DingTalk Rebuilds Its AI Ecosystem: Platform Launches Gynecology-Specific Large Model with Over 90% Accuracy

Recently, the "Doukou Gynecology Large Model," developed by Yisheng Jiankang (Hangzhou) Life Science Co., Ltd., achieved an accuracy of 90.2% in professional testing, becoming the first high-accuracy, highly practical domain-specific large model successfully trained on the DingTalk AI platform. The application of this model will effectively alleviate the shortage of specialized gynecologists and help more female users achieve science-based health management.

DingTalk Ecosystem Expands to AI Entrepreneurs

The launch of the gynecology large model marks that DingTalk's ecosystem has extended from SaaS, service providers, and consulting/delivery ecosystems to a new innovation ecosystem supporting AI entrepreneurs. As general-purpose large models become more widespread, various industries are actively promoting deep integration between large models and business scenarios. To address this, DingTalk has built a support system for industry/corporate large model development, covering the entire process from data annotation and training/inference to deployment and application, helping enterprises efficiently upgrade their productivity through AI.

Accuracy Rises to 90.2%, Demonstrating Clinical Utility

On the evening of June 30, at the Hangzhou LvFeng Center, the Yisheng Jiankang team announced that its Doukou Gynecology Large Model—trained on DingTalk's enterprise-specific AI platform—had completed 100 professional test questions. For six common gynecological symptoms—menstrual irregularities, abnormal bleeding, abnormal vaginal discharge, lower abdominal pain, and lower abdominal masses—the model achieved a diagnostic accuracy of 90.2%, demonstrating significant potential as a clinical decision-support tool.

"AI Gynecologist" Provides Instant Health Services

Wang Qiangyu, founder of Yisheng Jiankang, said that the Doukou Gynecology Large Model functions like an "AI gynecologist." Users simply select their symptoms in the "Bestie Doctor" app, and the system provides professional feedback, including primary diagnoses, potential diagnoses, recommended tests, treatment plans, and precautions. Compared with the average 30-minute wait time for traditional online consultations, the system can generate recommendations in just seconds—making it especially suitable for working women and users in remote areas who need quick guidance on whether to seek medical care.

The Yisheng Jiankang team combines expertise in the internet, obstetrics and gynecology, and biopharmaceuticals, aiming to use AI to address the shortage of gynecological medical resources and create dual social and industry value for medical aesthetic institutions and women at large.

From 77.1% to 90.2%: DingTalk Powers Technological Breakthrough

In the initial phase, the team trained a first-version model based on an open-source model, achieving an accuracy of 77.1%. While this met basic industry standards, it fell short of meeting medical safety requirements. Faced with performance bottlenecks, Yisheng Jiankang turned to DingTalk's enterprise-specific AI platform, receiving comprehensive support in data processing, computing power scheduling, and model optimization. Within just one month, the model's accuracy improved to 90.2%.

In the future, Yisheng Jiankang will continue to optimize the model's performance and integrate it into the "Bestie Doctor" product, enabling the AI doctor agent to serve more users. Wang Qiangyu envisions, "In the future, this model can be replicated in other specialties, such as dermatology, allowing ordinary people to receive preliminary health guidance at home that is comparable to what they would get from a professional medical institution."

DingTalk Builds Full-Stack Capabilities for Industry-Specific Large Models

Zhu Hong, CTO of DingTalk, noted that increasing the model's accuracy from 77.1% to 90.2% is akin to rapidly transforming a "generalist" into an "expert specialist." This process involves secure data preprocessing, efficient computing power scheduling, the establishment of evaluation mechanisms, and fine-tuned parameter optimization—demonstrating DingTalk's full-stack technical capabilities in co-creating industry-specific large models with partners.

The Doukou Gynecology Large Model is the first vertical-domain-specific large model empowered by DingTalk for industry development. Currently, while general-purpose models such as Qwen, DeepSeek, and GPT have become foundational infrastructure, various industries still face challenges in deploying AI due to differences in knowledge systems, scenarios, and processes.

DingTalk has found that companies often struggle with "not knowing where to start" or "uncertain outcomes" during the stages of defining needs and preparing data, and cloud-based model training poses risks related to data circulation. To address these issues, DingTalk has built a full-process support system covering data collection, cleaning, annotation, model training, evaluation, optimization, and engineering deployment. It also provides AI solution consulting and talent training services to ensure successful AI implementation for enterprises.

This initiative is driving a reconfiguration of the DingTalk ecosystem, attracting developers with domain-specific data to co-create vertical large models and AI agents, and enabling small and medium-sized enterprises to upgrade to intelligent operations through the DingTalk app marketplace. In the future, developers will be able to complete the entire closed loop of AI product development, deployment, and commercialization on the DingTalk platform, working together to drive AI transformation across thousands of industries.

DomTech is DingTalk's officially designated service provider in Macau, specializing in providing DingTalk services to a wide range of customers. If you'd like to learn more about DingTalk platform applications, feel free to contact our online customer service or reach us by phone at +852 95970612 or by email at cs@dingtalk-macau.com. We have an excellent development and operations team with extensive market service experience, ready to provide you with professional DingTalk solutions and services!