The first patient intelligent matching patent technology is landed domestically, and Wing Sail Medicine reconstructs the logic of recruiting clinical drug trials.

Recently, Beijing Yifan Fengshun Medical Technology Co., Ltd., a subsidiary of Water Drop Company, independently developed the "Intelligent Matching Technology for Clinical Drug Trial Patients" has officially obtained a national invention patent (patent number: ZL202210080502.4): matching methods, devices, and computer equipment for clinical drug trial patients. Water Drop Company had previously independently developed the Water Drop Shuoshou Big Model, driving business efficiency comprehensively. This new patented technology integrates innovative deep neural networks and natural language processing technology, completely solving the industry pain points of "high workload, low efficiency, and prone to errors" in traditional clinical drug trial patient recruitment, pressing the "acceleration button" for pharmaceutical innovation research and development. Water Drop Company is one of the earliest companies in the industry to deploy AI big model applications, investing nearly 300 million yuan in research and development annually. Currently, it has accumulated more than 100 artificial intelligence technology patents. The Water Drop Shuo Shou Big Model independently developed by Water Drop Company has become the "central nervous system" driving business efficiency, widely used in product innovation, user services, and many other business scenarios, significantly improving the efficiency and quality of the entire process of insurance and health services. Clinical drug trials are a key step in the launch of new drugs, and accurately recruiting patients who meet the inclusion criteria is the core prerequisite for the smooth progress of the trials. In traditional recruitment models, the inclusion criteria for drug trials are often described in unstructured natural language, and patient information is scattered in various medical record data. Staff need to manually compare and analyze them, which is not only time-consuming and laborious but also prone to inaccuracies due to information deviation and human errors, seriously affecting the progress of the trials. According to industry-related data, over 80% of clinical trials are obstructed due to recruitment delays. The intelligent matching technology launched this time innovatively adopts a dual-track parallel scheme of "numerical matching + semantic embedding matching", achieving end-to-end precise matching between patients and trial projects. In the technical process, the system first breaks down patient information into first numerical information (such as age, laboratory data, etc.) and medical information (such as medical history, treatment records, etc.), while splitting the inclusion criteria for drug trials into second numerical information and descriptive information; through a three-element extraction algorithm, the numerical information is transformed into "numerical - unit - meaning" triplets, achieving precise numerical comparison and quickly screening core matching items; for unstructured medical information and descriptive information, feature extraction and normalization are carried out using technologies such as one-hot encoding and BERT language model, completing semantic deep matching in a unified feature space, and ultimately outputting accurate matching conclusions based on the dual-track matching results. Compared to traditional manual matching modes, this technology has three core advantages: first, efficiency is increased several times, shortening the matching work that originally took several days to minutes, and overall recruitment efficiency is increased by over 300%; second, higher accuracy is achieved, with a matching accuracy of over 92% through global optimized end-to-end training, avoiding error accumulation in traditional segmentation, relationship extraction, etc.; third, strong adaptability, compatible with different formats of medical record data and diverse descriptions of trial inclusion criteria, covering the clinical trial requirements of various treatment areas such as solid tumors, hematologic diseases, cardiovascular diseases, autoimmune diseases, etc. The research and development leader of Beijing Yifan Fengshun Medical Technology Co., Ltd. stated that the practical application of this technology not only saves a significant amount of recruitment time and manpower costs for pharmaceutical companies but also allows eligible patients to quickly connect with high-quality clinical trial resources, bringing new treatment hopes to patients with difficult diseases. In the future, the Yifan Medical R&D team will continuously optimize technical models, expand functions such as multi-language adaptation and special matching for rare diseases, providing more efficient technical support for global pharmaceutical innovation. Currently, this technology has been pilot-tested in clinical trial projects in multiple top-tier hospitals and pharmaceutical companies, completing nearly a thousand patient matches, helping innovative drug clinical trial projects achieve recruitment goals 3-6 months ahead of schedule. It is understood that Yifan Medical is a digital clinical patient recruitment platform under Water Drop Company, relying on Water Drop Company's large patient group and digital AI capabilities to carry out patient recruitment and other businesses, focusing on solving the industry pain points of "slow recruitment and difficult matching" in clinical trials, and assisting well-known pharmaceutical companies at home and abroad to improve research and development efficiency.