Analysis of influencing factors of assisted reproduction and assisted pregnancy outcome of infertile male based on Logistic Regression and Decision Tree Model

doi:10.21203/rs.3.rs-3894248/v1

Download PDF

Research Article

Analysis of influencing factors of assisted reproduction and assisted pregnancy outcome of infertile male based on Logistic Regression and Decision Tree Model

https://doi.org/10.21203/rs.3.rs-3894248/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective: To study the influencing factors of assisted pregnancy outcome in infertile men receiving assisted reproduction.

Design: From January 2023 to June 2023, a total of 1037 infertile men who planned to undergo IVF/ICSI-ET assisted pregnancy in the Department of Assisted Reproductive Medicine of the First Maternal and Infant Health Hospital Affiliated to Tongji University were selected as the research objects. Logistic regression and classification decision tree model were used to study the influencing factors of infertile men. Receiver operating characteristic (ROC) curves were used to evaluate the effects of the two prediction models.

Subjects: Infertile men undergoing assisted reproduction

Main Outcome Measures: Assisted pregnancy outcome of infertile men and construction of prediction model based on Logistic and decision tree

Results: The two models showed that the percentage of grade A sperm, the percentage of grade B sperm, the sperm DFI, whether smoking or drinking alcohol were the influencing factors of assisted pregnancy outcome of infertile men. Logistic regression model showed that age, education level, daily exercise time, spermatozoa survival rate, anxiety, depression and insomnia were the factors affecting the outcome of assisted pregnancy in infertile men. Among them, the percentage of grade A sperm is the main influencing factor of infertile men. Compared with the two models, the sensitivity and specificity of Logistic regression model were 91.3% and 88.4% respectively. The sensitivity and specificity of decision tree model are 80.6% and 64.2% respectively.

Conclusion: Both Logistic regression and decision tree model have certain classification and prediction value, among which Logistic regression model has better prediction ability than decision tree model. Clinical medical staff can make predictive plans according to the prediction results, improve sperm quality as soon as possible, relieve negative emotions, and improve the outcome of assisted pregnancy with assisted reproductive technology.

Sterile Male

Assisted Reproduction

Logistic Regression

Decision Tree Model

Influencing Factor

According to the World Health Organization, male infertility refers to a symptom of male infertility caused by male factors caused by couples living together for more than 1 year without any contraceptive measures, accounting for about 50% of the infertility population [1]. Sperm quality is still an important index to evaluate male fertility [2], and in recent years, the quality of male sperm in China has generally shown a downward trend [3]. The quality of sperm is closely related to the clinical pregnancy outcome of assisted reproduction [4]. In addition to age, obesity and bad living habits, 30% to 50% of abnormal semen parameters cannot be identified as a clear cause [5]. However, more and more studies have pointed out that mental health problems such as anxiety, depression and stress can cause the decline of sperm quality. It aggravates infertility [6-7], and insomnia will also lead to an increase in the rate of sperm malformation and fragmentation, reduce the chance of conception [8], and have a negative impact on assisted reproduction and pregnancy outcomes. Although the success rate of IVF/ICSI-ET treatment has reached 40% to 50%, there are still many infertile families facing pregnancy failure. Considering the high cost of test-tube assisted pregnancy and the negative effects of treatment failure, it is necessary to study the related factors of assisted fertility outcomes in infertile men. Therefore, this study intended to screen the independent risk factors for the outcome of assisted reproduction of infertile men, construct a prediction model by using Logistic regression and decision tree, and select the optimal model to facilitate the early identification of potential risk of assisted pregnancy by medical personnel, and provide reference for the subsequent formulation of predictive treatment and nursing intervention.

1.1 Research object

This study was a cross-sectional survey, using convenience sampling method, and selected infertile men who underwent IVF/ICSI-ET assisted pregnancy in Department of Assisted Reproductive Medicine, Obstetrics and Gynecology Hospital Affiliated to Tongji University from March 2023 to September 2023 as the study objects. Inclusion criteria: (1) Meet WHO diagnostic criteria for male infertility; (2) The male reproductive system and physical examination are normal, no medical history affecting sperm quality; (3) The male has not received treatment that affects sperm quality before semen examination; (4) Informed consent and voluntary participation in the researcher. Exclusion criteria: (1) Patients who received sperm donation, surgical sperm extraction or frozen sperm during this assisted pregnancy cycle; (2) Patients with serious chronic diseases, tumors and other diseases; (3) Patients with poor ovarian reserve in their spouses were excluded: basal follicle-stimulating hormone (FSH) ≥10mIU/mL or sinus follicle number (AFC) < 5; Patients with uterine malformation, endometriosis, recurrent transmission; (4) chromosomal or genetic abnormalities of both spouses, preimplantation genetic testing (PGT); (5) Patients who have not completed embryo transfer, such as no embryo transfer cycle, embryo saving, embryo cryopreservation; (6) Patients with incomplete clinical data and incomplete information.

