(Hong Kong, Shanghai, 9 March 2021) Ping An Insurance (Group) Company of China, Ltd. (hereafter “Ping An” or the “Group”, HKEX: 2318; SSE: 601318) is pleased to announce that promising research on the artificial intelligence (AI) technology of Ping An AskBob Doctor has been published in The University of Oxford’s Journal of the National Cancer Institute (JNCI), an important international journal of oncology. The research shows the AI system can effectively categorize nasopharyngeal cancer (NPC) patients by risk level to recommend more refined treatment solutions.
Ping An Healthcare Technology, Ping An Smart City, Ping An Good Doctor and Sun Yat-sen University Cancer Center (SYSUCC) in Guangzhou jointly released the research paper, “A Prognostic Predictive System Based on Deep Learning for Locoregionally Advanced Nasopharyngeal Carcinoma”. NPC affects the nasopharynx, the upper part of the tube that starts behind the nose and runs down to the windpipe. NPC has high prevalence rates in South China, Southeastern Asia and North Africa.
The research team used magnetic resonance imaging (MRI) images and clinical data of over 3,000 NPC patients from four hospitals, including SYSUCC. The data was used to train the AI model to estimate the risk of the condition of NPC patients deteriorating in next five years and assign an overall score. Doctors input the original MRI images and clinical factors, then the system generated the prognoses for NPC patients automatically. The system displayed a concordance index of 0.78, which presented a 23% improvement on the conventional Tumor, Node, Metastasis (TNM) Staging System, the globally accepted standard to describe the extent of a primary tumor and the spread of cancer in a patient.
The incorporation of Deep Neural Networks and an ensemble learning model significantly improved the precision of AskBob Doctor’s prognoses. The overall scores were used to distinguish low-risk patients from high-risk patients. The system also analyzed the treatment standards and recommended more precise treatment solutions for patients in various risk groups, improving the treatment outcomes of high-risk patients. While high-risk patients are shown to benefit from a combination of concurrent chemoradiotherapy (CCRT) and induction chemotherapy (IC), IC does not benefit low-risk patients significantly more than CCRT alone. Differentiated treatments can shorten waiting times for radiotherapy for those who need it most, remove the financial burden of IC treatment for patients in low-risk groups.
Guotong Xie, Chief Healthcare Scientist of Ping An Group and co-corresponding author of the paper, said, “NPC is a kind of invasive epithelial malignancy. There is a high probability of relapse, metastasis or death for patients with advanced NPC. If we can accurately predict the risk of disease deterioration at an early stage, and develop a personalized treatment plan for patients according to different risks, then we can improve the treatment effect of high-risk patients and reduce the economic burden of low-risk patients.”
The AI prognostic predictive system is one of the key functions of AskBob Doctor’s system. It has been widely adopted in various application scenarios of Ping An Smart Healthcare and Ping An Good Doctor, China’s leading online healthcare services platform. As of December 2020, AskBob Doctor had cumulatively served 20,000 medical institutions and 650,000 doctors to support the entire process of disease management.