Blastocyst AI Assisted Morphokinetics in ART

BACKGROUND Assisted reproductive technologies encompass a range of medical and laboratory procedures designed to facilitate fertilization and embryo implantation when natural conception is not possible or practical. Despite continuous technological progress, successful pregnancy following embryo transfer remains uncertain. Registry data demonstrate that implantation and live birth rates remain moderate rather than guaranteed outcomes. For example, data reported by the Society for Assisted Reproductive Technology (SART) demonstrated an overall clinical pregnancy rate of approximately 45.7% per transfer in the United States in 2020, while European registry data published by the European Society of Human Reproduction and Embryology (ESHRE) reported pregnancy rates of approximately 34% per embryo transfer cycle. Given these limitations, improving embryo selection represents one of the most important opportunities to enhance ART efficiency. Identification of embryos with the highest developmental competence could reduce the number of treatment cycles required to achieve pregnancy, minimize patient exposure to repeated procedures, and decrease overall societal and healthcare expenditures. Consequently, substantial research efforts have focused on developing objective biomarkers capable of predicting embryo implantation potential. Currently, embryo selection in most IVF laboratories relies predominantly on morphological assessment at the blastocyst stage. Key parameters include blastocoel expansion, trophectoderm cell appearance, and inner cell mass organization, as described in standardized grading systems such as the Gardner classification. Although morphology correlates with implantation potential, assessment depends on visual interpretation by trained embryologists and therefore remains inherently subjective and operator-dependent, limiting reproducibility across centers. The present observational study is designed to generate clinical and technical performance data necessary to support hypothesis development and future regulatory evaluation of an embryo image-assessment technology. Clinical performance evaluation must demonstrate reasonable assurance of safety and effectiveness in predicting embryo developmental potential. Device performance will therefore be evaluated through measures including classification accuracy, sensitivity, specificity, positive predictive value, and negative predictive value at both embryo and patient levels. Evaluation of the AI-based morphokinetic analysis tool will include verification, validation, and risk assessment processes consistent with FDA expectations for software contained in medical devices. EMBRYO IMAGING Embryo imaging has been incorporated into in-vitro fertilization (IVF) laboratories since the earliest development of assisted reproductive technologies. Initial documentation methods relied on photographic film attached to light microscopes, producing static images that were subsequently developed as slides or printed photographs for clinical records and educational purposes. Analog imaging systems, including thermal paper printers adapted from ultrasonography platforms, were later adopted to facilitate routine embryo documentation. Prior to the digital era, real-time visualization of gametes and embryos was achievable through analog video microscopy systems, with recordings stored using conventional videotape technology. The transition to digital microscopy represented a major technological advancement, allowing rapid acquisition, storage, and retrieval of high-resolution embryo images and videos. Digital capture systems improved laboratory documentation practices and enabled retrospective evaluation of embryo development without substantially altering standard culture conditions. Subsequently, time-lapse incubation (TLI) systems were introduced into IVF laboratories with the objective of enabling continuous embryo monitoring while maintaining uninterrupted culture conditions. These systems integrate microscopy and imaging software within incubators, allowing sequential image acquisition throughout preimplantation development. Investigators proposed that morphokinetic parameters derived from time-lapse imaging could identify embryos capable of achieving key developmental milestones and cell-cycle checkpoints associated with implantation competence. Despite strong theoretical advantages, accumulated clinical evidence and systematic reviews have not consistently demonstrated improved live birth outcomes when TLI systems are compared with conventional bench-top tri-gas incubation combined with standard morphological assessment. As a result, the clinical superiority of time-lapse incubation for routine embryo selection remains uncertain. Recent advances in artificial intelligence (AI) and computer vision have renewed interest in non-invasive embryo assessment through real-time morphokinetic analysis. Automated image interpretation systems capable of extracting dynamic developmental features may provide objective and reproducible embryo evaluation while reducing operator dependency. Notably, AI-assisted embryo assessment methodologies have been successfully implemented for decades in animal reproduction, particularly within the bovine embryo transfer industry. In this setting, large numbers of embryos have been evaluated using stereomicroscopy combined with digital or mobile video acquisition to assess developmental kinetics and morphological dynamics associated with implantation success. Experience gained