30 Seconds to Customize Your Exclusive AI Assistant! YiduCore + DeepSeek Empowers Doctors at Sun Yat-sen University Cancer Center to Become AI Product Managers

When AI meets the diverse needs of clinicians, how can intelligent assistants be both professional and user-friendly?

Yidu Tech and Sun Yat-sen University Cancer Center (SYSUCC) have jointly delivered an innovative solution—the launch of China's first "My AI Assistant" platform that allows doctors to customize their own AI assistants. This enables every doctor to assemble their personalized AI assistant like building with LEGO bricks. No longer do doctors need to adapt to AI; instead, AI learns and adapts to the doctors.

Previously, Yidu Tech completed the private deployment of its AI middleware platform at SYSUCC, leveraging the full version of the domestic large language model DeepSeek and Yidu Tech's "AI Medical Brain" YiduCore to create a diagnostic assistant with functions such as decision support, medical record generation, and intelligent dialogue. The launch of "My AI Assistant" marks another milestone in the exploration of AI in healthcare by Yidu Tech and SYSUCC.

Solving the "One-Size-Fits-All" Dilemma in Clinical AI

Traditional AI healthcare solutions often face challenges due to their standardized designs, which fail to meet the diverse needs of doctors.

Professor Sun Ying, Vice President of SYSUCC, noted in an interview with China News Service that unified models or applications sometimes cannot meet the depth of expertise required by specialists or address the varying needs of doctors with different seniority levels and roles. For example, some doctors need patient education assistants to generate easy-to-understand explanations, while others require more professional and traceable conclusions. Standardized outputs often result in complaints of being "too simplistic" or "not professional enough."

Based on over a decade of experience in AI healthcare, Yidu Tech has deeply understood doctors' needs and proposed a solution: giving doctors the power to design AI. The newly launched "My AI Assistant" at SYSUCC leverages large language model technology, allowing doctors to build their own AI assistants on a unified model foundation. Doctors can customize the assistant based on their clinical experience, select patient data, configure workflow logic, and create application-specific intelligent agents as needed.

"When 1,000 doctors have 1,000 different needs, only by making AI adapt to humans, rather than humans adapting to AI, can AI truly take root in clinical practice."

Three Steps to Assemble Your Personalized Assistant

1. Name + Define Requirements: 30 Seconds to Create Your Exclusive AI
   

2. Start by naming your assistant, then describe its mission, and finally issue strategic instructions (e.g., "Please analyze tumor size in each imaging report and summarize the trends.").
   Don’t know how to write instructions? No problem! The platform provides multiple prompt templates for one-click use and a "Smart Generation" assistant that automatically completes instruction details based on semantic understanding, making it as simple as installing an app on your phone.

3. 
   

4. Dual-Model Switching: Meet Your Diverse Needs

5. 
   

6. YiduCore Medical Vertical Model: Built on Yidu Tech’s 10 years of multi-dimensional medical knowledge graphs, it offers high accuracy and professionalism, suitable for knowledge Q&A, medical record understanding, and generation, with fast response times.

7. • DeepSeek R1 (Deep Thinking Mode): DeepSeek’s deep thinking model is ideal for reasoning and complex decision-making, requiring longer response times.

8. 
   

9. Select Data + Set Time Limits: Precision-Feed Your AI
   Doctors can freely select key data types such as imaging exams, medical records, prescriptions, and pathology, excluding redundant information. They can also set time ranges, such as "only associate data from the last three outpatient visits," allowing AI to focus on core diagnostic data.

A Decade of Foundation: The "Iron Triangle" Supporting Personalized AI Implementation

1. 
   

2. Data "Living Map": Since 2015, Yidu Tech has assisted SYSUCC in building a specialized oncology big data platform. In 2020, China's first T+0 real-time updated oncology big data platform was established, integrating data from over 30 core business systems and processing more than 3,000 high-quality structured fields. This created a "living map" of medical data covering the entire treatment journey of 2 million patients. AI can accurately access these "exclusive resources," and the high-quality data significantly enhances AI precision.
3. 
   
4. Large Language Model "Engine": Yidu Tech has deployed the DeepSeek-R1 671B full-version large language model at SYSUCC, capable of flexibly handling complex medical scenarios. The more specialized and precise YiduCore medical vertical model provides medical logic constraints, enabling breakthroughs in specialized accuracy.
5. 
   
6. Scenario "LEGO-ization": Yidu Tech’s AI middleware platform provides clinicians with no-code capabilities to build intelligent agents and applications, allowing every doctor to easily "harness" AI. "My AI Assistant" takes customization further by integrating into SYSUCC’s doctor workstations and linking patient data, truly achieving "personalization for every doctor."

7. 
   

Like building with LEGO, data is the base, models are the connectors, and doctors are the designers.

"Clinicians are the best product managers for AI in healthcare," emphasized Xu Jiming, CEO and Co-founder of Yidu Tech, during the collaboration announcement. "Every second they spend in clinical scenarios accumulates the most genuine insights into pain points; every decision they make for patients defines the most critical functional standards. What Yidu Tech aims to do is transform clinicians' wisdom into reusable AI productivity through technology."

Vice President Sun Ying stated, "With configurable intelligent agent technology, every doctor can create N intelligent assistants tailored to their needs. This is a key step toward deeply scenario-based AI healthcare and marks the transition from 'tool empowerment' to 'capability co-creation,' providing a reusable technical paradigm for precise and personalized oncology diagnosis and treatment."

Previously, the "YiduCore + DeepSeek" partnership has flourished across the healthcare industry, with successful implementations at top-tier hospitals such as Xiangya Hospital of Central South University, The First Affiliated Hospital of Nanchang University, and Harbin Medical University Cancer Hospital.

In the future, Yidu Tech will continue to assist SYSUCC in rapidly implementing over 120 large language model application scenarios, driving oncology diagnosis and treatment toward greater precision, efficiency, and intelligence. The company also looks forward to deepening collaborations with more industry leaders to create quantifiable and replicable intelligent applications across more healthcare scenarios, enabling more patients to benefit from technological advancements.