# IVF Literature Summary

### 1. Individualized decision-making in IVF: calculating the chances of pregnancy

**Authors:** L L van Loendersloot, M van Wely, S Repping, P M M Bossuyt, F van der Veen

**Year:** 2013

**Outcome:** Pregnancy after first stimulated IVF cycle (fresh or frozen).

**Cohort:** Training data includes 2621 cycles in 1326 patients between January 2001 and July 2009, at the Centre for Reproductive Medicine of the Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands. Validation data includes 515 cycles in 440 couples treated between August 2009 and April 2011. Homologous. Excludes surgically retrieved spermatozoa, patients positive for human immunodeficiency virus, modified natural IVF and cycles cancelled owing to poor ovarian stimulation, ovarian hyperstimulation syndrome or other unexpected medical or non-medical reasons.

**Predictors:** Thirteen covariates:
1. Female age
2. duration of subfertility
3. previous ongoing pregnancy
4. male subfertility
5. diminished ovarian reserve
6. endometriosis
7. basal FSH
8. number of failed IVF cycles.

After the first cycle:

9. fertilization
10. number of embryos
11. mean morphological score per Day 3 embryo
12. presence of 8-cell embryos on Day 3
13. presence of morulae on Day 3

**Model:** Logistic regression

**AUC:** 0.63-0.73 in 95% confidence interval; mean 0.68.

### 2. Predicting Live Birth, Preterm Delivery, and Low Birth Weight in Infants Born from In Vitro Fertilisation: A Prospective Study of 144,018 Treatment Cycles

**Authors:** Scott M. Nelson, Debbie A. Lawlor

**Year:** 2011

**Outcome:** Live birth in all IVF cycles (fresh or frozen).

**Cohort:** Training data includes 144,018 cycles between 2003 and 2007 in the [Human Fertilisation and Embryology Authority database](https://www.hfea.gov.uk/about-us/our-data/).

**Predictors:** Templeton covariates:

1. Maternal age
2. Duration and cause of infertility
3. Previous number of IVF attempts
4. Number of previous spontaneous
5. IVF live births
6. Source of gametes
7. Cycle number

New covariates:

9. Causes of infertility
10. Source of the egg
11. Type of hormonal preparation used (antioestrogen, gonadotrophin, or hormone replacement therapy)
12. ICSI (yes or no)
13. Number of the treatment cycle (1, 2 or ≥3)

**Model:** Logistic regression with interaction terms

**AUC:** 0.6184 (0.6152–0.6217) for Templeton model. 0.6335 (0.6202–0.6367) for new model.

***Notes:*** They also study specifically the effect of ICSI, also risk factors for low birth weight. [The anonymized database](https://www.hfea.gov.uk/about-us/our-data/guide-to-the-anonymised-register/) is available for download from 1991-2015.

### 3. Factors that affect outcome of in-vitro fertilisation treatment

**Authors:** A Templeton, J K Morris, W Parslow

**Year:** 1996

**Outcome:** Live birth in fresh cycle

**Cohort:** Training data includes 36961 cycles from 26,389 women between August, 1991, and April, 1994 in the [Human Fertilisation and Embryology Authority database]. Cycles that involved gamete or embryo donation, frozen embryo transfer, or micromanipulation and unstimulated cycles were excluded.

**Predictors:**

1. Maternal age
2. Previous live-births or pregnancies and whether this was result of IVF treatment
3. Female causes of infertility
4. Duration of infertility
5. Number of previous unsuccessful IVF treatments

**Model:** Logistic regression with out interaction terms

**AUC:** unknown. Hosmer-Lemeshow text is p=0.73.

### 4. Multivariate Analysis of Factors Predictive of Successful Live Births in In Vitro Fertilization (IVF) Suggests Strategies to Improve IVF Outcome

**Authors:** Demetrios Minaretzis, Doria Harris, Michael M. Alper, Joseph F. Mortola, Merle J. Berger, and Douglas Power

**Year:** 1997

**Outcome:** Live birth in fresh cycle

**Cohort:** Training data includes 554 cycles at Boston IVF. All patients underwent ovarian hyperstimulation by the same protocol.