from animal reproductive biotechnology has historically served as a foundation for innovation in human ART. Several core technologies currently used in clinical IVF-including embryo cryopreservation, culture media optimization, and micromanipulation techniques-were initially developed and validated in animal models before translation into human reproductive medicine. Building upon these translational precedents, the present study seeks to generate structured imaging datasets suitable for development and validation of machine-learning algorithms capable of predicting embryo developmental competence in human IVF. Establishment of robust, standardized video-based datasets may enable creation of scalable AI-driven decision-support tools applicable across IVF laboratories worldwide, with the potential to improve embryo selection consistency and clinical efficiency while maintaining non-invasive assessment principles. STUDY DESIGN METHODOLOGY This investigation is designed as a prospective, observational, non-interventional study. The study population will include individuals undergoing assisted reproductive treatment between 18 and 45 years of age who generate at least one embryo reaching the blastocyst stage following five, six, or seven days of in-vitro embryo culture. Eligible participants will consist of patients or couples who voluntarily agree to participate and provide written informed consent using an Institutional Review Board (IRB)-approved Informed Consent Form (ICF). Potential participants will be informed of the study objectives, procedures, and voluntary nature of participation at the initiation of fertility treatment. Adequate time will be provided for review of the consent document and for discussion with clinical or research personnel prior to enrollment, consistent with ethical principles outlined in the Declaration of Helsinki and Good Clinical Practice (GCP) guidelines. Participation in this study will not modify clinical management, controlled ovarian stimulation protocols, laboratory culture conditions, or embryo transfer decision-making. Patients retain the unrestricted right to withdraw consent at any time without penalty or impact on medical care, physician-patient relationships, or access to fertility treatment services. Participants enrolled in the study will undergo standard assisted reproductive treatment, including controlled ovarian stimulation (COS), transvaginal ultrasound-guided follicular aspiration for oocyte retrieval, and fertilization performed according to routine embryology laboratory practices. Insemination will be conducted using conventional in-vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI), based on established clinical and laboratory indications. Following fertilization, embryos will be cultured under validated laboratory conditions maintaining strict control of temperature, pH stability, and gas composition consistent with established IVF laboratory standards designed to optimize embryo development. Embryos will be cultured until transfer, cryopreservation, or disposition according to routine clinical practice. At the blastocyst stage (day 5, 6, or 7 of development), and immediately prior to embryo transfer or cryopreservation, a brief real-time microscopic video recording of approximately thirty seconds will be obtained. Video acquisition will occur following routine blastocyst imaging already performed for clinical documentation and embryo transfer reporting purposes. All patients routinely receive standard blastocyst images as part of clinical care. The only study-related deviation from routine laboratory procedures consists of an additional estimated 15-20 seconds of light exposure during video capture. Imaging will be performed using an inverted microscope equipped with a temperature-controlled stage to maintain physiological culture conditions throughout observation. The embryo will remain within a microdroplet of culture medium covered by mineral oil, thereby preserving osmolarity, temperature stability, and pH buffering during the brief observation interval. Short-term microscopic exposure under controlled laboratory conditions has been demonstrated to be compatible with embryo viability when environmental stability is maintained and exposure duration remains limited. Accordingly, the additional imaging time introduced by this observational protocol is not expected to adversely affect embryo development or clinical outcomes. * Study Type: Observational * Study Design: Prospective observational cohort * Interventional Use: Increased light exposure. * Risk Level: Minimal risk * Estimated Duration: (e.g., 12-24 months) * Number of Sites: Multicenter (up to 20 sites) PARTICIPANT POPULATION Participant recruitment for this observational study will be conducted according to the following procedures: * Eligible patients will be identified during routine clinical consultations performed by the Principal Investigator, Sub-Investigators, or authorized study personnel as part of standard fertility care. Recruitment will occur exclusively within the normal clinical workflow of the participating institution. * Written informed consent will be obtained from all participants prior to study enrollment using an Institutional Review Board (IRB)-approved Informed Consent Form (ICF). Consent will be secured before initiation of any study-specific activity or data collection related to the research. * Upon enrollment, study participation will be documented within the participant's medical record, including the date of informed consent execution and