Clinical experts demonstrated and screened the factors that may affect the outcome of assisted pregnancy, including a total of 35 risk factors. According to the sample size calculation formula [9], 5 to 10 patients are required for each risk factor, and considering the sample loss rate of 10% to 20%, the pre-survey of small samples in our hospital shows that infertile men account for about 20.77% (43/207) of the population receiving assisted reproduction. The sample size of this study was 35×5× (1+0.2) ÷20.77%≈1011, and 1037 cases were eventually included. This study strictly adhered to the indications and complied with all laws, regulations and ethical principles, and was approved by the Ethics Committee of Obstetrics and Gynecology Hospital Affiliated to Tongji University (Ethics number: KS2313).

1.2 Methods

1.2.1 Survey Tools

1.2.1.1 General demographic information

The questionnaire was designed by oneself, including age, BMI, occupational status, household registration type, education level, annual household income, whether smoking, drinking alcohol, drinking tea, drinking cola, drinking coffee, daily sleep time and daily exercise time.

1.2.1.2 Sperm quality and secretion data collection

The results of the last semen examination before IVF/ICSI-ET were collected. The researchers were instructed to abstain from sex for 2 to 7 days before the examination, and to use the uniform semen and secretion treatment and analysis method. Semen routine (semen volume, sperm concentration, total sperm motility, etc.), sperm morphology (normal morphology rate, head malformation rate, mixed malformation rate), sperm DNA fragment index (DFI), sperm survival rate, anti-sperm antibodies, secretions (mycoplasma, chlamydia) and other data were extracted from the hospital's HIS system.

1.2.1.3 Sex hormones

3mL fasting venous blood was collected, serum was isolated and obtained, and serum total testosterone level was determined by chemiluminescence method.

1.2.1.4 Self-Rating Depression Scale (SDS) and Self-Rating Anxiety Scale (SAS) [10,11]

According to the symptoms of the patients in the last week, the 20 points of the scale were summed up to get the total crude score, standard score = crude score ×1.25. According to the Chinese norm results, the SDS standard was divided into 53 points, 53 ~ 62 points were classified as mild depression, 63 ~ 72 points were classified as moderate depression, and > 72 points were classified as severe depression. The SAS standard score is 50 points, of which 50 ~ 59 is classified as mild anxiety, 60 ~ 69 is classified as moderate anxiety, and > 70 and above is severe anxiety.

1.2.1.5 the Chinese version of perceived stress scale (CPSS)[12]

This scale was developed by Cohen equals in 1983 and revised by Chinese scholar Yang Tingzhong equals in 2003. Cronbach's alpha was 0.780, indicating high structural validity. The scale has 14 items in 2 dimensions. The total score between 11 and 26 indicates low perceived stress level, 27 to 41 indicates moderate stress level, and > 42 indicates high level.

1.2.1.6 Athens insomnia scale (AIS)

The scale was designed by Dan Sendmark[13] in 1985 and consisted of 8 items, with a total score ranging from 0 to 24, 0 to 3 as no sleep disorder, 4 to 6 as suspicious insomnia, and a total score > 6 as insomnia. After good reliability and validity test and strong diagnostic ability, Athens Sleep scale has become an internationally recognized self-assessment scale of sleep quality [14].

1.2.1.7 Observational indicators of assisted pregnancy outcome

The clinical pregnancy was observed in uterine cavity by B-ultrasonography 28 days after ET transplantation.

1.2.2 Investigation methods

After obtaining the consent of the patients, the male specialist nurses conducted a cross-sectional survey on the patients who met the criteria for scheduling on the day of operation. Semen and blood data were extracted from the clinical electronic medical record system. The unified guidance was used to ask patients to fill in the questionnaire on the spot according to the actual situation and give them enough time and an independent environment. A total of 1078 questionnaires were sent out, and 1037 were effectively collected, with an effective questionnaire recovery rate of 96.19%.

1.2.3 Statistical Methods

SPSS26.0 software was used for data analysis of the results of the two groups. The measurement data were expressed as mean ± standard deviation (±S), and the t test of two independent samples was used between groups. Counting data is expressed as "n (%)" and x2 test is used between the two groups. With assisted pregnancy outcome as dependent variable and statistically significant variables in univariate analysis as independent variables, binary Logistic regression model and chance Chi Square automatic interaction detection (CHAID) classification decision tree model were established respectively. Among them, the classification decision tree based on CHAID algorithm adopts x2 test or likelihood ratio x2 test results to determine the best grouping variables and segmentation points of the decision tree, and finally forms a classification tree [15]. In order to prevent the phenomenon of "over-fitting", the pre-pruning technology was applied to control the full growth of the decision tree [16] : the maximum tree depth was 3, the minimum sample size of the parent node was 100, and the minimum sample size of the child node was 50. Receiver operating characteristic (ROC) curve was drawn according to the predicted results of the model, and the difference between the two models was analyzed and compared by the area under ROC curve (AUC), sensitivity and specificity. Test level α=0.05.

