrial receptivity, and factors that neither conventional grading nor AI tools can fully predict. Selection methods improve the probability of choosing the best available embryo; they do not eliminate biological uncertainty.
Comparison: conventional grading vs AI-assisted selection
| Feature | Conventional Morphological Grading | AI-Assisted Selection |
|---|---|---|
| What it evaluates | Cell number, size, fragmentation, ICM/TE quality — at defined time points | Time-lapse kinetics + morphological parameters — continuously across culture |
| How embryologist is involved | Central — embryologist makes all observations and decisions | Central — AI provides a score; embryologist makes the final decision |
| Kinetic data (timing of divisions) | Captured only if embryologist observes at each division — impractical across all embryos continuously | Captured automatically from time-lapse images throughout culture |
| Chromosomal status | Not assessed — separate PGT-A test required | Not assessed — separate PGT-A test required |
| Reproducibility | Subject to some inter-observer variability | Consistent scoring model applied across embryos |
| Outcome guarantee | None — improves selection probability; does not guarantee implantation | None — improves selection probability; does not guarantee implantation |
| Availability | Standard at all IVF labs | Requires time-lapse incubator and AI platform; not universal |
| Replaces embryologist? | No | No |
How does the embryologist's role change — or stay the same?
AI-assisted tools change the information available to the embryologist; they do not change who is responsible for the selection decision. An experienced embryologist using AI output is similar to a radiologist using AI-assisted image analysis: the tool flags patterns and generates a score; the clinician integrates that information with the full clinical picture.
What does not change: the embryologist's knowledge of the patient's ovarian response, the lab environment during that culture period, the clinical plan for fresh versus frozen transfer, the number of embryos available, and the couple's personal preferences about multiple embryo transfer. AI scores are one input into that multi-factor decision.
What changes: the embryologist has access to kinetic data they would not otherwise have — timing of cell divisions captured automatically across five days of culture. This is genuinely additional information that can help differentiate between embryos that look visually similar at a static time point.
At Aansh, the embryology lab is led by Aayush Agarwal, Ph.D., who oversees both the conventional grading process and the integration of AI-derived scores into the selection framework. For more on blastocyst-stage culture and transfer decisions, see the blastocyst culture page.
Should all IVF patients seek AI-assisted embryo selection?
Not necessarily — and it would be misleading to suggest AI is required for a good IVF outcome.
For couples with a single embryo (the selection question is straightforward), couples pursuing PGT-A (chromosomal testing directly informs selection), or couples where clinical, cost, or access factors make time-lapse AI systems unavailable, conventional grading by an experienced embryologist remains the standard of care. A large body of IVF outcomes worldwide is built on conventional morphological grading.
AI-assisted selection is most useful when:
- Multiple embryos are available and are morphologically similar — AI kinetic data helps differentiate
- The lab has a time-lapse incubator system integrated with the AI platform
- PGT-A is not being performed (because PGT-A provides a direct chromosomal result that takes precedence in selection)
If you are considering IVF and want to understand which selection approach will be used for your cycle, and why, this is a direct question to ask the embryologist during your consultation. Ask: "Will my embryos be cultured in a time-lapse incubator? Will AI scoring be used, and how will it be integrated with your conventional grading?" The answer should be specific and clear.
To discuss your situation: WhatsApp or call +91 80056 85160.