Quan Yu from Shengjing Hospital: Hospital Big Data Applications Can Go Further

Recently, at the 2018 HIMSS Greater China Annual Conference, Quan Yu, Director of the Information Center at Shengjing Hospital of China Medical University, delivered a speech on the practical implementation and application of hospital data. In Director Quan’s view, hospital data can be leveraged for multiple applications, including medical record quality control, rational drug use, physician performance evaluation, and standardization of clinical pathways. The following are the key points compiled by VCBeat (WeChat official account: vcbeat):

Quan Yu, Director of the Information Center, Shengjing Hospital of China Medical University

Electronic medical records have been implemented in China for nearly a decade; only with paperless systems can big data move forward.

I recall that a few years ago, we often discussed on WeChat that paperless workflows seemed far out of reach for us; given the complexity of hospital processes, implementation appeared unlikely. Yet unexpectedly, Shengjing Hospital has now achieved paperless operations in over 95% of its processes.

Hospitals leverage big data to further enhance medical safety and improve the quality of care. I have often heard, on many occasions, that big data should provide a foundation for scientific research. However, if the data we obtain are inaccurate, the research findings will inevitably be flawed as well.

Big data has already been established in hospitals, yet some people are still comparing data volumes using carrier data, which is meaningless. What we truly need to compare is full-sample data.

Most hospital data remains idle and is not effectively utilized. True big data involves leveraging data analytics to identify issues, rather than merely conducting retrospective validation against a predefined objective after the fact. While that constitutes an application of big data, I believe we can go further.

Why Promote Shengjing Hospital’s Case? Because All Our Data Are Derived from Clinical Practice, Leveraging the Hospital’s Own Data to Support Its Big Data Applications.

Specifically, we leverage big data to support healthcare management and improve the quality of medical records (particularly subjective medical records), identify deficiencies in healthcare quality based on existing data, and thereby implement targeted improvements.

Currently, hospital medical record quality control is largely based on objective indicators, with little attention paid to the actual quality of the medical records. For the handwritten portions in subjective medical records, judgments are made by humans. How can computers leverage big data to analyze and utilize medical records, thereby improving healthcare quality?

Subjective Medical Record Quality Control: Comparing Textual Similarity Between Two Records. Shengjing Hospital has integrated algorithms into its medical record system to compare two ward round notes or two progress notes from the same patient, assessing whether their degree of overlap is identical. The results frequently reveal that the two records are completely identical, indicating that the quality of hospital medical records is far from what we might imagine.

Regarding chief complaints and diagnoses, assume a child is diagnosed with lobular pneumonia. If fever and cough are not mentioned in the chief complaint, the diagnosis is considered inconsistent with the chief complaint. Why? We extracted records of single-disease diagnoses spanning more than five years, separated them using IT techniques, and compiled high-frequency terms. This analysis revealed that fever and cough appeared with high frequency.

If fever and cough are not mentioned in the chief complaint, it raises suspicion of an issue. By having physicians manually review these medical records with subjective judgment, efficiency can be improved. We do not need to achieve 100% completion; reaching just 50% effectiveness would reduce labor costs by half.

The same applies to the development of clinical pathways. Around 2013, hospitals typically convened expert panels for deliberation, wherein leading specialists determined the appropriate diagnostic and treatment protocols for specific diseases. However, due to variations in expertise among experts, as well as differences in diagnostic capabilities and equipment across hospitals, the resulting clinical pathways varied from one institution to another.

How Can Hospitals Standardize Their Clinical Pathways and Continuously Optimize Them?

In response, Shengjing Hospital has leveraged big data analytics to identify single-diagnosis pathways for diseases with a duration of over five years. By determining the most prevalent diagnostic and treatment practices, the hospital has restructured these clinical pathways. Applying historical data to inform current practice and guide future care represents a significant advancement.

The effects are also evident. Three or four years ago, when a child was hospitalized, antibiotics were the first-line treatment on the first day. This is no longer the case; at Shengjing Hospital, the initial preference may now be medications aimed at boosting immunity. Antibiotics are typically not administered until the second or third day. This shift reflects changes in clinical diagnosis and treatment driven by big data.

Furthermore, big data can also be used to evaluate a physician's level of competence.

Typically, evaluating a physician’s competence requires examining the degree of concordance among admission diagnoses, discharge diagnoses, and confirmed diagnoses (including pathological diagnoses). Full concordance may indicate either that the disease case is relatively straightforward or that the physician possesses a high level of expertise. Conversely, significant discrepancies may suggest either limited physician proficiency or greater disease complexity. By extracting and providing this information to hospital administrators for analysis, it is possible to assess a physician’s capabilities in a rational and effective manner.

Medical Quality Analysis Can Serve Three Key Purposes:

1. Standardize the prescribing practices of clinicians;

2. Improve the level of rational drug use;

3. Provide data support to standardize medical practices.

"Analyzing the variety of medications through data can reflect the complexity of disease diagnosis."

How to Evaluate the Rationality of a Physician’s Medication Practices? For instance, one physician may treat Patient A, while another treats 10 additional patients; however, a higher patient volume does not necessarily indicate a greater workload. This is because physicians seeing fewer patients may be managing cases with higher clinical complexity, necessitating a comprehensive assessment. By extracting medication regimens to analyze the quality of prescribing, observing trends over time, and categorizing the types of antibiotics used annually, data can be provided to administrators to determine trends in antibiotic utilization.

Following the implementation of zero-markup pricing for pharmaceuticals, hospitals should place greater emphasis on both the volume and quality of medication use.

In terms of optimizing processes and rationally allocating medical resources, we conducted an analysis of outpatient efficiency and reconfigured service resources based on different workflows.

Among these, intelligent triage is also implemented based on big data. We analyzed the chief complaints of outpatient visits over a five-year period to automatically recommend the appropriate department for registration. The more keywords patients select, the more precise the recommended department becomes. Of course, due to the massive volume of data, real-time computation by the system would result in prolonged waiting times; therefore, Shengjing Hospital employs certain algorithms to address this issue.

After registration and consultation, the system will provide patients with follow-up visit reminders and corresponding health education.

Whether pursuing hospital information technology or healthcare management, it is essential to base efforts on process management optimization, leveraging historical data analysis to identify and address issues. Strengthening medical safety and quality control remains an enduring priority for hospitals.

Quantifying workload and performance evaluation is an anticipated development for doctors and nurses. Hospitals must fully mobilize the initiative of medical staff and ensure reasonable performance-based compensation. In this regard, since last December, Shengjing Hospital has been leveraging big data to assess the workload and performance of doctors and nurses, covering professionals such as technicians, pharmacists, anesthesiologists, and operating room nurses.

In summary, whether it involves healthcare management, information technology infrastructure development, or the utilization of big data, all represent ongoing processes of continuous improvement.