2.1 General data of infertile men and single factor analysis affecting assisted pregnancy outcome

The results showed that Different age, education level, smoking, drinking, tea habit, daily sleep duration, daily exercise duration, SAS score, SDS score, CPSS score, AIS score, percentage of grade A sperm, percentage of grade B sperm, percentage of grade D sperm, sperm normal form rate, DFI, sperm survival rate, and sterile men with or without chlamydia infection Assisted pregnancy has different outcomes. See Table 1 for details.

2.2 Binary Logistic regression analysis of influencing factors of assisted pregnancy outcome of infertile men

Binary Logistic regression analysis showed that age, education level, smoking, drinking, daily exercise time, percentage of grade A sperm, percentage of grade B sperm, sperm DFI, sperm survival rate, anxiety, depression and insomnia were the main influencing factors on the outcome of assisted pregnancy in infertile men. Compared with infertile men with high school education or below, the higher the education, the better the outcome of assisted pregnancy. Infertile men who slept more than 9h per day had better pregnancy outcomes than those who slept at other hours. Compared with infertile men with little daily exercise time, the longer the daily exercise time, the better the pregnancy outcome; Compared with infertile men without anxiety, infertile men with mild anxiety and severe anxiety had worse assisted pregnancy outcomes. Compared with infertile men without depression, infertile men with mild depression and moderate depression had worse assisted pregnancy outcomes. Compared with infertile men without insomnia, men with suspicious insomnia and insomnia had worse assisted pregnancy outcomes. For details, see Table 2 and Table 3.

2.3 Classification and decision tree analysis of influencing factors of assisted pregnancy outcome of infertile men by CHAID algorithm

According to the established growth and construction rules, a classification decision tree model was established, including 3 layers, 15 nodes and 9 terminal nodes, as shown in Figure 1. As can be seen from the model diagram, the percentage of grade A sperm, the percentage of grade B sperm, the sperm DFI, whether smoking or drinking alcohol are the main influencing factors for the assisted pregnancy outcome of infertile men. Among them, the root node is the percentage of grade A sperm, indicating that it has the highest correlation with the assisted pregnancy outcome of infertile men. The percentage of grade A sperm was divided into four subgroups: ≤10.1, 10.1 ~ 18.30, 18.30 ~ 21.00 and > 21.00. 90.2% of infertile men with grade A sperm percentage > 21.00 were clinically pregnant. Only 7.6% of infertile men in the subgroup with A sperm percentage ≤10.1 achieved clinical pregnancy. The percentage of grade A sperm was ≤10.1 subgroup, which was affected by alcohol consumption. Infertile men who did not drink alcohol had a higher probability of clinical pregnancy than those who drank alcohol. The percentage of grade A sperm > 21.00 subgroup was affected by sperm DFI, and the probability of clinical pregnancy in infertile men with sperm DFI≤10.03 was higher than that in infertile men with sperm DFI > 10.03. The subgroup of grade A sperm with percentage of 10.1 to 18.30 was affected by the percentage of grade B sperm. The probability of clinical pregnancy in infertile males with sperm percentage of grade B > 18.40 was significantly higher than that in infertile males with sperm percentage of grade B ≤18.40. In infertile males with sperm percentage of grade B ≤18.40, the assisted pregnancy outcome was affected by whether or not smoking was involved. In infertile men with grade B sperm percentage > 18.40, the outcome of assisted pregnancy was affected by alcohol consumption.

2.4 Comparison of analysis results of binary Logistic regression model and classification decision tree model on influencing factors of assisted pregnancy outcome of infertile men

ROC curves were drawn according to the prediction probabilities obtained by the two models as state variables, as shown in Figure 2, and the classification effects were shown in Table 4. The ROC curves of the two models were far away from the diagonal line. Among them, the area under ROC curve of the binary Logistic regression model was 0.975[95%CI(0.967, 0.983)], the sensitivity was 91.3%, and the specificity was 88.4%. The area under ROC curve of the classification decision tree model based on CHAID algorithm is 0.890[95%CI(0.870, 0.910)], the sensitivity is 80.6%, and the specificity is 64.2%, indicating that the accuracy of the two models is different. By comparing the ROC curves of the two models, we can find that Z=9.568, P < 0.001, and Z=9.568. The test results were statistically different, indicating that there were differences between the two models.

