Comparison of Clinical Characteristics and Pregnancy Outcome Prediction Models Between AIH and AID Patients Authors: Huijuan Wei, Pengyu Huang, Gangxin Chen Published: 2026 ABSTRACT AbstractObjectiveArtificial insemination by husband (AIH) and artificial insemination by donor (AID) are core assisted reproductive techniques for infertile couples; however, few head-to-head comparative prediction models exist for their pregnancy outcomes. This study aimed to compare clinical population characteristics, construct independent pregnancy outcome prediction nomogram models for AIH and AID patients, identify differential independent predictive factors, and evaluate model discrimination performance.Material and MethodsThis single-center retrospective observational cohort study enrolled 11651 AIH cycles and 3064 AID cycles from 2015 to 2025. All baseline demographic, clinical, treatment and laboratory parameters were collected and compared between groups. Binary univariate logistic regression was used to screen candidate predictors; variables with P<0.05 were included in multivariate forward stepwise logistic regression to identify independent predictive factors, visualized via forest plots. Visual nomogram prediction models were built using R software. Receiver operating characteristic (ROC) curve and area under the curve (AUC) were adopted to preliminarily assess model predictive ability. Employs the within-cohort bootstrap method for validation.ResultsBaseline profiles and pregnancy rates differed significantly between groups (P<0.05). Seven independent predictors were identified for AIH and three for AID. For AIH, advancing female age (OR=0.97, 95% CI:0.95~0.99), prolonged infertility duration (OR=0.96, 95% CI:0.93~0.98) and more treatment cycles (OR=0.92, 95% CI:0.86~0.98) reduced pregnancy probability, while thicker endometrium (OR=1.03, 95% CI:1.00~1.07), optimized ovulation induction (OR=1.27, 95% CI:1.13~1.42), unexplained infertility (OR=1.37, 95% CI:1.23~1.53) and double insemination (OR=1.18, 95% CI:1.06~1.32) elevated pregnancy odds (all P<0.05). For AID, longer infertility duration (OR=0.95, 95% CI:0.91~0.99) and repeated cycles (OR=0.88, 95% CI:0.82~0.95) impaired pregnancy, yet double insemination markedly improved outcomes (OR=4.75, 95% CI:2.78~8.10, P<0.05). The AIH and AID nomogram achieved AUC of 0.577 (95% CI:0.56~0.59) and 0.582 (95% CI:0.56~0.60) separately (P<0.05). Internal validation revealed acceptable calibration of the model, though its discriminative performance remained modest in both AID and AIH cohorts.ConclusionOvarian status and baseline fertility profiles differed substantially between the AIH and AID groups. The two nomogram models only demonstrated limited discrimination ability for predicting clinical pregnancy.Keywords: artificial insemination by husband, artificial insemination by donor, influencing factors, prediction model, predictive value Introduction Due to environmental pollution, economic pressure and food safety factors, there is a gradual decrease in human fertility.1 About 15% of the global population of reproductive age suffers from infertility.2 The use of assisted reproductive technology (ART) for infertility will be successful. Artificial insemination (AI) is the most practiced technique in ART which involves injecting optimized sperm directly into the uterine cavity of the woman during her ovulation period without sexual intercourse to help conception. AI is subdivided into artificial insemination by husband (AIH) and artificial insemination by donor (AID) based on semen origin, with entirely distinct clinical indications and baseline patient populations. AI has many advantages compared to surgical or pharmaceutical treatment, such as easy to operate, low price and fewer complications. This technology preserves the natural mix of sperm and egg and is the treatment of choice for patients with mild to moderate infertility.3 AIH is indicated for male patients with mild sperm defect and female patients with ovulatory disorder, partial tubal patency, uterine factors, sexual dysfunction as well as unexplained infertility.4 AID is indicated for male patient with azoospermia; his partner will usually have normal reproductive milestones.5 The existing models for predicting intrauterine artificial insemination rarely conduct stratified analysis of patients based on the source of sperm. The