Introduction
The convergence of healthcare and digital technologies has accelerated the emergence of Femtech, a rapidly expanding field that encompasses technology-enabled solutions for women’s health [1,2]. According to industry estimates, the global Femtech market was projected to exceed US $50 billion by 2025 [3], reflecting the rapid expansion of digital health technologies—including mobile applications, wearable biosensors, and artificial intelligence (AI)-based clinical decision systems—in healthcare delivery [2].
Within this landscape, reproductive health—and infertility in particular—has become a central area of innovation [4]. Infertility is a complex global health condition associated with substantial psychological, physical, and socioeconomic burdens [5].
Patients frequently report uncertainty, limited access to reliable information, and a lack of continuity in care throughout the treatment process [6]. In response, digital health technologies have been developed to improve accessibility, enhance patient engagement, and support clinical decision-making in reproductive medicine [7]. Despite rapid technological advances, robust evidence demonstrating improvements in clinically meaningful reproductive outcomes remains limited and inconsistent. This review critically examines the available evidence across three domains: consumer-facing fertility tools, digital health integration in infertility care, and AI applications in assisted reproductive technology (ART).
Methods
This structured narrative review was based on a comprehensive literature search of PubMed, Embase, and the Cochrane Library for studies published between January 2010 and July 2025. Web-based sources were accessed and verified through April 2026. Search terms included combinations of ‘Femtech,’ ‘infertility,’ ‘digital health,’ ‘telemedicine,’ ‘mobile health,’ and ‘artificial intelligence in reproductive medicine.’
Studies were included if they addressed clinical applications of digital technologies in reproductive health, including fertility tracking, telemedicine, mobile health (mHealth), or AI-based ART optimization. Randomized controlled trials and observational studies were considered, and review articles were included selectively to provide background context. Studies were excluded if they focused solely on contraception without relevance to infertility, lacked clinical outcomes, or were not available in English.
Given the heterogeneity of study designs and outcomes, a qualitative synthesis approach was used, with emphasis placed on study design, sample size, and potential sources of bias. Because this work was conducted as a structured narrative review rather than a systematic review, no formal risk-of-bias assessment tool was applied; instead, study design limitations, sample size, and potential sources of bias were considered narratively during synthesis.
Digital fertility tracking and at-home diagnostics: empowering the preconception journey
For many individuals and couples, the path to conception begins with consumer-facing technologies, such as smartphone applications, that identify the fertile window and track fertility indicators. Use of these applications has been associated with improved fecundability and may represent an early digital resource before clinical care is sought [8].
Fertility awareness–based method applications
An ecosystem of nearly 100 fertility and menstrual tracking applications—used for both contraception and conception planning—has emerged as an early form of digital reproductive self-care. Fertility awareness–based method applications educate users about biomarkers such as cervical fluid and basal body temperature, providing an initial resource for cycle understanding, increasing users’ sense of control, and supporting proactive reproductive management [9].
However, the strength of evidence varies substantially across fertility awareness tools. Calendar-based applications have consistently demonstrated limited accuracy because they rely on assumptions about cycle regularity that do not reflect physiological variability in ovulation timing [10]. In contrast, biomarker-integrated applications may offer greater precision, but much of the supporting evidence comes from observational cohorts of self-selected users, limiting causal inference and raising the possibility of selection bias [11].
Evidence supporting ovulation prediction should not be conflated with evidence supporting app-based interventions themselves. The 2023 Cochrane review found moderate-certainty evidence that timed intercourse guided by urinary luteinizing hormone (LH) probably improves live birth, but it did not establish comparable evidence for fertility applications as a class [12]. Therefore, current data suggest that these tools may improve fertility awareness and timing behavior, whereas their direct effect on clinically meaningful outcomes remains uncertain.
Advanced self-monitoring: wearables and quantitative hormone assays
Beyond manual entry, wearable biosensors, such as the Oura ring and Ava bracelet, enable passive, continuous measurement of physiological signals—including nocturnal skin temperature, heart rate, heart rate variability, and respiratory rate—that show cyclical changes correlated with the fertile window and the preovulatory LH surge [13,14]. These devices represent a promising step toward passive, continuous fertility monitoring, but the current evidence base remains preliminary. Most studies have focused on physiological correlations with menstrual or ovulatory markers rather than patient-centered reproductive outcomes, and many are limited by small samples and observational designs [13,14]. As a result, wearables appear technically feasible, but their incremental clinical value over simpler, established methods remains to be defined.
