In Vitro Fertilization (IVF) Cumulative Pregnancy Rate Prediction from Basic Patient Characteristics

Bo Zhang, Yuqi Cui, Meng Wang, Jingjing Li, Lei Jin, Dongrui Wu

Abstract

Tens of millions of women suffer from infertility worldwide each year. In vitro fertilization (IVF) is the best choice for many such patients. However, IVF is expensive, time-consuming, and both physically and emotionally demanding. The first question that a patient usually asks before the IVF is how likely she will conceive, given her basic medical examination information. This paper proposes three approaches to predict the cumulative pregnancy rate after multiple oocyte pickup cycles. Experiments on 11,190 patients showed that first clustering the patients into different groups and then building a support vector machine model for each group can achieve the best overall performance. Our model could be a quick and economic approach for reliably estimating the cumulative pregnancy rate for a patient, given only her basic medical examination information, well before starting the actual IVF procedure. The predictions can help the patient make optimal decisions on whether to use her own oocyte or donor oocyte, how many oocyte pickup cycles she may need, whether to use embryo frozen, etc. They will also reduce the patient's cost and time to pregnancy, and improve her quality of life.

Introduction

In Vitro Fertilization (IVF) is often a final hope for couples struggling with infertility, yet it presents significant financial and emotional hurdles. A major challenge is the "catch-22" of current predictive models: they often rely on data collected during the procedure, such as embryo morphology. By the time this information is available, patients have already invested months of time and thousands of dollars. This study addresses this gap by predicting the cumulative pregnancy rate using only baseline "Day 3" characteristics available at the first visit.

Comparison of the proposed rapid, low-cost approach versus conventional IVF prediction timelines and costs.

Methodology

The research compared three machine learning pipelines to handle the high variability in patient data:

  1. Clustering: Using k-means to group patients with similar profiles and assigning the mean success rate of the cluster.
  2. Support Vector Machine (SVM): A standard supervised learning approach using a Radial Basis Function (RBF) kernel.
  3. C-SVM (Hybrid): A novel sequential approach that first clusters patients to group similar biological profiles and then trains a specific SVM for each individual cluster.

The models were trained on a dataset of 11,190 patients from Tongji Hospital in Wuhan, China, using 11 key features identified through logistic regression, including female age, BMI, infertility duration, and biomarkers like Antral Follicle Count (AFC) and Anti-Mullerian Hormone (AMH).

Results and Discussion

The C-SVM model outperformed both standalone clustering and standard SVMs. By creating "micro-models" for specific patient types, such as older patients with low ovarian reserve versus younger patients with PCOS, the C-SVM approach achieved higher stability and lower error rates.

ROC curves demonstrating the superior performance of the C-SVM model compared to standalone clustering and SVM.

The C-SVM achieved an Area Under the Curve (AUC) of 0.70 for the first cycle. While an AUC of 0.70 is considered "fair" in general machine learning, it is robust for biological predictions based on limited baseline data. Furthermore, the C-SVM showed a significantly lower Root Mean Squared Error (RMSE of 0.0267) compared to the standard SVM (0.0794), indicating its predictions were much closer to actual observed pregnancy rates.

Conclusion

This study demonstrates that baseline "Day 3" medical data is sufficient to provide meaningful pre-treatment predictions. The proposed C-SVM model offers a rapid (1-3 days) and economic (~$50) alternative to conventional methods that cost thousands of dollars and take months. While the model's current scope is limited to a Chinese population, it provides a foundational framework for personalized IVF success estimation that can significantly improve a patient's quality of life and decision-making process.

Summary updated

Abstract

Tens of millions of women suffer from infertility worldwide each year. In vitro fertilization (IVF) is the best choice for many such patients. However, IVF is expensive, time-consuming, and both physically and emotionally demanding. The first question that a patient usually asks before the IVF is how likely she will conceive, given her basic medical examination information. This paper proposes three approaches to predict the cumulative pregnancy rate after multiple oocyte pickup cycles. Experiments on 11,190 patients showed that first clustering the patients into different groups and then building a support vector machine model for each group can achieve the best overall performance. Our model could be a quick and economic approach for reliably estimating the cumulative pregnancy rate for a patient, given only her basic medical examination information, well before starting the actual IVF procedure. The predictions can help the patient make optimal decisions on whether to use her own oocyte or donor oocyte, how many oocyte pickup cycles she may need, whether to use embryo frozen, etc. They will also reduce the patient's cost and time to pregnancy, and improve her quality of life.
BibTeX
@inproceedings{zhang2019vitro,
  author = {Bo Zhang and Yuqi Cui and Meng Wang and Jingjing Li and Lei Jin and Dongrui Wu},
  title = {{In Vitro Fertilization (IVF) Cumulative Pregnancy Rate Prediction from Basic Patient Characteristics}},
  booktitle = {IEEE Access},
  year = {2019},
  url = {https://openreview.net/forum?id=aql2FeBRJ31},
}
Figure