Establishment of a Fertility Management Service Platform Based on Infertility Big Data

Overview

In response to Korea’s declining fertility rate, the project utilized AI-driven analysis of infertility treatment data to develop an innovative digital fertility management service.

The service seeks to develop an innovative digital platform that boost fertility through:

1. Predicting pregnancy likelihood and providing solutions based on infertility data
2. Building an innovative platform to enhance fertility, by providing self-management services based on lifestyle and health management factors.

Project description

In 2024, South Korea entered the ranks of countries with an ultra-low birth rate. As of the fourth quarter of 2023, the total fertility rate reached a record low of 0.65. Despite the government's various birth support policies, the birth rate continues to decline. Meanwhile, the number of people diagnosed with infertility and the number of procedures performed are rapidly increasing. However, due to a lack of awareness and support for fertility management to achieve pregnancy, effective responses are not being implemented. Therefore, this project aims to develop innovative services, such as pregnancy probability prediction and fertility management services, by introducing big data and AI technologies.

• Data Collection and Processing

  • Collected and refined infertility treatment data accumulated from 2018 to 2023 at eight medical institutions in the Yeongnam region and established a big data-based analysis environment. The main data types are infertility treatments history, patient health status, and treatment response data. This data was standardized and integrated into a data lake.

• Dataset Composition

  • After review by the Busan Pseudonymised Data Utilization Center, pseudonymization was performed, and data labeling for machine learning was conducted. Subsequently, to enhance the accuracy of pregnancy probability prediction, external health data, genetic information, and lifestyle habit data were linked and combined to enable multidimensional analysis.

• AI Model Training Design

  • The constructed dataset was designed to enable AI model training, comprising refined infertility treatment data, health information, and environmental factor data. This established the foundation for developing a more reliable pregnancy probability prediction model.

Key technological features

• Machine Learning and Data Mining-Based Predictive Model: Predicts pregnancy likelihood by training AI models using infertility treatment and lifestyle data, analyzing infertility factors, and providing optimal solutions.
• Core Algorithms: Random Forest (RF), XGBoost
• Utilizing Data Lake: Model performance optimized using SparklingSoDA

Expected outcomes

By digitizing previously handwritten infertility treatment data, approximately 35,000 records from medical institutions were selected, refined, accumulated, and shared. Based on this dataset, the service provides daily care support for infertile individuals and predicts fertility success rates.

By managing information such as infertility treatment history and women's occupational data, the service established the foundation for providing personalized health management solutions that enhance fertility outcomes for individuals facing infertility.

Furthermore, it contributed to raising public awareness about infertility prevention and reproductive health through a self-diagnosis feature available to the public.

Evaluation of the AI algorithm model for predicting pregnancy success rates achieved an accuracy exceeding 90%. Through continuous outreach and dissemination, it contributed to improving public awareness regarding infertility prevention and health management.

Moving forward, the project will continue to collect and refine additional infertility and fertility- related data to further enhance the performance of its AI learning models. Continuous improvement of the algorithms will enable more accurate and effective data analysis and utilization.

To scale the service beyond the current Daegu and Gyeongbuk regions, the project plans to establish a nationwide collaboration network with public health centers and hospitals, and to develop a user- friendly platform accessible via both mobile and web interfaces.

Launch year: 2023

Attribution

Case study courtesy of the National Information Society Agency, Republic of Korea, Apolitical's Content Partner. Sourced from Use Cases of Public AI Service, Vol.1,  https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=28997&parentSeq=29001