existing prediction models overlook the fundamental differences between AIH and AID candidates in terms of infertility etiology, female baseline fertility, and treatment response. Accurate pre treatment pregnancy risk stratification can help couples set realistic expectations, avoid repeated treatment cycles, and optimize ovulation stimulation and fertilization plans. This study retrospectively analyzed clinical data from a large sample of patients undergoing AIH and AID, compared the clinical characteristics between the two groups, and constructed a nomogram model to predict clinical pregnancy outcomes. Material and Methods Materials This study is a retrospective cohort study. A total of 11,651 cases of AIH patients and 3064 cases of AID patients who visited the Reproductive Medicine Center of Fujian Provincial Maternal and Child Health Hospital from January 2015 to June 2025 were included. The research protocol obtained formal ethical approval from the Institutional Review Board of Fujian Maternity and Child Health Hospital (Approval No: 2023KY093). All couples provided written informed consent for retrospective clinical data analysis, and all patient information was fully de-identified to protect privacy. The study strictly complied with the Declaration of Helsinki. Inclusion criteria: ① Female recipient age < 43 years old; ② Normal ovarian reserve markers (AMH, basal FSH within institutional reference range); ③ Unilateral or bilateral patent fallopian tubes confirmed by hysterosalpingography; ④ Normal uterine cavity morphology without submucous fibroids, severe endometrial polyps or uterine malformations; ⑤ Complete ultrasound follicle monitoring, semen processing and 14-day post-procedure serum HCG follow-up data available. Exclusion criteria: ① Recurrent spontaneous abortion history ≥ 2; ② Severe endometriosis stage III–IV; ③ Pre-cycle hormonal therapy interfering with follicular development; ④ Lost to follow-up without biochemical/clinical pregnancy confirmation; ⑤ Missing key clinical variables required for regression modeling.6 Methods Treatment Protocol Selection Transvaginal ultrasound was used to monitor follicle development and endometrial growth. Patients without ovulatory disorders who developed dominant follicles were treated via a natural cycle. Patients with ovulatory disorders, or those who had previously failed natural cycle treatment, received ovulation induction with letrozole or clomiphene. The medication regimen was adjusted individually based on ultrasound monitoring results, with the number of dominant follicles controlled to < 2.7 AID was performed for patients without sperm, while AIH was performed for patients with sperm. Semen Processing On the day of the procedure, semen was collected from the male partner or provided by a sperm bank, liquefied in a 37°C incubator, and active sperm were separated using a density gradient centrifugation method.8 A volume of 0.3–0.5 mL of sperm sediment was collected for later use. Surgical Process AI procedures were performed the day the dominant follicle reached a size of ≥ 18 mm and peak luteinizing hormone level was observed. Using a soft catheter, active sperm were injected into the uterine cavity of patients placed in a supine position, supplemented with knee flexion. After 1 hour post-operation, the patients were discharged and a follow-up ultrasound was conducted after 24 hours to check ovulation. The same procedure was followed on the same day if ovulation did not occur. Pregnancy Monitoring Oral dydrogesterone was given on the day of surgery for luteal phase support. On postoperative day 14, HCG level in the peripheral blood was measured to confirm biochemical pregnancy. Transvaginal ultrasound examination was performed at day 21 of gestation with positive HCG to check the size of gestational sac and confirm clinical pregnancy. Statistical Analysis Data analysis has been performed via SPSS 26.0 software. Continuous data were non-normally distributed and were presented as median (interquartile range). Intergroup comparisons were done using Mann-Whitney U-Test. The rates (%) were presented for categorical data and intergroup comparison was done by χ2 test. Single factor binary logistic regression analysis was used to screen factors related to pregnancy outcomes, and multiple factor binary logistic regression analysis was used to determine predictive