assignment of a unique study identification code to ensure subject confidentiality and traceability. * Following enrollment, participants will continue assisted reproductive treatment according to routine clinical practice established by the fertility clinic. Standard procedures may include controlled ovarian stimulation, oocyte retrieval, fertilization via IVF or ICSI, embryo culture, embryo transfer, or elective embryo cryopreservation. In certain clinical circumstances, all embryos may undergo vitrification (e.g., preimplantation genetic testing cycles or provider-directed "freeze-all" strategies), consistent with standard medical decision-making. * For study purposes, authorized research personnel will obtain a thirty-second microscopic video recording of eligible blastocysts. All digital files will be coded using the assigned study identification number to permit subsequent morphokinetic analysis while maintaining participant confidentiality. * Clinical outcomes associated with embryo utilization will be prospectively collected, including results of preimplantation genetic testing when available and outcomes following embryo transfer. Transfer outcomes will be recorded in accordance with predefined study endpoints. * In cases where patients elect to discard cryopreserved embryos, individuals will be informed of the study and provided the opportunity to participate. Written informed consent will be obtained prior to thawing, culture, video acquisition, and subsequent embryo disposition in accordance with institutional policies and applicable regulations. No additional recruitment strategies, advertising, or external enrollment methods will be employed for this study. This investigation will be conducted in accordance with internationally recognized ethical and regulatory standards, including the ethical principles outlined in the Declaration of Helsinki, the Belmont Report, and the International Council for Harmonization (ICH) Guideline for Good Clinical Practice (E6\[R2\]). The study will comply with applicable United States federal regulations governing human subjects research, including the Common Rule (45 CFR Part 46), as well as requirements under the Health Insurance Portability and Accountability Act (HIPAA) for protection of participant privacy and confidentiality. Participating institutions may determine enrollment targets based on clinical capacity and study feasibility considerations. Patients enrolled in this study won't receive any benefit for participating. No result will be given to patients regarding the morphokinetic characteristics of their embryos. In other words, no decision making will be influenced for performing this evaluation. The VIABLE-AIM study does not currently utilize a clinically validated embryo scoring system or diagnostic algorithm. At the present stage of development, the study is observational and exploratory in nature and is designed to evaluate whether quantitative morphokinetic features extracted from short-duration blastocyst videos are associated with known embryo characteristics and reproductive outcomes. The current analytical approach consists primarily of non-machine-learning statistical analysis of normalized pixel activity over time derived from approximately 30-second real-time microscopic embryo videos. Video processing methods include background subtraction and frame-by-frame quantification of temporal pixel-intensity changes. From these processed video signals, multiple quantitative morphokinetic variables may be extracted, including but not limited to: Mean activity Standard deviation of activity Peak count Valley count Peak amplitude Temporal fluctuation patterns Burst activity metrics Coordinated rhythmicity measurements Additional descriptive statistical features derived from normalized activity curves These variables are analyzed for potential associations with known embryo outcomes, including implantation outcome, pregnancy outcome, and chromosomal status when available. At the current stage, these measurements are exploratory research variables only and are not used for clinical decision-making, embryo selection, or patient management. The study does not currently generate a validated clinical "Viable AIM Score." Any references to "AIM score" within the protocol represent a conceptual or investigational framework rather than an established diagnostic classifier. The present study is intended to generate foundational datasets necessary for future model development and hypothesis generation. Model development may include training supervised machine-learning or deep-learning models using outcome-linked datasets generated in this study. Potential model architectures under evaluation for future research include: Convolutional Neural Networks (CNN) Long Short-Term Memory (LSTM) networks Transformer-based architectures Other spatiotemporal deep-learning approaches These future models may ultimately generate probabilistic confidence outputs representing the likelihood of predefined outcomes (for example implantation potential or euploid status), typically expressed on a normalized scale between 0 and 1. However, such models are currently investigational, have not been clinically validated in humans, and are not being used prospectively in patient care within the current study. The purpose of the present study is therefore to establish feasibility, characterize morphokinetic signal patterns, and generate structured datasets suitable for future validation studies and regulatory evaluation of AI-assisted embryo assessment technologies.