At-home quantitative hormone monitors, such as Mira and Inito, measure urinary estrone-3-glucuronide, LH, and pregnanediol glucuronide. These systems demonstrate strong analytical performance and improved detection of luteal-phase dynamics compared with qualitative ovulation predictor kits [15,16]. Most supporting evidence consists of validation studies demonstrating agreement with established laboratory or commercial monitoring systems, such as Clearblue; one validation study reported high precision in identifying the LH surge (r=0.94) [15,16].
However, analytical accuracy does not necessarily translate into improved reproductive outcomes, and data linking these devices to higher pregnancy or live birth rates remain limited. Integrating home-based quantitative hormone monitoring with telemedicine in medically assisted reproduction (MAR) cycles has been proposed as a way to reduce reliance on frequent in-clinic phlebotomy while maintaining individualized surveillance, although its clinical effectiveness in improving ART outcomes has not yet been established [17]. Notably, the current evidence base is derived largely from validation and observational studies, limiting causal inference and clinical generalizability.
Home-based semen analysis
Male factors contribute to approximately half of infertility cases, yet evaluation is often delayed because of stigma, logistical barriers, and discomfort with in-clinic sampling [18,19]. At-home semen tests vary in sophistication. Simple immunodiagnostic devices, such as SpermCheck Fertility (Princeton BioMeditech Corporation), qualitatively assess sperm concentration through sperm protein-10 (SP-10) antigen detection, using thresholds such as 20 million/mL [20]. Although these tests offer convenience and ease of use, their clinical utility is inherently limited by their qualitative nature and by their inability to provide a comprehensive assessment of semen parameters, including motility and morphology [20,21]. The SwimCount test isolates and quantifies progressively motile sperm; its performance correlates well with World Health Organization–standard semen analysis and may better predict fertility potential than total sperm count alone. Validation studies have shown good diagnostic reliability (area under the curve [AUC], 0.85; sensitivity: 87.5% for total progressively motile sperm count <5 million/mL) [22]. However, these findings are derived from diagnostic validation studies, and their effect on clinical decision-making or reproductive outcomes remains unclear.
Among the most advanced at-home options are smartphone-based systems. The YO device (Medical Electronic Systems) uses a phone-mounted optical attachment to capture microscopic videos of unprocessed semen samples; in validation testing, it achieved 97.8% accuracy for classifying motile sperm concentration compared with a laboratory analyzer [23]. The O’VIEW-M PRO® system (Intin Inc.), formally evaluated in a multicenter study, showed an overall accuracy of 84.6% for concentration and motility relative to computer-assisted semen analysis (CASA) [24]. These studies support the analytical validity of smartphone-based systems; however, they primarily assess agreement with laboratory parameters rather than effectiveness in improving reproductive outcomes. Most currently available home-based tests assess only selected semen parameters and do not provide a comprehensive evaluation of male fertility, particularly in domains such as sperm morphology, vitality, or DNA fragmentation. Therefore, although these technologies may lower barriers to initial assessment and facilitate early screening, they should be regarded as adjunctive tools rather than replacements for standard laboratory semen analysis; further studies are needed to determine their role in improving fertility outcomes [21].
Integration of digital health into clinical infertility management
Beyond preconception, digital technologies are increasingly being embedded in infertility care pathways. Home-based urinary hormone monitoring combined with telemedicine in MAR cycles is positioned as an adjunct—not merely as a consumer tool—to streamline monitoring, reduce clinic burden, and promote patient-centered care [17].
Telemedicine in reproductive care
The coronavirus disease 2019 pandemic catalyzed telehealth adoption in reproductive endocrinology and infertility. In one survey, all responding providers implemented telehealth during the pandemic, and most intended to continue using it after the pandemic [25]. Evidence now supports high patient satisfaction with virtual infertility consultations, with some series reporting satisfaction of 91% (95% confidence interval [CI], 80% to 96%) and citing convenience, reduced travel time and costs, and improved access for rural or remote patients [26].
