The inclusion criteria for the participants are presented in Table 1. In group 1, the eligibility criteria were designed to minimize factors adversely affecting spermatogenic function for quality control of DFI measurement values. Specifically, group 1 included only men who were sexually active young adults (18-24 years old), had a high frequency of ejaculation (average of ≥10 per month), had a standard body type (BMI 18.5 to <25 kg/m²), and had never smoked. Group 2 had broader inclusion criteria to explore DFI risk factors in men from diverse backgrounds. The target population included men of general reproductive age (18-41 years) with minimal ejaculatory function (average of ≥1 time per month). However, to ensure that the blood and urine test results were accurate, compliance with visits during the designated time (8:00 to 11:00 A.M.), fasting the previous night, and a two-visit schedule was mandatory for this group.

The exclusion criteria were identical for both groups. We excluded individuals with a history of congenital anomalies or diseases of the male reproductive organs, varicocele, malignant tumors, endocrine diseases, or a history of hormone medication use. Additionally, individuals who might experience unjust disadvantages from participating, those who had difficulty providing consent because of cognitive impairment, or those judged by the study physician to be in poor physical condition on the day of the survey (eg., acute illness symptoms) were considered ineligible. If a participant in group 1 also met the requirements of group 2 and wished to participate, co-enrollment in both groups was permitted. Such cases were identifiable through study registration records. Because groups 1 and 2 were established for different purposes rather than as formal independent comparison arms, co-enrolled records were retained as group-specific records but were not treated as statistically independent in any cross-group descriptive comparison or pooled participant-level analysis.

To reduce clinic-based sampling bias (ie, overrepresentation of men seeking fertility evaluation), we recruited men residing in Japan who were registered in a volunteer panel managed by InCROM Co, Ltd (Nishiwaki, Japan) rather than through infertility clinics. After eligibility confirmation, participants were referred to the Osaka Pharmacology Clinical Research Hospital, a clinical research implementation facility, where written informed consent was obtained. This study concerns semen-based measures of biological male reproductive function; gender identity and sexual orientation were not assessed, and sex/gender was not analyzed as a separate covariate in this initial phase. Given that study participation required morning fasting visits and (for group 2) two visits, we anticipated potential overrepresentation of individuals with greater scheduling flexibility. To characterize this quantitatively, age and recruitment-related sociodemographic variables (eg, employment status, shift work, education, and income) were systematically collected. Their distributions, including “prefer not to answer” responses where applicable, will be reported.

Considering biological sample processing, assay capacity, and the budget in this initial phase, the target group-specific enrollment was set at 25 for group 1 and 100 for group 2. Because co-enrollment was permitted, these targets refer to group-specific enrollment records; therefore, the total number of unique individuals across both groups may be smaller than the sum of the two group-specific counts. The two groups address different objectives. Group 1 is intended for quality control and descriptive characterization of the DFI measurement system in a relatively homogeneous low-risk population, rather than formal hypothesis testing. Group 2 is intended for feasibility-oriented initial association analyses between sperm image/video data and the DFI and exploratory analyses of candidate factors. Therefore, the sample size was determined pragmatically rather than via a formal power calculation for a specific effect size. We acknowledge that this initial phase may be underpowered to detect small associations or to support multivariable models with many predictors. Accordingly, effect estimates, CIs, and consistency of direction will be emphasized, and the findings will be used to inform variable prioritization and formal sample size planning for subsequent phases.

Data on baseline characteristics were collected through these structured self-administered questionnaires and through physician-conducted interviews for clinically complex eligibility items. Demographic and socioeconomic factors included age, height, weight, educational attainment, marital status, and household income [46]. Regarding medical background, current and past medical histories were obtained. Regarding current medications, we verified the use of treatments for androgenetic alopecia, drugs for erectile dysfunction, and zinc supplementation. Furthermore, as part of the reproductive background, the partner’s pregnancy history, history of infertility treatment [47,48], the participants’ desire for children [49], and experience using home sperm-testing kits [50] were investigated.

Regarding environmental and occupational exposures, employment status, including shift work, was assessed [51]. Work in high-temperature environments and the frequency of laptop use on the lap were also assessed to evaluate potential thermal effects on spermatogenesis [52].

Lifestyle factors, psychological status, and sexual function were evaluated multidimensionally using structured self-administered questionnaires, incorporating validated scales and specific self-developed questions. Where available, validated Japanese versions or established Japanese translations of standard instruments were used. Regarding lifestyle habits, data on smoking habits (Brinkman index; daily number of cigarettes × years) [53], alcohol consumption (Alcohol Use Disorders Identification Test–Consumption [AUDIT-C]) [54], and sauna/bathing habits (self-developed questionnaire) were collected. Nutritional intake was assessed using the Brief-type Self-Administered Diet History Questionnaire (BDHQ) [55]. Sleep quality was evaluated using the Japanese version of the Pittsburgh Sleep Quality Index (PSQI-J) [56], chronotype using the reduced Morningness-Eveningness Questionnaire (rMEQ) [57], and physical activity levels using the International Physical Activity Questionnaire (IPAQ) [58].

