Medical Imaging Series Report No.13: PaiYiPai – Starting from Lab Report Interpretation to Build a Big Data Platform in Healthcare
Open the mobile app, take a photo of your lab test report, and upload it; within three minutes, you will receive a comprehensive and detailed interpretation report. This app, aptly named “Pai Yi Pai,” was officially launched in mid-June this year.
In addition to providing rapid and detailed interpretations of lab reports uploaded by users, Pai Yi Pai can also convert paper-based medical records into electronic health records (EHRs) and establish cloud-based personal health profiles. By leveraging cloud storage for patient data, Pai Yi Pai aims to track patients’ medical data throughout their lifetimes, thereby delivering the most personalized health guidance.
Why Start with Laboratory Test Reports?
In 2014, the total number of outpatient visits in China’s healthcare institutions reached 7.8 billion, with an average of 5.8 visits per person. Each visit may generate multiple laboratory test reports; conservatively, the total number of such reports exceeded 15 billion. According to statistics from the National Health and Family Planning Commission, there were more than 2.3 million physicians nationwide, each interpreting dozens of laboratory test reports on average per day. In reality, however, Chinese physicians spend only 3–5 minutes per patient on average, meaning that patients inevitably cannot understand the implications of abnormal indicators on their laboratory test reports within such a short timeframe.
Moreover, patients generate a large volume of medical test reports, making management complex. The loss of partial laboratory data can delay optimal timing for diagnosis and treatment. Furthermore, due to patients’ limited understanding of the extensive laboratory data, it is difficult for them to effectively integrate their health information on their own, thereby hindering their ability to monitor changes in their health status. These factors may impede physicians from gaining a comprehensive understanding of the patient’s condition and making timely and effective clinical judgments.
Pai Yi Pai addresses these pain points by enabling patients to interpret their laboratory test results through photo-based inquiries, delivering timely explanations to help them understand their health status conveniently and efficiently. In addition to storing test reports, the platform provides second opinions from authoritative medical experts in Beijing. With cloud storage of laboratory reports, patients’ medical information is continuously accumulated, consolidating medical records from different institutions into a unified personal cloud, thereby laying the foundation for healthcare providers to deliver enhanced services.
Wu Shizhan, CEO of Pai Yi Pai, stated that from an entrepreneur’s perspective, hospital data remains siloed, and patients cannot directly access their electronic records. Therefore, photographing, recognizing, and interpreting laboratory test reports serves as the optimal entry point for helping patients establish personal electronic health records.
Machine Intelligence Recognition Technology
PaiYiPai is the first mobile health app in the industry to offer image recognition and interpretation of laboratory test reports. Prior to its launch, several teams had attempted to develop similar applications for interpreting lab results, but most failed due to unresolved technical challenges in image recognition. As the CEO of PaiYiPai stated, “Our team has a strong technical background, including world-leading experts in computer vision. We automatically recognize laboratory test reports through our system, extract data from them, and provide interpretations via an expert system, thereby improving physicians’ efficiency in reviewing test results.”
Paiyi Pai has currently completed OCR (Optical Character Recognition) modeling for laboratory test reports, is undergoing data training for the model, and has filed multiple patent applications. The patents filed by the team include: layout analysis technology for medical documents based on machine learning (capable of recognizing laboratory test reports with different layouts); and a core image text recognition engine for multi-character sets in the medical field (capable of recognizing different names for the same term).
PaiYiPai CEO Wu Shizhan further introduced, “We hold a national invention patent that allows users to capture laboratory test reports without manual photography. By simply pointing the smartphone camera at the report area, the system automatically initiates scanning. If a test report is detected, the app provides real-time guidance—such as moving closer, correcting excessive tilt, or holding steady—and subsequently captures the image, automatically cropping and rectifying the document.”
Furthermore, Pai Yi Pai possesses robust machine learning capabilities; the more data it processes through machine recognition, the “smarter” it becomes.
In other words, during the recognition process, PaiYiPai selects content it cannot identify or is uncertain about and transfers it to human reviewers for identification. After manual review, the results are fed back into the system to provide training data for the machine. Through iterative learning, PaiYiPai’s accuracy in recognizing laboratory test parameters continues to improve. “This process is akin to how a child learns about the world. When a child does not recognize an object, their mother tells them what it is; after repeated exposure, the child remembers it. This is essentially a learning process,” emphasized Wu Shizhan. “As data accumulates, the machine’s recognition accuracy will ultimately approach that of the human eye.” Once the machine accurately identifies the laboratory test parameters, it can make corresponding judgments.
