JMIR Publications

ABSTRACT

Background:

Reproductive healthcare faces challenges in predictive accuracy, timely intervention, and equitable resource distribution. General-purpose large language models (LLMs) further lack the medical expertise required for clinical applications. This creates a critical need for specialized AI solutions to enhance precision and accessibility in full-cycle fertility care.

Objective:

To address challenges in fertility health such as difficult prediction, delayed intervention, and uneven resources, as well as limitations of general LLMs (e.g., lack of medical expertise and hallucinations), this study aims to develop an LLM-based platform for full-cycle fertility health management.

Methods:

A knowledge base was built by integrating authoritative guidelines and clinical resources. The model was optimized using Retrieval-Augmented Generation (RAG) and deep reasoning, supporting a dual-track system for doctors (full diagnosis-treatment process) and patients (full health management cycle).

Results:

The platform successfully delivers three core functions: For clinicians, it provides individualized risk assessment, etiological analysis, and decision-making support in key areas like maternity care, cervical diseases, and reproductive immunity; For patients, it offers intelligent triage, pre-consultation, personalized education, and an AI Q&A service; It effectively empowers primary healthcare institutions by standardizing processes, enhancing their capabilities in risk assessment and standardized referral.

Conclusions:

This platform establishes a new intelligent service paradigm through clinical intelligence, patient-centered care, and primary care standardization, offering a solution for full-cycle health protection.