Southwest Securities: It is recommended to focus on iFLYTEK (02506), which has a large model with a self-developed base

The Zhitong Finance App learned that Southwest Securities released a research report saying that China's medical artificial intelligence industry has a large space, and policies drive rapid development of the industry; iFLYTEK (02506) has a self-developed base model, and the upper performance limit continues to rise. Spark Healthcare's Big Model X1 can significantly reduce illusions and improve the accuracy, professionalism and interpretability of reasoning; it constructs a diversified GBC customer product matrix based on the big model and big data technology platform. First coverage, recommended.

The main views of Southwest Securities are as follows:

China's medical artificial intelligence industry has a lot of space, and policies drive the rapid development of the industry

The application of artificial intelligence in medical scenarios is expected to revolutionize changes. According to Frost & Sullivan data, China's medical artificial intelligence market grew significantly, from 2.7 billion yuan in 2019 to 8.8 billion yuan in 2023, with a compound annual growth rate of 33.8% during the period. The market size is expected to grow further to RMB 315.7 billion by 2033, and the compound annual growth rate from 2023 to 2033 is expected to reach 43.1%. Furthermore, the policy guides investment in the field of medical artificial intelligence. In 2024, the National Health Commission, in conjunction with the State Administration of Traditional Chinese Medicine and the State Administration for Disease Control and Prevention, issued the “Reference Guidelines for Artificial Intelligence Application Scenarios in the Health Industry” to strengthen support for the AI medical industry and promote its deep integration with the medical industry.

The company has a large self-developed base model, and the upper performance limit continues to rise

In March 2025, the company officially released the “iFLY Spark Medical Model X1” based on deep reasoning technology. This model is currently the only large-scale medical in-depth inference model that has been completely completed using national production computing power training. In the real scenario test, the reasonable rate of general adjuvant diagnosis reached 94.0%. In July 2025, the iFLY Spark Medical Big Model completed a capacity upgrade, and many performance indicators were further improved. Specifically, the reasonable rate of adjuvant diagnosis in general medicine was raised to 95.0%, the reasonable rate of interpretation of medical examination reports was raised to 86.3%, the response rate for health consultation was raised to 91.5%, and the reasonable rate of hospital-level cardiovascular medicine diagnoses increased to 91.2%. Testing shows that in key medical scenarios, its performance continues to be far ahead of mainstream international models such as OpenAI O3 and DeepSeek R1.

Build a diversified GBC customer product matrix based on big models and big data technology platforms

By the end of 2024, the company had provided products and services to more than 70,000 primary medical institutions, covering more than 30 provinces, 670 districts and counties, and more than 500 grade hospitals across the country, including more than 40 of China's top 100 hospitals and 7 of the top ten hospitals. 1) To G primary care service business 25H1 revenue growth rate of 52.3%; regional management platform solution 25H1 revenue growth rate of 178.1%: primary care services+regional management platform solutions are basic. The primary care service business line is mainly composed of intelligent medical assistants and chronic disease management. Regional management platform solutions are mainly composed of smart health and smart health insurance; 2) To B business 25H1 revenue fell 10%: mainly due to the continuous update of the big model. Businesses include smart hospitals+AI diagnosis and treatment assistants to improve medical service capabilities and efficiency; 3) To C business 25H1 revenue growth rate of 10%: including imaging cloud platform+post-diagnosis management+iFLY+ intelligent hardware. The AI post-diagnosis patient management product developed by the company is based on the Spark Medical model upgrade. It can understand patients' electronic medical records, automatically organize key diagnosis and treatment information, generate personalized rehabilitation plans based on evidence-based disease management knowledge, and naturally interact with patients through various technologies to help doctors carry out disease science popularization, rehabilitation guidance and condition assessment.

Profit forecast

The estimated revenue for 2025-2027 is 850 million yuan, 1.01 billion yuan, and 1.21 billion yuan, respectively, and the corresponding PS is 10.7 times, 9 times, and 7.5 times, respectively. The AI medical market space is large. The company has a large self-developed base model, and the upper performance limit continues to rise.

Risk warning: Risk of market expansion falling short of expectations, risk of delayed contract execution, risk of increased market competition.