Psychological status was assessed using the Perceived Stress Scale (PSS) [59] and the Kessler Psychological Distress Scale (K6) [60]. Sexual function and urological symptoms were comprehensively evaluated using the Sexual Health Inventory for Men (SHIM) [61,62], a modified SHIM replacing “sexual intercourse” with “masturbation” in questions 3-5 [63]; the Erection Hardness Score (EHS) [64,65]; the Male Sexual Health Questionnaire–Ejaculatory Dysfunction–Short Form (MSHQ-EjD-SF) [66,67]; and a modified version of the first question of the MSHQ-EjD-SF, where “sexual intercourse” was replaced with “masturbation” (self-developed modification). The self-estimated intravaginal ejaculation latency time [68] and the self-estimated ejaculation latency time during masturbation (self-developed question) were additionally assessed. The frequencies of sexual intercourse, masturbation, nocturnal erection, and feelings of sexual desire were evaluated using point scales similar to those used in the 2023 National Survey in Japan [69].

Furthermore, the Aging Males’ Symptoms Scale (AMS) score [70-72] was used as an indicator of physical and psychological changes associated with androgen decline. The National Institutes of Health Chronic Prostatitis Symptom Index (NIH-CPSI) [73] was used to evaluate chronic prostatitis symptoms. These measures were included to support exploration of potential infertility risk factors and hypothesis generation.

Published studies support the reliability/validity or relative validity of the BDHQ [55], PSQI-J [56], rMEQ [57], IPAQ [58], AUDIT-C [54], PSS [59], K6 [60], AMS [71], and NIH-CPSI [73] in Japanese populations. A Japanese version of the SHIM [62] and MSHQ-EjD-SF [67] were used. The EHS is a single-item measure, for which internal consistency is not applicable. Modified or self-developed items were treated as exploratory measures.

For group 2 participants, we conducted detailed physical function evaluations and physical measurements based on standardized procedures. The weight, BMI, body fat percentage, muscle mass, and segmental muscle mass of the limbs and trunk were measured using a multifrequency body composition analyzer (MC-780A, TANITA, Tokyo, Japan). The skeletal muscle mass index [74], fat mass index [75,76], phase angle [77,78], and extracellular water–to–total body water ratio [79] were also measured. These indicators were collected to evaluate the detailed body composition, including regional characteristics and cellular status, in addition to indicators such as BMI and body fat percentage commonly used in previous studies [80,81]. Measurements were performed with participants barefoot and wearing light clothing, with corrections made for the weight of the clothing.

Subsequently, muscle strength was evaluated as a physical indicator of androgen status and whole-body muscle strength. In addition to grip strength [82,83], a widely used representative indicator of whole-body muscle strength and physical function, back muscle strength [84] was also measured for a more comprehensive evaluation of muscle strength status, including that of the trunk. Digital grip strength (TAKEI 5401, SANKA Co, Ltd, Health Equipment Division, Tokyo, Japan) and back muscle strength meters (TAKEI 54021, SANKA Co, Ltd) were used. Grip strength was measured twice for each hand alternately in the standing position, and the maximum value was adopted [85]. Back muscle strength was measured twice, with the upper body inclined forward at 30° and the knees extended (but not locked), and the maximum value was adopted [86].

Anatomical circumferences were measured based on standardized procedures using a nonstretchable measuring tape (SECA 201, Hamburg, Germany). Specifically, the waist circumference was measured at the midpoint between the lower rib margin and the iliac crest [87]. The maximum circumferences of the hip [88], thigh [89], and calf [90] were also measured. These circumferences were collected exploratorily for a simple evaluation of the regional body shape and fat/muscle distribution, which could not be captured using whole-body composition indicators. For quality control, each site was measured twice, and if the difference was ≥1 cm, a third measurement was performed. Sexual function may be associated with complex physical factors, such as systemic metabolic status, muscle mass/strength, fluid distribution, and physical activity. Thus, data on body composition, muscle strength, and morphological indicators were comprehensively collected to evaluate the whole-body physical status multidimensionally.

Following an ejaculatory abstinence period of 48 to <144 h (verified upon arrival) [91], semen was collected via masturbation in a designated room at the clinical research facility and analyzed in an onsite laboratory. The samples were liquefied in a 37°C incubator or at room temperature, with a maximum liquefaction time of 60 minutes. Incomplete liquefaction was defined as a thread length of ≥2 cm during pipetting, in which case mechanical agitation using an 18-gauge needle was performed. Semen volume was calculated by weighing the collection container, assuming a semen density of 1.0 g/mL.