**Predictors:**

1. Maternal age
2. Cause for intervention
3. Donor insemination
4. Rank of attempt
5. Serum LH and E2 levels on day of hCG administration
6. Embryo transfer catheter (flexible vs rigid)
7. Number of embryos transferred of each morphologic type and developmental stage

Additional:

9. Sperm parameters (concentration, percentage motility and rate of progression) before and after Percoll processing
10. Sperm concentration at insemination
11. Number and quality of retrieved oocytes
12. "Human" factor

**Model:** Logistic regression with out interaction terms

**AUC:** unknown. Model fit was not evaluated.

***Note:*** Additional covariates showed little correlation with outcome.

### 5. Computational prediction of implantation outcome after embryo transfer

**Authors:** Behnaz Raef, Masoud Maleki, Reza Ferdousi

**Year:** 2020

**Outcome:** Pregnancy

**Cohort:** Training data includes 500 patients (one cycle each) the East Azerbaijan ACECR ART center (Tabriz, Iran) from April 2016 to February 2018. 251 samples positive β-HCG and 249 samples negative β-HCG. Excluded other infertility treatment procedures at this clinic that does not include embryo transfer.

**Predictors:**

Clinical data (patient-related data):

1. Age of female
2. Age of male
3. BMI (body mass index)
4. Family relation of couples
5. Family relation in parents of couples Smoking
6. Type of infertility
7. Infertility duration
8. Contraception duration
9. Infertility in family
10. G (gravida/gravidity)
11. P (para/parity)
12. Ab (abortion)
13. EP (ectopic pregnancy)
14. L (living children)
15. D (dead children)
16. Comorbidity diseases
17. Anemia
18. Thyroid disease
19. Prolactin hormone disorders
20. Drug usage

Female pathology data:

22. Amenorrhea (absence of menstruation) Dysmenorrhea (painful periods)
23. Period status
24. Hirsutism (excessive body hair in women)
25. Galactorrhea (abnormal milky breast discharge) Gynecological surgery
26. Oocyte donation
27. AFC (antral follicle count)
28. Endometrium (tissue lining of the uterus) thickness Three-line (regular/normal) endometrium
29. Uterus depth
30. Size of follicles
31. Tubal factor
32. Pelvic factor
33. Cervical factor
34. Ovulatory factor
35. PCOS (polycystic ovary syndrome)
36. Uterine factor
37. Endometriosis (abnormal growth of endometrium in outside of the uterus cavity)
38. Endometrial factor
39. Vaginitis
40. RIF (repeated implantation failure)
41. RPL (recurrent pregnancy loss)

Male pathology data:

42. Male factor
43. Male genital surgery
44. Varicocele (abnormal enlargement of the testicular veins)
45. TESE (testicular sperm extraction)
46. PESE (percutaneous epididymal sperm extraction) Fresh/freeze sperm

Semen analysis data:

47. Sperm count
48. Normal morph
49. Immotile

Lab tests:

50. FSH (follicle-stimulating hormone)
51. LH (luteinizing hormone) Estradiol
52. vitD3 Levels

Oocyte stimulation and morphology:

53. FSH/HMG (human menopausal gonadotropin) dosage
54. GnRH (gonadotropin-releasing hormone) antagonists Dosage
55. GnRH agonists dosage
56. Duration of stimulation (days)
57. Estradiol dosage
58. No. estradiol days
59. Number of retrieved oocytes
60. Number of MII (metaphase II) quality oocytes
61. Number of MI (metaphase I) quality oocytes
62. Number of GV (germinal vesicle) quality oocytes
63. Number of degenerated quality oocytes
64. Quality of injected MII oocytes

Embryological data:

65. Number of 2PN (pronuclear)
66. Number of developed embryos
67. Quality of developed embryos
68. Quality of vitelline space
69. ET (embryo transfer) strategies
70. ET day
71. Number of transferred embryos
72. Number of blastomeres
73. Quality and stages of transferred embryos Experience of ET

Others:

74. PRP (platelet-rich plasma)
75. β-HCG (human chorionic gonadotropin)

**Model:**
Six models
1. KNN
2. SVM
3. Neural Networks
4. Naive Bayes
5. Random Forest
6. Decision Tree

with feature selection (78-82 features selected).

**AUC:** .87 to .97

***Note:*** No mention of how the model is tuned. I don't trust these results.

### 6. Can we predict the IVF/ICSI live birth rate?

**Authors:** José Luis Metello, Claudia Tomás, Pedro Ferreira

**Year:** 2019

**Outcome:** Live birth in fresh cycle

**Cohort:** 739 IVF/ICSI cycles at the Division of Reproductive Endocrinology and Infertility, Garcia de Orta Hospital, Almada, Portugal between 2012-2016. Only cycles with a live birth delivery after 24 weeks, or cycles with no sur- plus embryos left were considered. Women’s age at oocyte retrieval varied between 18-39 years old.