factors. The predictive factor forest plot was created using GraphPad Prism 6 software. The OR is displaying the serious of relationship, and statistically significance is particularized as p-value < 0.05. A nomogram was developed using R language and its performance was evaluated using the ROC curve. An AUC of 0.5 represents no predictive, AUC values between 0.5–0.7 indicate limited/low discriminative performance. The Bootstrap is employed for internal validation, and the internal validation calibration curve is plotted. Results Comparison of Clinical Characteristics of Patients The clinical pregnancy rate of the AID cohort reached 26.50%, which was significantly higher than the 13.18% pregnancy rate observed in the AIH group (P < 0.01). There were significant differences between the two groups in terms of age, duration of infertility, proportion of secondary infertility, number of treatments, proportion of double fertilization, total number of motile sperm, proportion of ovulation induction cycles, and proportion of unexplained infertility (P < 0.05) (Table 1). Table 1Comparison of Clinical Characteristics Between Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025)CharacteristicsAIH Group (n=11651)AID Group (n=3064)Z/X2PFemale age (years)31.00 (29.00, 34.00)29.00 (27.00, 32.00)−25.47<0.01Male age (years)33.00 (30.00, 35.00)31.00 (28.00, 34.00)−20.18<0.01Duration of infertility (years)3.00 (2.00, 4.00)3.00 (2.00, 5.00)−6.84<0.01Number of treatment (time)1.00 (1.00, 2.00)2.00 (1.00, 3.00)−11.72<0.01Endometrial thickness (mm)10.10 (9.00, 11.50)10.20 (9.10, 11.50)−1.010.31Total number of active sperm (×106)63.40 (40.00, 90.00)8.60 (7.30, 9.70)−84.28<0.01Two insemination attempts (%)42.22 (4918/11,651)93.67 (2870/3064)25.78<0.01Secondary infertility (%)32.54 (3792/11,651)25.85 (792/3064)50.75<0.01Ovulation induction cycle (%)59.53 (6936/11,651)54.57 (1672/3064)24.61<0.01Normal uterine morphology (%)93.25 (10,865/11,651)93.54 (2866/3064)0.310.58Unexplained infertility (%)50.19 (5848/11,651)0 (0/3064)25.52<0.01Pregnancy rate (%)13.18 (1536/11,651)26.50 (812/3064)32.08<0.01 Comparison of Clinical Characteristics Between Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025) Comparison of Factors Influencing Pregnancy Outcomes in Two Groups The female age, duration of infertility, number of treatments, and proportion of unexplained infertility in the AIH pregnancy group were significantly lower than those in the non-pregnant group (P < 0.05); the proportion of double fertilization, endometrial thickness, and ovulation induction cycles were significantly higher than those in the non-pregnant group (P < 0.05). In the AID pregnancy group, the female age, male age, duration of infertility, and number of treatments were significantly lower than those in the non-pregnant group (P < 0.05); the proportion of double fertilization and endometrial thickness were significantly higher than those in the non-pregnant group (P < 0.05) (Table 2). Table 2Comparison of Factors Influencing Pregnancy Outcomes in Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025)CharacteristicsAIH (n=11651)Z/X2PAID (n=3064)Z/X2PNon-Pregnant Group (n=10115)Pregnancy Group (n=1536)Non-Pregnant Group (n=2252)Pregnancy Group (n=812)Female age (years)31.00 (29.00, 34.00)31.00 (29.00, 33.00)−2.440.0229.00 (27.00, 32.00)29.00 (26.00, 32.00)−2.110.04Male age (years)33.00 (30.00, 35.00)33.00 (30.00, 35.00)−1.350.1831.00 (28.00, 34.00)31.00 (28.00, 33.00)−2.400.02Duration of infertility (years)3.00 (2.00, 4.00)3.00 (2.00, 4.00)−3.88<0.013.00 (2.00, 5.00)3.00 (2.00, 4.00)−3.33<0.01Number of treatment (time)2.00 (1.00, 2, 00)1.00 (1.00, 2.00)−3.70<0.012.00 (1.00, 3.00)2.00 (1.00, 2.00)−3.15<0.01Endometrial thickness (mm)10.10 (9.00, 11.50)10.30 (9.20, 11.70)−1.980.0510.10 (9.10, 11.50)10.40 (9.30, 11.60)−2.040.04Total number of active sperm (×106)63.00 (40.00, 90.00)64.70 (40.50, 90.18)−0.860.398.60 (7.40, 9.70)8.50 (7.20, 9.68)−0.680.49Two insemination attempts (%)41.52 (4200/10,115)46.74 (718/1536)14.91<0.0192.05 (2073/2252)98.15 (797/812)37.46<0.01Secondary infertility (%)32.48 (3285/10,115)33.01 (507/1536)0.170.7026.20 (590/2252)24.88 (202/812)0.540.46Ovulation induction cycle (%)58.64 (5931/10,115)65.43 (1005/1536)25.55<0.0154.40 (1225/2252)55.05 (447/812)0.100.75Normal uterine morphology (%)93.33 (9441/10,115)92.71 (1424/1536)0.840.3693.38 (2103/2252)93.97 (763/812)0.330.56Unexplained infertility (%)51.28 (5187/10,115)43.03 (661/1536)36.27<0.010 (0/2252)0 (0/812)–– Comparison of Factors Influencing Pregnancy Outcomes in Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025) Univariate Binary Logistic Regression Analysis of the Two Groups in the AIH group, female age, duration of infertility, number of treatments, endometrial thickness, number of fertilizations, ovulation induction method, and whether the cause of infertility was clear were significantly associated with pregnancy outcomes (P < 0.05). In the AID group, female age, duration of infertility, number of treatments, and number of fertilizations were significantly associated with pregnancy outcomes (P < 0.05) (Table 3). Table 3Univariate Binary Logistic Regression Analysis for Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025)CharacteristicsClinical Pregnancy AIHClinical Pregnancy AIDOR (95% CI)POR (95% CI)PFemale age (years)0.98 (0.96, 1.00)<0.010.97 (0.96, 0.99)0.02Male age (years)0.99 (0.98, 1.00)0.160.98 (0.96, 1.00)0.07Duration of infertility (years)0.95 (0.93, 0.98)<0.010.95 (0.92, 0.98)<0.01Number of treatment (time)0.91 (0.85, 0.98)<0.010.89 (0.83, 0.95)<0.01Endometrial thickness (mm)1.03 (1.00, 1.07)0.031.05 (1.00, 1.10)0.07Total number of active sperm (×106)1.01 (0.99, 1.02)0.280.98 (0.95, 1.02)0.31Number of insemination attempts (1)1.24 (1.11, 1.38)<0.014.59 (2.69, 7.82)<0.01Sterile type (1)1.02 (0.91, 1.15)0.680.93 (0.78, 1.12)0.46Ovulation induction methods (1)1.34 (1.19, 1.49)<0.011.03 (0.87, 1.21)0.75Uterine morphology (1)0.91 (0.74, 1.12)0.361.10 (0.79, 1.54)0.56Infertility causes (1)1.39 (1.25, 1.55)<0.01––Note: (1) Represents only one surgery was performed, primary infertility, natural cycle, normal uterine morphology and unknown cause of infertility. Univariate Binary Logistic Regression Analysis for Artificial Insemination by Husband and Artificial Insemination by Donor (2015–2025) Note: (1) Represents only one surgery was performed, primary infertility, natural cycle, normal uterine morphology and unknown cause of infertility. Multivariate Binary Logistic Regression Analysis of Two Groups The female age, duration of infertility, number of treatments, endometrial thickness, number of fertilizations, ovulation induction methods, and whether the cause of infertility is clear are independent influencing factors of pregnancy outcomes in the AIH group (P < 0.05). The clarity of the cause of infertility is the most significant influencing factor for successful pregnancy in AIH (OR=1.37, 95% CI: 1.23–1.53). In the AID group, the duration of infertility, number of treatments, and number of fertilizations are independent influencing factors of pregnancy outcomes (P < 0.05). The number of fertilizations is the most significant influencing factor for successful pregnancy in AID (OR=4.75, 95% CI: 2.78–8.10) (Figure 1). Comparison of predictive factors for pregnancy outcomes in artificial insemination by husband and artificial insemination by donor. (A) Factors associated with pregnancy outcomes in artificial insemination by husband. (B) Factors associated with pregnancy outcomes in artificial insemination by donor. Comparison of Prediction Nomograms for Pregnancy Outcomes in Two Groups Prediction models for clinical pregnancy outcomes in the AIH and AID groups were developed in the form of nomograms using R language. The prediction probability of clinical pregnancy can be obtained from the nomogram of the total score, which is the summative score of the predictive indicators in the two models. The AUC values obtained from the ROC curves for both prediction models are 0.577 and 0.582 (P < 0.05) (Figures 2 and 3). Line chart of the artificial insemination by husband and artificial insemination by donor pregnancy prediction model. (A) Line chart of the artificial insemination by husband pregnancy prediction model. (B) Line chart of the artificial insemination by donor pregnancy prediction model. To predict the probability of pregnancy for patients undergoing artificial insemination, locate the corresponding score for each predictor on the “Points” axis according to individual clinical characteristics. Sum all scores to obtain the “Total Points”. Finally, draw a vertical line downward from the total points scale to the “Pregnancy rate” axis to read the corresponding predicted probability of