Semen analysis was performed using an automated semen quality analysis system (LensHooke X3 PRO, Bonraybio, Taichung, Taiwan) [92]. Sperm concentration and total motility were measured in accordance with the 6th Edition of the World Health Organization (WHO) Manual for Human Semen Analysis [93,94]. The following specific kinematic parameters were also analyzed: categorization of sperm motility (fast progressively motile, slow progressively motile, nonprogressively motile, and immotile), average path velocity, straight line velocity, curvilinear velocity, amplitude of lateral head displacement, beat cross frequency, linearity, straightness, and wobble [95]. In addition, the percentage of normal morphology, pH, and round cell concentration were measured using an automated image recognition system. To evaluate sperm oxidative stress, the static oxidation-reduction potential was measured using a sperm oxidative stress measurement system (MiOXSYS, Caerus Biotech, Vilnius, Lithuania) [96]. Microscopic videos generated by the system were stored in native output format, and file metadata (eg, frame rate and resolution) were retained for subsequent image-based analyses.

The participants additionally completed a Subjective Assessment at Collection questionnaire immediately after semen collection. The questionnaire recorded the abstinence period, presence of spillage, time taken for collection on the study day, subjective intensity of orgasm on the study day (4-point Likert scale) [97], subjective semen volume compared to usual (modified question 3 of the MSHQ-SF), subjective time taken for collection compared to usual (self-developed question), and stress level during collection (10-point Likert scale, self-developed question). This was intended to aid in assessing the influence of psychological and situational factors on sample quality.

Residual samples after analysis were preserved as smears for Diff-Quik staining. Residual semen for downstream exploratory analyses was aliquoted according to the volume available after onsite testing and stored at −80°C before shipment to collaborating institutions or external laboratories. The smears are being morphologically evaluated based on Kruger’s strict criteria [98]. Depending on residual sample availability, frozen semen aliquots are being used for exploratory analyses of the sperm DFI; high DNA stainability (HDS) using SCSA [99-101]; trace and major elements and chemical speciation elements in semen and seminal plasma; and additional biomarkers, such as inhibin B and insulin-like 3 (INSL3).

Group 2 participants provided early-morning urine and fasting blood samples. Fasting was defined as no breakfast on the study day, and the participants were instructed to finish dinner by approximately 9:00 P.M. on the day before the visit. Water intake was permitted. To account for diurnal fluctuations in analytes, such as testosterone [102] and zinc [103] levels, blood samples were collected between 8:00 and 11:00 A.M. To avoid potential alterations in parameters such as postejaculation prolactin, blood sampling was performed before semen collection [104]. A portion of the blood was processed into plasma or serum and stored frozen at −80°C; these samples were shipped to collaborating institutions or external laboratories for exploratory analysis.

Urine assays include measurements of exposure to environmental chemicals (phthalate esters, parabens, and pesticides), soy isoflavones, trace and major elements, chemical speciation of elements, creatinine, specific gravity, and urea nitrogen. Blood assays include measurements of serum and plasma gonadotropins and sex steroid profiles (follicle-stimulating hormone, luteinizing hormone, testosterone, free testosterone, dehydroepiandrosterone sulfate, sex hormone–binding globulin, dihydrotestosterone, estradiol, prolactin, anti-Müllerian hormone, inhibin B, and INSL3), metabolic markers (serum albumin, fasting blood glucose, and insulin), and selected exposure substances (per- and polyfluoroalkyl substances, trace and major elements, and chemical speciation of elements) in whole blood, serum, and plasma.

The primary outcome measure is the continuous DFI. The DFI refers to the proportion of sperm with fragmented or acid-denaturable DNA. In this study, the DFI is measured using SCSA after acid treatment and acridine orange staining with flow-cytometric analysis [12-14]. The primary analysis will assess whether predefined microscopic sperm image-/video-derived summary variables are associated with a continuous DFI, as a feasibility-oriented proof-of-concept evaluation of whether sperm microscopy data contain information relevant to sperm DNA integrity. Secondary and exploratory outcome measures include a categorical DFI based on literature-informed or study-specific exploratory thresholds, HDS, and general semen analysis parameters. Clinical, environmental, physical, and lifestyle variables will be examined as candidate correlates or confounders in secondary exploratory analyses [105].

Continuous variables will be summarized as means (SDs) and/or medians (IQRs), depending on the distribution characteristics, and categorical variables will be summarized as frequencies (percentages) [106]. The DFI will be treated as a continuous variable. Categorization using literature-informed thresholds (eg, 15% and 27%) [107-110] will also be used in secondary exploratory analyses. In addition, if data characteristics permit, a study-specific candidate threshold will be explored using receiver operating characteristic (ROC) analysis against a prespecified binary semen quality classification based on the lower fifth percentile values for basic semen parameters described in the 6th Edition of the WHO Manual for Human Semen Analysis [94,111,112]. Any such data-driven threshold will be considered hypothesis generating only [113].