3.1 Analysis of influencing factors of IVF/ICSI-ET assisted pregnancy outcomes for infertile men

3.1.1 Analysis of influencing factors of psychological factors on assisted pregnancy outcome of infertile men

This study shows that anxiety, depression and sleep status of infertile men are negatively correlated with assisted pregnancy outcome, while daily sleep time is positively correlated with assisted pregnancy outcome, which is consistent with previous research results [17-18]. The reason is that in addition to the pressure brought by reproductive defects, infertile men also need to face the pressure from family and society, which is easy to produce more complex and diverse negative emotions; Anxiety and depression will lead to changes in hormone levels in the body, resulting in temporary dysfunction of reproductive function [19]. At the same time, hormone level changes will also have a negative impact on sperm production and sperm function, and affect the outcome of assisted pregnancy; At present, the relatively consistent conclusion is that the decrease of sperm density or the increase of malformed sperm is significantly related to higher psychological pressure, which is the main reason for the decline of male fertility [20]. In addition, sleep quality is closely related to mental health and is of great significance to human reproduction. Long-term sleep disorders will lead to anxiety and depression [21], which will lead to disturbance of reproductive endocrine function. The spirits of people with long-term sleep deprivation are often in a state of excitement, which is not conducive to the secretion of male testosterone, and the decrease of androgen level has a certain impact on male spermatogenic function and sexual desire, leading to infertility [22] and affecting the outcome of assisted pregnancy.

3.1.2 Analysis of influencing factors of social factors on assisted pregnancy outcome of infertile men

This study showed that the age, education, smoking and drinking of infertile men were negatively correlated with assisted pregnancy outcome, while the daily exercise time was positively correlated with assisted pregnancy outcome. A large number of studies have shown that age is a key factor affecting male fertility, and sperm quality will decline with the increase of male age [23]. In this study, infertile men with a master's degree or above have better clinically assisted pregnancy outcomes. The reason is that people with higher education usually have better health literacy and medical awareness, better understanding and compliance with doctors' suggestions and guidance [24], and access to healthcare resources through multiple channels to help improve sperm quality and increase clinical pregnancy rate. Other studies have shown that tobacco and alcohol will increase sperm DNA fragmentation, reduce sperm concentration and motility, and produce abnormal sperm morphology [25-26], thus negatively affecting the outcome of assisted pregnancy. Moderate exercise can improve fertility potential. According to the research results of Vaamonde et al. [27], aerobic exercise ≥3 times per week for at least 2 to 4 hours per time can improve male fertility. However, according to the research of Riachy et al. [28], the sperm count and motility of men who cycle for more than 5 hours per week show a downward trend, and the relative success rate of assisted pregnancy also declines, which may be related to the type and intensity of exercise. This study only investigated the exercise time, and detailed investigation on exercise-related studies can be made in the future.

3.1.3 Analysis of influencing factors of sperm quality on assisted pregnancy outcome of infertile men

This study showed that sperm DFI was negatively correlated with assisted pregnancy outcome, while the percentage of grade A sperm, percentage of grade B sperm and sperm survival rate were positively correlated with assisted pregnancy outcome. A higher proportion of forward motile sperm may improve the success rate of fertilization and embryonic development. The combination of class A motile sperm and sperm-egg zona pellucida test can effectively predict whether IVF fertilization fails [29], which is similar to the results of Verheven G et al. [30]. Sperm DFI is an important indicator of sperm quality and plays an important role in early embryonic differentiation [31]. However, high DFI affects the integrity of sperm DNA, resulting in the loss of genetic information and the impact on embryonic development [32], thus reducing the clinical pregnancy rate of assisted pregnancy. Spermatozoa survival rate is also related to male fertility. Studies have shown that spermatozoa DFI combined with spermatozoa survival rate has a high predictive value for clinical pregnancy outcomes, and relevant prevention work can be done in advance according to the predicted results [33].

3.2 The prediction conclusions of IVF/ICSI-ET assisted pregnancy outcomes of infertile men by the two models were not completely consistent

Both models showed that the percentage of grade A sperm, the percentage of grade B sperm, the sperm DFI, whether smoking or drinking alcohol were the influential factors for the assisted pregnancy outcome of infertile men. However, there were differences between the two models. In the binary Logistic regression, age, education level, daily sleep time, daily exercise time, sperm survival rate, anxiety, depression and sleep quality had statistical significance. According to the standardized regression coefficient, smoking was the most important predictor of assisted pregnancy outcome for infertile men. In the classification decision tree model, the percentage of grade A sperm is the primary basis for predicting the outcome of assisted pregnancy in infertile men.

3.3 Analysis of predictive performance of the two models

The results of this study show that the performance of Logistic regression model is better than that of classification decision tree model, which is consistent with previous research results [34-35]. Logistic regression can calculate the quantitative dependency relationship between each meaningful independent variable and dependent variable, and the result is easy to interpret. The effect of independent variable on dependent variable can be quantified by OR value, which can better reflect the information of the change relationship between independent variable and dependent variable than decision tree. In addition, Logistic regression has strong robustness and is not prone to overfitting [35]. However, the decision tree model can eliminate the collinearity between variables, reflect the interaction between variables, go deep into the details of the data, intuitively show the importance and interaction of the predictor, and clearly express it in the form of a tree graph [35]. In clinical work, the advantages of the two models can be combined. First, meaningful independent variables can be screened by Logistic regression, and then the interaction between variables can be analyzed by classification decision tree model to achieve the optimal combination.