Patient preferences support a hybrid model: in a national survey, 60.5% of respondents preferred in-person visits for initial consultations, where establishing a therapeutic relationship may be especially valued, whereas telehealth was more acceptable for follow-up visits [27]. A systematic review and meta-analysis found no significant difference in pregnancy rates between telemedicine-assisted monitoring and in-clinic care (odds ratio, 1.02; 95% CI, 0.83 to 1.26), supporting the safety and comparable clinical outcomes of decentralized ART management, although variation in study design and patient selection should be considered [26]. This model may separate expert consultation and routine monitoring from the physical ART center and thereby broaden access to specialized care.
mHealth for patient support during ART
Infertility treatment is medically complex and emotionally demanding and is often accompanied by significant psychological distress [28]. In response, mHealth applications have evolved from static information tools into more integrated platforms designed to support patient engagement and care delivery throughout ART. The Infotility application was designed to deliver gender-informed medical, lifestyle, psychosocial, and legal content; usage analytics indicated that lifestyle content was accessed most frequently by both men and women [29]. The WiStim application functions as a secure, bidirectional communication platform for ovarian stimulation, enabling clinicians to deliver daily instructions and educational content while allowing patients to manage treatment schedules. In a prospective observational study, users reported high satisfaction, ease of use, and perceived usefulness, while clinical staff reported time savings of approximately 2 hours per day and a potential reduction in treatment-related errors [30]. Similarly, MediEmo, which was developed collaboratively with patients and clinicians to address both medical and emotional needs during ART, demonstrated high feasibility and uptake: 79.8% of eligible patients used the medication timeline function, and usability and acceptability ratings were favorable [31].
Data generated by integrated mHealth platforms—including medication adherence, patient-reported symptoms, and emotional well-being—can be incorporated into electronic health records, giving clinicians real-time insights between visits and enabling a more proactive, personalized model of care [32]. Collectively, evidence for these mHealth applications comes primarily from observational and implementation-based studies focused on user engagement, feasibility, and satisfaction. Although these findings support the acceptability of mHealth platforms and their potential to enhance patient support, they do not establish causal effects on adherence, treatment success, or reproductive outcomes. Thus, these platforms appear feasible and well accepted, but their clinical utility in routine infertility care remains uncertain.
AI: the new frontier in ART
AI and machine learning are increasingly being used to enhance sperm and embryo assessment, predict outcomes, and support individualized treatment planning in infertility care [33].
AI-driven gamete and embryo selection
Embryo selection, a cornerstone of in vitro fertilization (IVF), has traditionally relied on morphological grading, which is widely used but subjective and operator-dependent. Interpretable AI may offer a more objective and reproducible alternative. In a prospective cohort study, AI-assisted embryo selection was associated with a higher implantation rate than conventional selection (80.9% vs. 68.2%, p=0.022), without adverse neonatal or pregnancy outcomes [34].
For sperm selection, AI-enhanced CASA achieves high accuracy in assessing morphology, including a support vector machine model AUC of 88.6%, and motility, with approximately 89.9% accuracy; emerging approaches also aim to predict DNA fragmentation [35]. AI has also been applied to oocyte assessment using microscopic imaging, with analyses of features such as cytoplasmic transparency, zona pellucida thickness, and perivitelline space that correlate with blastocyst usability [35].
Deep learning models, particularly convolutional neural networks, have further advanced embryo evaluation by analyzing static and time-lapse images to predict implantation potential. Some studies have reported that AI-based models outperform experienced embryologists in identifying embryos with higher implantation potential [36]. Predictive performance may be enhanced through integration with time-lapse imaging, which enables analysis of morphokinetic parameters that are not readily detectable by human observation [37,38].
Taken together, the available evidence is derived largely from prospective cohort, validation, and retrospective studies focused on algorithm performance and does not yet establish causal effects or consistent improvements in clinically meaningful outcomes, such as fertilization rates, embryo quality, or live birth rates. The limited number of prospective studies and the reliance on single-center or dataset-specific models raise further concerns about external validity, reproducibility, generalizability, and potential dataset bias. Accordingly, although AI shows potential to improve the consistency and objectivity of gamete and embryo assessment, it should currently be regarded as a decision-support tool rather than a replacement for expert clinical judgment, pending further multicenter prospective validation and standardized evaluation frameworks.
