Life Singularity CEO Liu Liyu: Unlocking True Value from Healthcare Data through Advanced Analytics

“To achieve true big data analytics, a specialized foundational platform for secondary data analysis is required—namely, a medical big data platform. Its distinguishing features include a positioning and technical architecture oriented toward data analytics, a comprehensive data governance framework with data quality management at its core, and the capacity to generate in-depth data analytics applications that truly ‘mine gold’ from data.” On the evening of July 12, Liu Liyu, Founder and CEO of Singularity Life, made these remarks at a seminar on medical data and applications themed “Empowering Data-Driven Smart Healthcare,” hosted by Singularity Life.

This event is part of the China Hospital Information Network Conference (CHIMA). As one of the most influential industry conferences in the field of healthcare informatization in China, this year’s CHIMA conference is themed “Implementing the Healthy China Strategy.” It features dozens of specialized academic reports and discussion forums, promoting disciplinary development and institutional advancement in hospital informatization. Thousands of representatives from hospitals across China have gathered at this prestigious event to further improve the quality of medical services and enhance levels of teaching, scientific research, and practical application.

Liu Liyu, Founder of Life Singularity

As a leader in China's medical big data sector, Life Singularity made its appearance at this prestigious event.

In his presentation titled “Governance and Application of Medical Data,” Liu Liyu, Founder of Life Singularity, stated that the journey of hospital informatization has evolved from the comprehensive digitalization of various business systems (“comprehensiveness”), to information integration and data consolidation (“connectivity”), and now to genuine data analysis and utilization (“intelligence”). He outlined a vision for leveraging system-based data governance and data application as key drivers to effectively implement the governance and application of medical big data. Additionally, he shared Life Singularity’s data governance architecture, core technologies such as standardization and structuring, as well as the company’s achievements in data governance and application.

Liu Liyu stated that traditional Clinical Data Repositories (CDRs) are unable to support in-depth data analysis and applications, lacking systematic data governance and technical architectures suitable for data analytics. To effectively translate current data into practical applications, data quality issues must be resolved; however, there remains a significant gap between our current capabilities and the high-quality clinical data characterized by breadth, longitudinal depth, granularity, accuracy, precision, and completeness. Effective data governance requires the collaborative efforts of a highly interdisciplinary team. The presentation resonated strongly with attending experts and received high acclaim, with one expert remarking, “This industry needs companies and a mindset dedicated to the diligent work of data governance and improving data quality. Otherwise, AI will remain nothing but castles in the air.”

Here are the key highlights from the speech by Liu Liyu, Founder and CEO of Life Singularity:

China's healthcare informatization has undergone a comprehensive process of digitizing business systems and is currently in the midst of a surge to build "interconnectivity." The subsequent focus will be on enhancing the quality of decision support based on data and realizing intelligent applications.

However, can traditional Clinical Data Repositories (CDRs) effectively support intelligent innovative applications based on big data? Our practice in recent years has revealed that the positioning and technical architecture of traditional CDRs are oriented toward business data sharing, merely achieving data aggregation. They still fall short of enabling in-depth data analysis, which explains why many CDR implementations have failed to deliver sufficient perceived value to business departments.

To achieve true big data analytics, a specialized foundational platform for secondary data analysis is required—namely, the Medical Big Data Platform, which we may refer to as CDR 2.0. Its key features include: an orientation and technical architecture designed for data analytics; a comprehensive data governance framework centered on data quality management; and the capability to generate in-depth data analytics applications, thereby truly “mining gold” from data.

In 2016, Life Singularity launched VitArk, a big data platform capable of integrating clinical data, biospecimens, and genomics data simultaneously. It established a productized medical data governance system, exploring and accumulating expertise in data integration, standardization, structuring, and quality improvement.

Data governance is a buzzword that has sparked intense discussion in recent years. But what exactly does it encompass? The framework diagram summarized by HIMSS in 2012 offers valuable reference points. Furthermore, the core objective of our data governance efforts is to ensure data quality. So, what constitutes high-quality data? From which dimensions should we examine and address data-related issues?

We have summarized several dimensions and attempted to encapsulate each with a single word: length, width, granularity, accuracy, standardization, and completeness.

The entire data governance framework is divided into three tiers: the foundational standards and specifications layer; the data governance functional layer; and the data application layer. The core activities of data governance include data integration, data standardization, data structuring, as well as quality control and improvement of business data.

From the perspective of data integration, we have explored and implemented an automated pipeline capable of rapid integration. This solution enabled a large tertiary Grade A hospital to consolidate all historical data in less than one month, while incorporating a quality control rule library that updates daily with new rules. In terms of data standardization, we have developed Life Singularity’s proprietary common data model, terminology system, and unified query and computation framework.

In terms of data structuring, we have built a corpus of millions of entries through three years of accumulation, continuously refining annotation and models. The F1 score for named entity recognition has reached 97.9%, achieving an industry-leading level. Based on the outcomes of data governance, we have explored numerous data applications in partner hospitals.

In the field of clinical research, we and our partner hospitals have identified several candidate biomarkers. A landmark academic paper is set to be published shortly, with us as direct participants in the study—a key distinction from other big data companies that are involved only in the preliminary data organization phase.

We have also achieved exploratory results in clinical decision support and operational management. Leveraging the foundation of data governance, we can unlock substantial business value through big data applications. Of course, data application is not the endpoint of data governance; a robust data governance framework must be a closed-loop system characterized by continuous iteration. During the process of data analysis and application, numerous issues with existing data are inevitably uncovered. Through data applications, we can identify data quality problems originating from business systems, and then implement targeted data quality control measures by integrating intelligent technologies with management strategies.

Such a comprehensive data governance framework has already been implemented in many hospitals across China. We will continue to explore, constantly accumulate experience and lessons learned in data governance, pioneer the path for medical big data governance in China, and contribute our expertise to the industry. Only with solid and in-depth data governance can the applications of medical big data and artificial intelligence be truly realized.