Functional Development and Clinical Validation of a Diagnostic Tool Based on Artificial Intelligence for the Assessment of Sperm Quality and the Selection of the Optimal In Vitro Fertilisation (IVF) Treatment

Infertility is a growing global health problem affecting millions of couples worldwide, with male infertility accounting for approximately half of all cases. In the physiological environment, sperm go through an exhaustive selection process in the female reproductive tract before reaching the oocyte. During this journey, progressive mobility and morphology are key parameters for achieving fertilisation. Therefore, before starting an assisted reproduction treatment, it is essential to analyse and process the semen sample to assess the fertile potential, select the most optimal sperm and determine the most appropriate treatment.

Conventional methods of semen processing, such as density gradient centrifugation (DGC) and Swim-up washing of motile sperm, have significant limitations. These include interobserver and interlaboratory subjectivity, as well as damage to sperm DNA caused by centrifugation. Alternatively, microfluidics, which simulates natural selection, allows higher counts of morphologically normal, progressive motile sperm to be obtained. On the other hand, the CASA (computer-assisted sperm analysis) system has improved the standardisation and quality of semen analysis. Furthermore, the incorporation of Artificial Intelligence (AI) into semen quality analysis represents a promising opportunity, as it improves efficiency, accuracy and standardisation, and has the potential to increase success rates in assisted reproduction treatments.

This project aims to develop an innovative AI-based diagnostic tool to address male infertility. The tool will integrate microfluidic technology and the CASA system to analyse semen quality, calculate fertilisation potential and recommend personalised treatments with an estimate of success. Trained with large volumes of biological and clinical data, it will provide a comprehensive and patient-specific diagnosis by identifying complex relationships between multiple variables. Finally, a comparative study will be conducted to evaluate laboratory indicators and clinical outcomes of cycles using this tool versus those using conventional methods.