The sample size of this study is relatively small, and there may be sampling bias. The representativeness of a single center is insufficient, and the conclusion may have some limitations. In addition, in terms of controlling the influencing factors of assisted pregnancy outcome, female factors were not included, only male factors were considered; The outcome of assisted pregnancy was only considered clinical pregnancy, and the long-term outcome of assisted pregnancy was not explored. External verification has not been carried out in the verification of the model, in order to gradually improve it in the subsequent research. Clinical workers can use the model to predict the outcome of assisted pregnancy before assisted pregnancy, and timely intervene the negative emotions and bad lifestyle of infertile men to optimize sperm quality, so as to improve the outcome of assisted pregnancy.

Funding Statement:

National Natural Science Foundation of China Youth Project (82201882)

Author Contribution

K Wang: Methodology; Supervision; Writing-draftY Xu: Methodology; Supervision; Writing-review & editingJX Zheng: Data curation; Resources; Supervision; ValidationNX Qin: Formal analysis; Funding acquisition; Validation; Writing-review & editing J Bai: Project administration; Resources; SoftwareY Sun: Software; Supervision; ValidationYY Dong: Resources; Formal analysis; SoftwareZY Li: Conceptualization; ValidationAll authors reviewed the manuscript.

Agarwal A, Sharma R, Harlev A, et al. Effect of varicocele on semen characteristics according to the new 2010 World Health Organization criteria: a systematic review and meta-analysis[J]. Asian J Androl. 2016;18(2):163–70.
Keihani S, Verrilli LE, Zhang C, et al. Semen parameter thresholds and time-to-conception in subfertilecouples: How high is high enough? Hum Reprod. 2015;36(8):2121–33.
Lv MQ, Ge P, Zhang J et al. Temporal trends in semen concentration and count among 327373 Chinese healthy men from 1981 to 2019: A systematic review. Hum Reprod, 21,36(7):1751–75.
Mengchi S, Huawei W, Meng R, et al. The relationship between male oligoasthenospermia and embryo quality and pregnancy outcome in in vitro fertilization-embryo transfer [J]. Chin J Androl. 2019;35(01):7–10.
Salonia ACB, CarvalhoJ, Corona G, et al. EAU Guidelines on Sexual and Reproductive Health. EAU Annual Congress Milan 2021. Arnhem, The Netherlands: EAU Guidelines Office; 2020. pp. 106–16.
Wan Wan ZHU, Yajie H, Ping Y, Yujuan CHE, Xiaoyan. Research progress of male infertility health management in China [J]. Chin J Sexuality,202,31(08):9–13. (in Chinese).
Yana CAI, Yuezhi D. A meta-analysis of the correlation between psychological stress and semen quality in men [J]. Chin J Reprod Contracept, 2017, 37(04) : 320–6. (in Chinese).
Huaqiong BAO, Sun LAN, Xuedu Y, Jie D, Shiyuan M, Liu Y. Xu Xiaoou. The status quo and influencing factors of semen quality in men before pregnancy in Chongqing [J]. Chin J Androl 2018,32(03):41–6. (in Chinese).
CHEN Y, DU H, WEI B H. CHANG X N. Development and validation of risk-stratification delirium prediction model for critically ill patients: A prospective, observational, single-center study[J]. Med 2017, 96(29), e7543.
Zung WWK. A Fating instrument for Anxiety Disorders. Psychosomatics,1971, 12:371–379.
Zung WWK. A self-rating depression scale. Arch Gen Psychiattry, 1965,63–70.
Yang Tingzhong H, Hanteng. An epidemiological study on the psychological stress of urban residents in social transformation [J]. Chin J Epidemiol. 2003;24(9):760–4. (in Chinese).
Athens Insomnia Scale [J]. Medical World,2008(05):39. (in Chinese).
SOLDATOS C R, DIKEOS D G, PAPARRIGOPOULOS TJ. Athens insomnia scale: Validation of an instrument based on ICD-10 criteria[J]. J Psychosom Res. 2000;48(6):555–60.
Zixuan WEI, Anuo L, Mei Y, et al. Analysis of influencing factors of psychological capital of nurses in Anhui Province based on decision tree model and Logistic regression model [J]. J Nurs Res. 2019;37(16):2862–70.