pregnancy. ROC curve of the artificial insemination by husband and artificial insemination by donor pregnancy prediction mode. (A) ROC curve for the pregnancy prediction model based on artificial insemination by husband. (B) ROC curve for the pregnancy prediction model based on artificial insemination by donor. Internal Validation and Calibration of Predictive Models Randomly select 1000 samples from the AID and AIH queues for Bootstrap resampling for internal validation, to evaluate the discriminative ability and calibration performance of the prediction model (Figure 4). In the AID queue, the apparent AUC value is 0.582 (95% confidence interval: 0.560–0.604), and the adjusted AUC value is 0.581 (95% confidence interval: 0.536–0.625). In the AIH queue, the apparent AUC value is 0.577 (95% confidence interval: 0.562–0.592), and the adjusted AUC value is 0.572 (95% confidence interval: 0.542–0.603). Both do not exhibit overfitting. The calibration curve corrected for Bootstrap bias matches the ideal diagonal height within the observed prediction probability range. The calibration curve corrected for bias shows acceptable consistency between the predicted risk and the observed event occurrence rate. Bootstrap internal validation of the prediction model in the AID and AIH cohorts. (A) ROC curve for the AIH cohort. (B) ROC curve for the AID cohort. (C) Calibration curve for the AIH cohort. (D) Calibration curve for the AID cohort. Discussion Artificial insemination techniques offer numerous advantages in the treatment of infertility; however, their clinical pregnancy rates remain relatively low. Current data indicate that the pregnancy rate for AIH is 10–15%, while AID is 20–25%.9,10 In contrast, the pregnancy rate for in vitro fertilization (IVF) can reach 40–50%.11 Infertile patients often face a decision-making dilemma when choosing between AI and IVF. Research has shown that the pregnancy outcomes of AIH are influenced by various factors, including the patient’s baseline condition, the cause of infertility, the ovulation induction protocol, and the quality of the semen.12,13 The factors affecting pregnancy outcomes in AID are similar to those in AIH.5,14 Although both procedures share similarities in terms of surgical operation, indications, and exclusion criteria,6 there are significant differences in their clinical pregnancy rates.9,10 This suggests that the characteristics and independent influencing factors of the two populations undergoing AIH and AID may differ. This study systematically compares the differences in population characteristics, influencing factors for pregnancy, and predictive models between AIH and AID. The present study showed a significantly higher pregnancy rate in the AID group than in the AIH group (26.50% versus 13.18%). The AID group presented with younger age, shorter infertility duration and a lower proportion of secondary infertility, indicating a better underlying reproductive status in this population. It is worth mentioning that the AID group had a substantially lower total motile sperm count at insemination (8.60×106 vs 63.40×106). Despite this difference, its pregnancy rate remained superior. Collectively, these observations reveal no straightforward linear correlation between sperm quantity and pregnancy rate in artificial insemination. Sperm count exerts no additional effect on pregnancy outcomes after crossing the critical threshold. The result is in line with published literature.15–17 Additionally, the fewer treatment cycles and the higher proportion of unexplained infertility in the AIH group may also contribute to its lower pregnancy rate. In summary, our results demonstrate that the female baseline condition in the AID group is superior, and although the quantity of sperm used for insemination is lower, it meets the required standards, which is a reason for its higher pregnancy rate. Our findings indicate that AIH and AID share commonalities as well as differences in the factors influencing pregnancy outcomes. Comparing the influencing factors between the two groups, it can be seen that the decisive factors for pregnancy outcomes are similar, which is consistent with previous research results.5,12 Ogistic regression analysis shows that the duration of infertility, number of treatments, and number of inseminations are common influencing factors for both groups, but the odds ratio (OR) for the AID group is higher than that for the AIH group. Furthermore, AIH patients are also influenced by factors such as ovulation induction therapy, endometrial thickness, and the clarity of infertility etiology, which aligns with literature reports.9,18 Univariate regression analysis revealed an association between age and AID; however, multivariate analysis confirmed that age was not an independent influencing factor, likely due to the younger age and frequent reproductive history among AID patients. The differences in predictive models for pregnancy outcomes between AIH and AID reflect the heterogeneity of the population and the varying influencing factors. The factors negatively correlated with AIH pregnancy outcomes were identified through multiple logistic regression analysis, including female age, duration of infertility, number of treatments, and causes of infertility; The positively correlated factors include endometrial thickness, number of fertilizations, and ovulation induction protocol (OR=1.37). Factors negatively correlated with AID pregnancy outcomes included duration of infertility and number of treatments; the only positively correlated factor was the number of fertilizations (OR=4.33). Both results indicated that the pregnancy rate with two fertilizations is higher than with one; however, the weight of influencing factors is significantly greater in the AID group. Our research constructed a visual nomogram model using R language, converting relevant factors into standard scores to facilitate quick calculations of individualized pregnancy probabilities for clinicians. The model validation results showed that the AUC value for the AIH model was 0.577, while the AID model had an AUC value of 0.582. These results have been validated by internal testing and calibrated using the validation model. These results indicate that both models exhibit relatively weak predictive performance, which is consistent with findings from previous studies.19,20 Some studies have demonstrated that increasing relevant clinical and laboratory parameters, along with optimizing model algorithms, can improve predictive accuracy to 70–80%.21,22 Furthermore, results indicate that strict control of major influencing factors such as age and hormone levels can elevate the pregnancy rate of IUI from the original 15% to 30%.23 These finding suggests that the factors influencing pregnancy rates in artificial insemination are complex and their individual characteristics are dispersed. Some studies have shown that factors such as women’s anti-müllerian hormone (AMH) and baseline hormone levels may also affect AI pregnancy outcomes,20,24 while other studies have indicated no correlation.25 The relatively poor predictive performance of this study may be attributed to the exclusion of these factors. Conclusion This 10-year large-sample retrospective cohort study confirmed prominent heterogeneity in baseline fertility characteristics and independent pregnancy predictive factors between AIH and AID recipients, even under unified insemination inclusion criteria and operative protocols. We established separate logistic regression nomogram models to quantify the relative impact of clinical variables on conception probability for each treatment subgroup. However, the low AUC values reveal limited discrimination performance of both prediction tools. The visualized forest plots and nomograms only provide comparative quantitative reference of risk factor weights for reproductive clinicians. Funding Statement This work was supported by funding from the Joint Funds for the Innovation of Science and Technology, Fujian province (No.2023Y9385). Data Sharing Statement The datasets of current study are available from the corresponding author (Gangxin Chen) on reasonable request. Ethics Approval and Consent to Participate The study protocol received ethical approval from the Institutional Review Board of Fujian Maternal and Child Health Hospital, China (Approval No.: 2023KY093). Our study complies with the Declaration of Helsinki. Consent for Publication All authors consented to publication. Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors report no conflicts of interest in this work. References Associated Data Data Availability Statement The datasets of current study are available from the corresponding author (Gangxin Chen) on reasonable request.