What is AI sperm analysis and is it reliable? The level of evidence and limits of home sperm tests, electronic witnessing and digital lab technologies.
AI sperm analysis is the automatic assessment by software of the number, movement and shape of the sperm in a semen sample. Favourable results have been reported in some measurements, for example in the repeatability of the shape assessment; however, no effect on pregnancy or live birth has been shown, and the method needs to pass the laboratory's own validation. Your doctor interprets the result. Digital laboratory technologies such as electronic witnessing, unlike sperm measurement, are concerned with verifying that samples belong to the right person.
How is standard sperm analysis done?
Sperm analysis (spermiogram) is the examination of a semen sample under the microscope in the laboratory. Current international laboratory guidelines offer standardised, evidence-based procedures for the examination and processing of semen; the aim is to preserve comparability between laboratories. The results are reported within this standard framework, and their interpretation belongs to your doctor.
What do artificial intelligence and computer-assisted analysis look at in sperm?
Computer-assisted systems and artificial intelligence software aim to count sperm from the microscope image, to follow their movement and to classify their shape. Research has concentrated most on assessment of shape (morphology) and movement. The software's output is a measurement result; it does not make a diagnosis and does not explain the cause of a problem.
Under what conditions is the benefit of automated analysis assessed?
According to the current sperm morphology recommendations of an expert group, automated systems based on examination of cells after staining are viewed favourably, provided that the person using them is competent and the laboratory validates performance within its own setting. For systems that estimate the "percentage of typical forms" with software on an unstained sample, it is stated that there is not enough evidence, and the laboratory is expected to build a separate screening strategy. The same assessment states that the overall level of evidence is low.
A current study reviewing the use of artificial intelligence in male infertility has reported promising accuracy in areas such as shape and movement. The authors emphasise that the studies are small-sample, mostly retrospective and trained on particular populations, and that external validation and large, prospective studies are needed. Measuring technical accuracy is not the same as improving your pregnancy or live birth outcome.
Why does the laboratory need to carry out its own validation?
The success of software depends on the type of sample it was trained on and on the image conditions. For this reason it cannot be assumed automatically that a system that works well in one place will give the same result in another laboratory; in the recommendations mentioned above too, the laboratory's own validation of analytical performance and the competence of the person using it are counted as conditions.
Does a home test or a test done with a phone take the place of a spermiogram?
The result of home-type and smartphone-based tests does not take the place of a spermiogram done in the laboratory. If you have obtained a result with these tests, it is sensible to talk with your doctor about how it should be interpreted.
Has sperm selection with artificial intelligence been clinically validated?
The use of artificial intelligence in selecting sperm before microinjection (ICSI) is also being researched. Practices in this area are still at an early stage, and reviews in this area state that clinical validation is needed.
What does electronic witnessing verify?
In IVF and andrology laboratories, verifying that samples belong to the right person is a critical safety step. According to the current expert view on this subject, critical steps should be verified in real time by a second person and/or a certified electronic witnessing system. The same view also states that electronic systems must be selected, validated, staffed with trained personnel and audited regularly.
What can you ask your doctor about sperm analysis and sample safety?
- By which method was my spermiogram assessed, manually or computer-assisted?
- Does the laboratory validate automated or AI-assisted measurement in-house?
- How should I assess the result if I have done a test at home?
- How is the identity of the sample verified?
- How are my results and data stored and used?
Frequently Asked Questions
If a home sperm test comes out normal, do I need to see a doctor?
A home test result that looks normal cannot be regarded as assurance on its own and does not take the place of a laboratory spermiogram. Whatever the result, if you have a concern about fertility, it is sensible to see your doctor.
Does artificial intelligence remove errors in the spermiogram?
In the current recommendations on this subject, it is stated that automated systems are superior to manual analysis only in inter-observer repeatability of the percentage of normal forms and that no superiority has been shown in other measures. The same assessment also includes that some systems count abnormal forms as fewer than they are. Consequently the software can contribute to the reliability of the measurement, but it cannot be said to remove the possibility of error.
Does electronic witnessing completely prevent sample mix-ups?
According to the current expert view on this subject, electronic witnessing can lead to working alone and to complacency. At some stages, such as the ICSI or insemination step where gametes are mixed and identity verification before transfer, manual verification with a second person is needed; the disposal of gametes or embryos is also verified manually with a second person. The same view also states that deviations from protocol may be unavoidable in complex processes; for this reason the presence of the system cannot be regarded as sufficient on its own.
In what other areas is artificial intelligence being researched in male infertility?
According to a current mapping review, studies have also been done in areas such as predicting the likelihood of retrieving sperm surgically and predicting the outcome of IVF. The same review emphasises that the applications are still at an early stage and need to be validated.
Does the measurement by artificial intelligence take the place of a diagnosis?
No. The software's output is a measurement result; it does not make a diagnosis or explain the cause of a problem. Your doctor interprets it.
Does measurement accuracy show that the pregnancy outcome improves?
No. Technical accuracy in assessing shape and movement is not the same as an improvement in pregnancy or live birth outcome. No effect on clinical outcomes has been shown.
Does automated analysis give the same result in every laboratory?
No such assumption can be made. Performance depends on the sample type and image conditions; the laboratory's own validation and the competence of the person using it are required.
Is the evidence the same for stained and unstained samples?
No. Automated cell assessment after staining is viewed favourably under certain conditions. For systems that estimate the percentage of typical forms on an unstained sample there is not enough evidence; a separate screening strategy is also expected.
Is sperm selection with artificial intelligence an established practice?
The practices in this area are at an early stage. Its use in sperm selection before ICSI is being researched, and clinical validation is needed.
Does electronic witnessing measure sperm count or movement?
No. Electronic witnessing is concerned with verifying that the sample belongs to the right person. Sperm measurement and identity verification serve different purposes.
If there is an electronic system, are staff training and audit still needed?
Yes. Alongside the selection and validation of the systems, staff training and regular audit are also needed. The presence of the system is not sufficient on its own.
Does electronic witnessing completely remove verification by a second person?
No. Manual verification with a second person is needed at the stages specified, such as ICSI or insemination, identity verification before transfer, and the disposal of gametes or embryos.
Get Your Free Consultation
Fill out the form below and our team will contact you