Xue Wei. Decision tree technology and its application in data mining [J]. Stat Inform Forum,2002(02):4–10.
Gerber M, Brand S, Herrmann C, et al. Increased objectively assessed vigorous- intensity exercise is associated with reduced stress, increased mental health and good objective and subjective sleep in young adults. [J] Physiol Behavior. 2014;135(4):17–24.
Gollenberg AL, Liu FC, Drobnis EZ, et al. Semen quality in fertile men in relation to psychosocial stress [J]. Fertil Steril. 2010;93(4):1104–11.
Coban O, Serdarogullari M, Onar-Sekerci Z, et al. Evaluation of the impact of sperm morphology on embryo aneuploidy rates in a donor oocyte program[J]. Syst Biol Reprod Med. 2018;64(3):169–73.
Abu -- Musa A, Nassar AH, Hannoun AB. Usta. Effect of the Lebanese civil War on sperm parameters[J]. Fertil Steril. 2007;88(6):1579–582.
Liu Weizhong W, Chengmin J, Erni, et al. Relationship between insomnia, depression and anxiety in Shenzhen [J]. J Clin Psychiatry. 2019;33(06):457–61. (in Chinese).
Li Fengrui L. Effects of sleep disorders on infertility [J]. J Neuropharmacol. 2019;10(05):48–54. (in Chinese.
Zheying HU, Yu G, Ruijun G, et al. The relationship between sperm parameters and age and body mass index in male infertility patients [J]. Marker Immunoass Clin. 2019;29(06):913–7. (in Chinese).
Zheng Rong Z, Zili. Zhang Suan-suan. Higher education, Public health awareness and Health behavior: Impact of university enrollment Expansion on the prevention and control of COVID-19 [J]. Econ Manage Rev. 2019;36(06):5–15.
Zhiqiang WANG, Zhongjun D, Wensheng S, et al. Effect of sperm DNA damage on pregnancy outcome of in vitro fertilization-embryo transfer [J]. Chin J Sex Sci. 2019;32(01):11–4.
Minguez-Alarcon L, Chavarro JE, Gaskins AJ. Caffeine, alcohol, smoking, and reproductive outcomes among couples undergoing assisted reproductive technology treatments[J]. Fertil Steril. 2018;110(4):587–92.
Vaamonde D, Silva-Grigoletto D, Edir M, et al. Physically active men show better semen parameters and hormone values than sedentary men[J]. Eur J ApplPhysiol. 2012;112(9):3267–73.
Riachy R, McKinney K, Tuvdendorj DR. Various factors may modulate the effect of exercise on testosterone levels in men[J]. J Funct Morphol Kinesiol,2020:5(4):81.
Sifer C, Sasportes T, Barraud V, et al. World Health Organization grade‘a’motility and zona-binding test accurately predict IVF outcome for mild male factor and unexplained infertilities [J]. Hum Reprod. 2005;20(10):2769–75.
Verheyen G, Tournaye H, Staessen C, et al. Controlled comparison of conventional in-vitro fertilization and intracytoplasmic sperm injection in patients with asthenozoospermia. Hum Reprod. 1999;14(9):2313–9. (in Chinese).
Wang J, Yuqiang Z. Analysis of combined detection of sperm DNA fragmentation rate and sperm survival rate in predicting pregnancy outcomes of in vitro fertilization-embryo transfer [J]. J Reprod Med. 2019;32(05):671–6.
Zhou Chunhong X, Anran W, Xiujuan et al. Progress in the clinical application of sperm DNA integrity [J]. Chin J Eugenics Genet 2018,26(12):125–6. (in Chinese).
Rahban R, Priskorn L, Senn A. etc. NICER Working Group. Semen quality of young men in Switzerland: a nationwide cross-sectional population-based study. Andrology. 2019;7(6):818–26.
Geng JX, Xu JN, Yu L, et al. Analysis of multiple factors influencing pregnancy outcome and establishment of prediction model for intrauterine artificial insemination [J]. Chin J Reprod Contracept. 2019;43(8):792–8.
Wenwen W, Xiaodong T, Donghan S, et al. Application of Logistic regression analysis model and decision tree analysis in early warning indicators of hypertension and diabetes co-morbidity [J]. Chin J Disease Control. 2019;26(07):827–33.

Table 1 General data of infertile men and univariate analysis of outcomes affecting assisted pregnancy (n=1006)