Predictive modeling for IVF outcomes
Beyond the laboratory, machine learning enables individualized prognostic estimates that may surpass traditional age-based estimates. Models such as XGBoost and LightGBM that integrate age, body mass index, ovarian reserve markers—including anti-Müllerian hormone and antral follicle count—diagnosis, and treatment history have reported strong predictive performance for clinical pregnancy and live birth (AUC up to 0.999 and 0.913, respectively) [39]. By incorporating multiple patient-specific variables, these models may support individualized counseling and decision-making [40].
Machine learning center-specific (MLCS) models trained on clinic-level data may outperform generalized national registry–based models, such as those from the Society for Assisted Reproductive Technology (SART). MLCS models have shown substantially higher precision-recall AUC and F1 scores (p<0.05), reducing both false positives and false negatives [41]. Notably, one MLCS model identified 11% of patients with an estimated live birth probability ≥75%—a category in which the SART model identified no patients—with an observed live birth rate of 81% in that group [41].
The available evidence, which is derived largely from retrospective, single-center, or dataset-specific studies, has not demonstrated consistent improvements in clinically meaningful outcomes across diverse populations. Moreover, despite high reported predictive performance, these models remain susceptible to overfitting, poor calibration, and limited external validity. (Table 1)
Discussion
The integration of Femtech and digital health into reproductive medicine offers meaningful opportunities to improve access, convenience, and personalization of care; however, the strength of the supporting evidence varies substantially across domains [2,4]. Telemedicine in infertility care is supported by relatively mature evidence, including systematic reviews demonstrating high patient satisfaction and clinical outcomes comparable to conventional in-person management in selected settings, supporting its role within hybrid care models [26,27]. In contrast, most consumer-facing technologies, such as fertility tracking applications, wearable biosensors, and home-based diagnostic tools, primarily enhance patient engagement, monitoring convenience, and workflow efficiency, whereas evidence for improvements in clinically meaningful outcomes, including pregnancy and live birth rates, remains limited or inconsistent [12]. Similarly, AI in ART shows promise for embryo selection and outcome prediction, but much of the current literature is based on observational or validation studies, with limited external validation and uncertain generalizability; therefore, these tools are best regarded as decision-support systems rather than stand-alone clinical solutions [33,34].
Across the literature, several recurring limitations warrant cautious interpretation. Study designs are heterogeneous, outcome measures are not standardized, and many studies rely on surrogate endpoints rather than patient-centered outcomes. In addition, selection bias is common in studies of digital tools that involve self-selected, technology-engaged users, and the rapid evolution of AI systems raises concerns about reproducibility and transportability across clinical settings. These factors underscore a consistent pattern in which analytical performance and technical feasibility outpace robust evidence of clinical effectiveness.
At the same time, the expansion of Femtech raises important ethical and regulatory challenges. The collection and use of sensitive reproductive health data require rigorous standards for privacy, security, and informed consent. In addition, the regulatory distinction between general wellness tools and medical devices remains unclear for many applications, particularly AI-driven systems; clearer validation frameworks, international regulatory standards, and appropriate oversight mechanisms are therefore needed [42].
Beyond these regulatory considerations, equity remains a critical concern because disparities in digital access and literacy may limit the equitable benefits of these technologies. These disparities underscore the need to address gaps in user understanding and engagement. In this context, structured user education and clinician-guided counseling are essential to support appropriate interpretation of results and mitigate the risk of misinterpretation associated with consumer-facing technologies.
Looking forward, well-designed prospective and multicenter studies with larger populations, together with standardized evaluation frameworks and clinical guidelines, will be essential to validate these technologies and support consistent, safe clinical integration. Prospective randomized controlled trials, standardized clinical guidelines, international regulatory frameworks, and comprehensive user education are needed to determine whether these technologies improve clinically meaningful reproductive outcomes beyond feasibility or analytical performance.
For AI applications, external validation, transparent model development, and standardized performance metrics will be essential to ensure reliability and safe clinical integration. In parallel, integrated care ecosystems that connect home-based monitoring, telemedicine, and clinical decision-making may enable more personalized and efficient reproductive care, provided that innovation remains aligned with evidence, regulation, and equity considerations.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