No formal between-group inferential comparison between groups 1 and 2 is a primary objective of this study. Any comparison of baseline characteristics between the two groups will therefore be descriptive/exploratory only. If participant overlap between groups is present, co-enrolled individuals will be counted only once in such cross-group summaries or in any pooled participant-level analysis requiring independence. Under the primary handling rule, the group 1 record will be retained and the corresponding group 2 record excluded, because group 1 had stricter eligibility criteria and a quality control purpose. Alternative handling rules, such as retaining the group 2 record or excluding co-enrolled participants entirely, may be examined as sensitivity analyses. This issue does not affect the main group-specific analyses, in which group 1 is used for DFI quality control and group 2 for exploratory association analyses.

All statistical tests will be two-sided and interpreted using a nominal significance level of P<.05. However, the secondary analyses are exploratory and involve multiple related comparisons and sensitivity analyses. Thus, emphasis will be placed on effect sizes, 95% CIs, and consistency of direction rather than on statistical significance alone, and no formal multiplicity adjustment is planned. As exposure indices (eg, environmental chemicals) may generally exhibit skewness, transformations (eg, logarithmic transformations) will be considered according to the distribution [114]. For urinary indices, a correction using urinary creatinine or specific gravity will be applied [115,116].

Data from group 1 will be used for quality control of the DFI measurement system. Specifically, using SCSA for DFI and HDS measurements, we will evaluate intraday, interday, and interfacility variability using quality control semen samples. Given that SCSA-derived DFI values can be influenced by the assay protocol, instrument settings, and gating strategy, this quality control step is required before the main association analyses [14,117,118]. After evaluating measurement validity, the flow cytometry gating settings will be fixed to ensure measurement reliability. Using data from group 1, we will confirm the distribution in young, healthy men; exclude outliers; and describe the reference value range.

For the primary analysis, predefined microscopic sperm image-/video-derived summary variables will be assessed for association with the continuous DFI (primary outcome measure) as a preliminary proof-of-concept evaluation of whether image/video data contain information relevant to sperm DNA integrity. Associations with the continuous DFI will initially be assessed using correlation analyses and simple linear regression. When the DFI is categorized using prespecified literature-informed thresholds or a study-specific exploratory candidate threshold, simple logistic regression will be used as a secondary exploratory analysis. Any predictive modeling (including AI approaches) will be considered exploratory and will be specified in subsequent phases after database expansion. Exploratory image-DFI association analyses will be reported separately by group. If any additional pooled participant-level analysis across groups 1 and 2 is performed, the nonoverlapping dataset described earlier will be used.

For group 2 data, associations among the DFI, microscopic sperm image/video data, and clinical variables will be analyzed to identify candidate predictors and confounding factors for image-based analyses. Core analyses will prioritize standard semen parameters, questionnaire data, and physical measurements. The DFI will be analyzed primarily as a continuous outcome, with secondary analyses using thresholds [107-110]. Building on the reference values established in the quality control analysis, we may explore a study-specific DFI threshold against the WHO criteria [93] using ROC analysis. The area under the ROC curve and Youden index–derived cut points will be reported as hypothesis-generating results.

Candidate variables will first be examined using correlation analyses, group comparisons, or simple linear/logistic regression, as appropriate to variable type. Age and recruitment-related participant characteristics (eg, employment status, shift work, education, and income) will also be examined as candidate covariates or stratification factors to evaluate potential selection-related bias. The data of participants selecting “prefer not to answer” for education or income will be retained in descriptive analyses, and sensitivity analyses will compare alternative handling of these responses (eg, separate-category vs exclusion approaches). Given that a population-based sampling frame is not available, these analyses are intended to assess robustness rather than fully remove selection bias. Multivariable linear or logistic regression models will be considered only when data completeness and the number of informative observations is adequate. To reduce overfitting, the number of predictors simultaneously included will be restricted, with priority given to variables selected based on prior knowledge and univariable results. Multicollinearity will be assessed using the variance inflation factor, with values >3 considered to indicate potential concern. In such cases, we will preferentially retain representative variables or use dimension reduction, clustering, or mixed-exposure approaches as exploratory analyses.

Participants with missing data for primary outcome measure analysis (DFI or images) will be excluded from the primary analysis, and the reasons for missingness will be described. Considering that the status of questionnaires or samples, such as self-reports of spilling all or part of the sample during collection [119] or recent infectious diseases (eg, a cold, influenza, or COVID-19) with or without a high fever (≥38°C) within the past 3 months [120,121], may affect the results, sensitivity analyses excluding such participants will be conducted, as necessary. If azoospermic samples are observed, sensitivity analyses will be performed both including and excluding them because their underlying etiologies cannot be fully characterized in this non-clinic-based setting [122].