item	Research group 403 people	Control group 634 people	t/ x²	P
Age	34.00±4.59	35.96±5.65	-5.854	0.000^***
BMI	24.64±3.21	25.35±3.43	-3.313	0.143
Occupational status			0.959	0.327
Be on the job	346（85.86%）	530（83.59%）
On the job	57（14.14%）	104（16.41%）
Type of household registration			1.955	0.162
Towns	215（53.35%）	310（48.89%）
Village	188（46.65%）	324（51.11%）
Educational level			23.019	0.000^***
High school and below	113（28.04%）	255（40.22）
College / Undergraduate	193（47.89%）	287（45.27%）
Master degree or above	97（24.07%）	92（14.51%）
Gross annual household income			2.911	0.406
Less than 100,000 yuan	122（30.27%）	172（27.13%）
110,000 to 500,000yuan	233（57.82%）	365（57.57%）
51 to 1 million yuan	32（7.94%）	66（10.41%）
More than 1 million yuan	16（3.97%）	31（4.89%）
Whether you smoke or not			247.559	0.000^***
Yes	29（7.20%）	344（54.26%）
No	374（92.80%）	282（44.48%）
Whether to drink			144.786	0.000^***
Yes	82（20.35%）	370（58.36%）
No	321（79.65%）	264（41.64%）
Have a habit of drinking tea			10.557	0.001^***
Yes	131（32.51%）	270（42.59%）
No	272（67.49%）	364（57.41%）
Have the habit of drinking coke			0.782	0.376
Yes	84（20.84%）	118（18.61%）
No	319（79.16%）	516（81.39%）
Have a coffee habit			0.330	0.566
Yes	91（22.58%）	153（24.13%）
No	312（77.42%）	481（75.87%）
Daily Sleep Duration (hour)			104.625	0.000^***
Less than 5	5（1.24%）	14（2.21%）
5～7	91（22.58%）	334（52.68%）
7～9	274（67.99%）	272（42.90%）
More than 9	33（8.19%）	14（2.21%）
Daily Exercise Duration (hour)			65.410	0.000^***
Hardly	54（13.40%）	209（32.97%）
0～0.5h	138（34.24%）	223（35.17%）
0.5～1h	131（32.51%）	139（21.92%）
More than 1h	80（19.85%）	63（9.94）
SAS score			20.625		0.000^***
No	276（68.49%）	367（57.89%）
Mild	85（21.09%）	146（23.03%）
Moderate	33（8.19%）	72（11.36%）
Severe	9（2.23%）	49（7.72%）
SDS score			23.830		0.000^***
No	279（69.23%）	351（55.36%）
Mild	102（25.31%）	206（32.49%）
Moderate	15（3.72%）	57（8.99%）
Severe	7（1.74%）	20（3.16%）
CPSS score			31.610		0.000^***
No	137（34.00%）	122（19.24%）
Moderation	163（40.44%）	292（46.06%）
High pressure	96（23.82%）	194（30.6%）
Extremely high pressure	7（1.74%）	26（4.1%）
AIS score			46.190		0.000^***
No	132（32.75%）	99（15.62%）
Suspicious insomnia	151（37.47%）	255（40.22%）
Insomnia	120（29.78%）	280（44.16%）
Grade A sperm percentage	20.62±7.66	12.96±4.75	19.885		0.000^***
Grade B sperm percentage	24.59±7.85	17.06±6.05	17.366		0.000^***
Grade C sperm percentage	16.37±6.10	16.99±5.45	-1.691		0.091
Grade D sperm percentage	38.86±12.53	53.08±10.20	-19.985		0.000^***
Head deformity rate	78.90±4.14	80.43±31.67	-0.964		0.441
Mixed malformation rate	17.53±4.15	17.47±3.98	0.215		0.357
Normal form rate	3.81±1.98	3.42±1.76	3.303		0.023^*
High stainable sperm index	6.52±3.69	6.70±4.04	0.708		0.424
DFI	8.52±4.49	17.29±9.78	-16.898		0.000^***
Sperm survival rate	83.53±8.51	70.88±16.01	14.597		0.000^***
Sperm concentration	26.82±18.41	24.93±17.78	1.643		0.960
Semen volume	3.55±1.44	3.47±1.47	0.915		0.109
Semen PH	7.25±0.17	7.26±0.17	-1.262		0.207
Spermatocyte	6.78±1.55	6.59±1.52	1.844		0.982
Antisperm antibody			1.852		0.174
Positive	34（8.44%）	70（11.04%）
Negative	369（91.56%）	564（88.96%）
Mycoplasma			0.016		0.900
Positive	50（12.41%）	77（12.15%）
Negative	353（87.59%）	557（87.85%）
Chlamydia	19.59±3.00	19.31±2.58	7.442		0.006^**
Positive	36（8.93%）	93（14.67%）
Negative	367（91.06%）	541（85.33%）
Serum testosterone	5.52±2.00	5.53±1.92	0.076		0.328

Notes：Significant at P **< 0.05, P **<0.01, P **<0.001

Table 2 Variable assignment

Variable	Assignment mode
Educational level	High school and below=0，College / Undergraduate=1，Master degree or above=2
Whether you smoke or not	No=0，Yes=1
Whether to drink	No=0，Yes=1
Have a habit of drinking tea	No=0，Yes=1
Daily Sleep Duration	Less than 5h=0, 5～7h=1，7～9h=2，More than 9h=3
Daily Exercise Duration	Hardly=0，Within 30min=1，30～60min=2，More than 60min=3
Chlamydia	Negative=0，Positive=1
Anxiety	Anxiety free=0，Mild anxiety=1，Moderate anxiety=2，Severe anxiety=3
Depressed	No depression=0，Mild depression=1，Moderate depression=2，Severe depression=3
Perceived pressure	No pressure=0，Mild pressure=1，Moderate pressure=2，Severe pressure=3
Insomnia	No insomnia=0，Suspicious insomnia=1，Insomnia=2