**Predictors:**

1. AMH
2. AFC
3. women’s and men’s age
4. body mass index (BMI) both for men and women
5. smoking status
6. previous diagnosis
7. type of treatment (IVF/ICSI)
8. having had previous deliveries

**Model:** Binary regression with out interaction terms

**AUC:** 0.688 (0.649-0.728)

***Note:*** the categorical variables have many possible values yielding many covariates (63 total).

### 7. Predicting the chances of a live birth after one or more complete cycles of in vitro fertilisation: population based study of linked cycle data from 113 873 women

**Authors:** David J McLernon, Ewout W Steyerberg, Egbert R Te Velde, Amanda J Lee, Siladitya Bhattacharya

**Year:** 2016

**Outcome:** Cumulative chances of a first live birth for a couple having up to six complete cycles of IVF. One complete cycle included all fresh and frozen embryo transfers resulting from one episode of ovarian stimulation.

**Cohort:** 253,417 women who started IVF (including intracytoplasmic sperm injection) treatment in the UK from 1999 to 2008 using their own eggs and partner’s sperm. After exclusion 113,873 women with 184,269 complete cycles. All licensed IVF clinics in the UK. National data from the Human Fertilisation and Embryology Authority register.

**Predictors:**
Predictors available in the HFEA dataset. Predictors available: body mass index of the woman, ethnicity, smoking status, alcohol intake, and measures of ovarian reserve such as antral follicle count.

**Model:** Pre-treatment and post-treatment models. Discrete time logistic regression model to predict the chance of a live birth after a maximum of six cumulative complete cycles of IVF or ICSI, where a complete cycle included a fresh embryo transfer and any associated frozen-thawed embryo transfers. Used a random intercept for the effect of IVF centre. Variable selection thru backwards selection.

**AUC:** pretreatment model 0.73 (0.72 to 0.74); post-treatment model 0.72 (0.71 to 0.73)

***Note:*** treatment year was highly associated with live birth, signifying improvements in technology over time.

### 8. Personalized prediction of live birth prior to the first in vitro fertilization treatment: a machine learning method

**Authors:** Jiahui Qiu, Pingping Li, Meng Dong, Xing Xin and Jichun Tan

**Year:** 2016

**Outcome:** Cumulative live birth chance of the first complete IVF cycle

**Cohort:** 7,188 women at the Reproductive Medical Center of Shengjing Hospital, China from January 2014 to December 2018. Excluded women with no pregnancy outcomes follow-up and incomplete cases with missing data in any study feature. Exclusion criteria also included previous IVF/ICSI attempts, using frozen gametes, donor oocyte/sperm cycles and PGD/PGS cycles.

**Predictors:**
Pre-treatment predictors:
1. age
2. AMH
3. BMI
4. duration of infertility
5. previous live birth
6. previous miscarriage
7. previous abortion
8. type of infertility (Tubal, Anovulatory, Male factor, Others, Unexplained)

**Model:**
1. Logistic regression
2. Random forest
3. SVM
4. XGBoost

**AUC:**
1. Logistic regression 0.72 $\pm$ 0.01
2. Random forest 0.73 $\pm$ 0.01
3. SVM 0.72 $\pm$ 0.01
4. XGBoost 0.74 $\pm$ 0.01

### 9. Factors predicting the cumulative outcome of IVF/ICSI treatment: a multivariable analysis of 2450 patients

**Authors:** Q.F. Cai, F. Wan, R. Huang, and H.W. Zhang

**Year:** 2011

**Outcome:** Clinical pregnancy after a completed IVF/ICSI cycle (fresh plus cryopreserved embryos transferred from one stimulated cycle)

**Cohort:** 2,450 women with one complete cycle each at the Women and Children’s Hospital of Guangdong Province, P.R. China between between January 2002 and December 2007. Excluded: (i) cycles that involved oocyte or sperm donation or in vitro maturation; (ii) unstimulated cycles; (iii) cycles that result in neither fresh nor frozen – thawed embryo transfer in a completed treatment cycle; (iv) patients who had not become pregnant but still had frozen embryos left; and (v) cycles that involved blastocyst transfer, to keep embryo quality and quantity variables consistent and comparable for all patients.

Validation data consists of 256 patients who attended the reproductive medicine center
at Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, P.R. China, between January and May 2010.

**Predictors:**
1. total number of good-quality embryos
2. total number of embryos
3. progesterone level on the day of hCG injection
4. endometrial thickness on the day of hCG injection
5. antral follicle count (AFC)
6. duration of infertility
7. age
8. fertilization rate
9. number of follicles measuring between 10 and 14 mm in diameter on the day of hCG injection

**Model:** Logistic regression with no interaction terms and some variables polynomially transformed. Variable selection for multicolinearity (selecting for clinical significance and larger variance). Then bootstrap variable selection.