In the future, Pai Yi Pai will also provide users with precise medical consultation information services. Meanwhile, under the premise of protecting user privacy, it will offer data mining services to companies in the insurance, pharmaceuticals, medical equipment, and consumables sectors through accurate information analysis, ultimately realizing the vision of serving users with big data and building a patient-centered healthcare ecosystem.
CEO Wu Shizhan stated, “We will first perfect the recognition of laboratory test reports. Subsequently, we will expand to cover all types of paper-based medical documents, as well as medical imaging. With sufficient data, whether it be laboratory test reports, prescription slips, or medical images, Pai Yi Pai is capable of learning.”
Pai Yi Pai will also introduce an interactive, AI-powered Q&A feature. Upon reviewing medical documents, the system identifies indicators corresponding to various stages and outcomes in a patient’s condition. It then asks users about their current stage of diagnosis or treatment—such as being three months pregnant or currently taking specific medications—and provides personalized recommendations.
“I hope to see more and more companies join us in expanding this industry, enabling patients to better understand themselves and become their own guardians.” Wu Shizhan further stated that Pai Yi Pai is considering opening up its optical character recognition (OCR) technology for paper documents to collaborate with mobile health companies in need, working together to grow the industry.
The Pai Yi Pai team continues to expand, having completed its angel round of financing and is currently in discussions for a new round of funding.
Team Background
PaiYiPai’s team is a technology-driven force, comprising numerous experts in computer vision and machine learning, including core OCR specialists from companies such as Baidu, Tsinghua Unigroup, and Hanwang.
Wu Shizhan, Founder and CEO, is a MySQL database expert. He previously served as Chief DBA and Director of the Technical Assurance Department at Ganji.com, and led Baidu’s commercial database team, overseeing data systems for all revenue-generating products, including paid search rankings, Fengchao (Baidu’s PPC advertising platform), and the Alliance Network. He possesses extensive experience in internet development, operations, and maintenance, as well as big data analytics and mining. He has conducted in-depth research and accumulated substantial practical expertise in big data operations within high-concurrency, high-availability environments.
Co-founder Liu Li is an expert in computer vision and machine learning. After earning his Ph.D. from the Illinois Institute of Technology, he conducted research on digital imaging technologies at Panavision and Omnivision. He returned to China in 2009 and continued to lead the R&D of machine vision and imaging products at companies such as Panavision. He has filed more than 20 patents covering areas such as facial recognition, motion image processing, and drone piloting, and boasts over a decade of extensive experience in the fields of image processing, machine vision, and machine learning.
Another co-founder, Yang Jinsong, previously served as the Medical Director at Haodf.com, where he managed a triage team of 60 members. Prior to this role, Dr. Yang practiced medicine for 15 years and worked as an attending physician in the Department of Critical Care Medicine at Peking University People’s Hospital. During his clinical career, he participated in multiple medical data development initiatives at Peking University People’s Hospital and possesses extensive experience in the deep mining of big medical data.
VCBeat’s Series on Medical Imaging Startups
Medical Imaging Series Report 1: Yilian Builds an Instagram + Quora Exclusively for Doctors
Medical Imaging Series Report 5: Weiyun Imaging Becomes the Fastest Medical Image Viewing Platform
Medical Imaging Series Report 6: Aipeike: The U.S. Medical Imaging Expert Platform Available 24/7
Medical Imaging Series Report 7: Figure.1, An Instagram Exclusively for Doctors
Medical Imaging Series Report 8: 51 Read Film Network: Why Are We Establishing a Medical Imaging Expert Group?
Medical Imaging Series Report 9: Haina Yixin: How Did This Medical Imaging Company Sell Its Products to the United States?
Medical Imaging Series Report 10: M&S: The Core of Remote Medical Imaging Platforms Lies in Providing Diagnostic and Treatment Services
Medical Imaging Series Report 11: Yingda Technology: Deepening Expertise in Mobile Imaging Technology
Medical Imaging Series Report 12: Huaxia Pathology Network: How It Retains Over 90% of Pathologists in China