Table 3 Results of binary Logistic regression analysis of assisted pregnancy outcomes of infertile males (n=1037)

Research factor	B	S.E.	*Wald*	P	OR	95% CI
Research factor	B	S.E.	*Wald*	P	OR	Lower limit	Upper limit
Constant	-10.274	2.852	12.981	0.000^***	0.000
Age	-0.078	0.029	7.471	0.006^**	0.925	0.874	0.978
Educational level (High school and below)			7.113	0.029^*
College / Undergraduate	0.566	0.290	3.799	0.049^*	1.761	0.997	3.110
Master degree or above	0.973	0.394	6.081	0.014^*	2.645	1.221	5.730
Whether you smoke or not (No)	-2.662	0.352	57.090	0.000^***	0.070	0.035	0.139
Whether to drink (No)	-1.123	0.299	14.071	0.000^***	0.325	0.181	0.585
Have a habit of drinking tea (No)	0.164	0.280	0.345	0.557	1.178	0.681	2.039
Daily Sleep Duration（Take less than 5 hours as reference）			25.025	0.000^***
5～7h	-0.214	0.882	0.059	0.809	0.808	0.143	4.553
7～9h	0.950	0.869	1.194	0.274	2.586	0.471	14.208
More than 9h	2.298	1.041	4.875	0.027^*	9.955	1.294	76.562
Daily Exercise Duration（Take hardly as a reference）			16.294	0.001^**
Less than 30min	0.954	0.358	7.093	0.008^**	2.597	1.287	5.241
30～60min	1.360	0.383	12.583	0.000^***	3.894	1.837	8.254
More than 60min	1.570	0.463	11.504	0.001^**	4.806	1.940	11.907
Grade A sperm percentage	0.192	0.031	38.129	0.000^***	1.212	1.140	1.288
Grade B sperm percentage	0.154	0.033	21.765	0.000^***	1.166	1.093	1.244
Grade D sperm percentage	0.041	0.023	0.838	0.080	1.042	0.995	1.091
Normal form rate	-0.078	0.077	1.010	0.315	0.925	0.795	1.076
Sperm DFI	-0.191	0.027	52.009	0.000^***	0.826	0.784	0.870
Sperm survival rate	0.099	0.014	51.381	0.000^***	1.104	1.074	1.134
Chlamydia (negative)	-0.296	0.443	0.445	0.505	0.744	0.312	1.773
Anxiety (with no anxiety as reference)			16.387	0.001^**
Mild	-0.953	0.325	8.593	0.003^**	0.386	0.204	0.729
Moderate	-0.765	0.507	2.273	0.132	0.465	0.172	1.258
Severe	-1.872	0.592	9.995	0.002^**	0.154	0.048	0.491
Depression (with no depression as reference)			12.831	0.005^**
Mild	-0.726	0.323	5.045	0.025^*	0.484	0.257	0.912
Moderate	-1.977	0.618	10.252	0.001^**	0.138	0.041	0.464
Severe	-1.074	1.580	0.462	0.497	0.342	0.015	7.556
Fertility pressure (with no stress as reference)			1.513	0.679
Moderate pressure	-0.276	0.331	0.695	0.404	0.759	0.397	1.452
High pressure	-0.126	0.374	0.113	0.737	0.882	0.423	1.837
Heavy pressure	-864	0.822	1.106	0.293	0.421	0.084	2.110
Insomnia (with no insomnia as reference)			9.192	0.010^*
Suspicious insomnia	-1.092	0.365	8.965	0.003^**	0.336	0.164	0.686
Insomnia	-0.887	0.378	5.512	0.019^**	0.412	0.196	0.864

Notes：Significant at P **< 0.05, P **<0.01, P **<0.001

Table 4 Comparison of classification effect between binary Logistic regression model and classification decision tree model

Model	Area under the curve	Standard error	Asymptotic significance	Asymptotically 95%CI
Model	Area under the curve	Standard error	Asymptotic significance	Lower limit	Upper limit
Binary Logistic regression model	0.975	0.004	＜0.001	0.967	0.983
Classification decision tree model	0.890	0.010	＜0.001	0.870	0.910

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Analysis of influencing factors of assisted reproduction and assisted pregnancy outcome of infertile male based on Logistic Regression and Decision Tree Model

Status:

Version 1

Abstract

Figures

Introduction

1 Materials And Methods

2 Results

3 Discussion

4 Research limitations and suggestions

Declarations

Funding Statement:

Author Contribution

References

Tables

Additional Declarations

Status:

Version 1