**AUC:** 0.778 (0.763 – 0.801)

### 10. A prediction model for selecting patients undergoing in vitro fertilization for elective single embryo transfer

**Authors:** Claudine C Hunault, M.D., Marinus J.C Eijkemans, Math H.E.C Pieters, M.D., Ph.D., Egbert R te Velde, M.D., Ph.D., J.Dik F Habbema, Ph.D., Bart C.J.M Fauser, M.D., Ph.D., Nicholas S Macklon, M.D., Ph.D.

**Year:** 2002

**Outcome:** pregnancy and twin pregnancy following the transfer of two embryos.

**Cohort:** 642 undergoing their first IVF treatment cycle in which no more than two embryos were transferred at the University Hospital Rotterdam between December 1993 and December 1998.

**Predictors:**
1. Woman’s age (per y)
2. Duration of infertility (per y)
3. Secondary type of infertility

Indication for IVF:
4. Tubal
5. Male factor
6. Idiopathic infertility
7. Others

8. Total no. of sperm cells (per 10−7/mL)
9. Progressive motile sperm cells (per %)
10. Estrogen level (per 10−3 pmol/L)
11. No. of preovulatory follicles (per follicle)
12. No. of retrieved oocytes (per oocyte)
13. Proportion of oocytes fertilized (per 10%)

Day of ET:

14. Day 3
15. Day 4
16. Day 5
17. No. of embryos suitable for transfer (per embryo)

Stage development of the best embryo:

18. Retarded
19. Appropriate
20. Advanced

Stage development of the second best embryo:

21. Retarded
22. Appropriate
23. Advanced<br><br>
24. Morphology score of the best embryo (range 1–4)
25. Morphology score of the second best embryo (range 1–4)

**Model:** Logistic regression with no interaction terms. Variable selection

**AUC:** 0.68 for pregnancy, 0.71 for twin pregnancy.

### 11. Prediction model for live birth in ICSI using testicular extracted sperm

**Authors:** A.M. Meijerink, M. Cissen, M.H. Mochtar, K. Fleischer,
I. Thoonen, A.A. de Melker, A. Meissner, S. Repping, D.D.M. Braat,
M. van Wely, and L. Ramos

**Year:** 2016

**Outcome:** live birth in couples undergoing ICSI after successful testicular sperm extraction (TESE-ICSI). An ICSI cycle is a fresh cycle and the corresponding cryo embryo cycle(s) derived from it.

**Cohort:** 526 couples undergoing 1006 TESE-ICSI cycles between September 2007 and May 2014 at the Radboud university medical center, The Netherlands (Radboudumc). Validation set is 289 couples undergoing 553 TESE-ICSI cycles between August 2007 and September 2015 in the Academic Medical Center (AMC).

**Predictors:**
1. Type of infertility (primary/secondary);
2. Duration of infertility (months);
3. Female age (years);
4. Parity (n);
5. Average menstrual cycle length (days);
6. Uterine abnormalities (yes/no);
7. Antral follicle count before stimulation (number of follicles ,11 mm);
8. Alcohol use (self-reported; yes/no) for male and female;
9. Smoking status (self-reported; yes/no) for male and female;
10. BMI at baseline (kg/m2) for male and female;
11. Male age (years);
12. Male testosterone (nmol/l);
13. Male inhibin B (ng/l);
14. Male FSH (IU/l);
15. Male LH (IU/l);
16. Total testicular volume (cc);
17. Suspected primarily diagnosis of azoospermia (OA/NOA) before sperm retrieval.

**Model:** Logistic regression with no interaction terms. Variable selection

**AUC:** 0.63 (0.59 – 0.67)

---

## Surveys of Predictive Models in IVF

1. A systematic review of the quality of clinical prediction models in in vitro fertilisation, *M.B.Ratna, S.Bhattacharya, B.Abdulrahim, and D.J.McLernon*
2. Making IVF more effective through the evolution of prediction models: is prognosis the missing piece of the puzzle? *Mara Simopoulou, Konstantinos Sfakianoudis, Nikolaos Antoniou, Evangelos Maziotis, Anna Rapani, Panagiotis Bakas, George Anifandis, Theodoros Kalampokas, Stamatis Bolaris, Agni Pantou, Konstantinos Pantos & Michael Koutsilieris*
3. Predicting the outcomes of assisted reproductive technology treatments: a systematic review and quality assessment of prediction models, *Ian Henderson M.Sc., Michael P. Rimmer M.Sc., Stephen D. Keay M.D., Paul Sutcliffe Ph.D., Khalid S.Khan M.Sc., Ephia Yasmin Ph.D., Bassel H. Al Wattar Ph.D.*