Smart Medical Records: The Next Growth Catalyst in the DRG Era?
Medical records are the original documentation of a patient’s entire diagnosis and treatment process in a hospital, reflecting the quality of care and technical proficiency, and are closely linked to medical safety. Specifically, medical records provide essential data for hospital medical quality control, teaching, scientific research, resolution of medical disputes, clinical decision-making, and the implementation of the Diagnosis-Related Groups (DRGs) payment system, underscoring their undeniable importance.
However, since medical record documentation occurs at the end of the clinical pathway and does not immediately impact the quality of patient care, physicians often become lax in this area when overwhelmed by a high patient volume.
Prior to the era of smart hospitals, the value of information exchange and integration was limited. Even when physicians documented complete and clear medical records for patients, these records were often overlooked in most cases due to patients’ errors in preserving them and the lack of mutual recognition among different hospitals. However, with the large-scale advancement of medical informatization today, the barriers to patient data transmission across different departments and even different hospitals have been broken, and the value of data is now emerging.
As the carrier of patient information, medical records should provide foundational support for informatization reforms. However, as illustrated by the bucket effect, deficiencies in medical record quality are constraining the development of the entire informatization system.
In analyzing the issues within medical records, Ying Jiaoqian, Deputy Director of the Medical Affairs Department at China-Japan Friendship Hospital, attributes quality defects primarily to four factors: first, incomplete completion of the medical record face sheet and improper selection of diagnoses; second, non-standardized use of medical terminology, resulting in inconsistent descriptions for the same disease; third, overly simplistic progress notes that render performed diagnostic and therapeutic procedures untraceable, thereby posing legal risks; and fourth, incomplete medical record information.
To address the aforementioned four issues, it is necessary to both improve physicians’ efficiency and implement post-hoc quality control of medical records. The key lies in focusing on three core aspects: standardization, intelligent automation, and effective quality control.
Standardization refers to the standardization of medical record information, ensuring that text, charts, images, and other content generated during patient care comply with standardized clinical terminology for storage in a standardized database.
For a long period, inconsistencies in the information systems used by various medical specialties and the lack of standardized terminology for data annotation collectively contributed to the widely discussed problem of “information silos.” However, with the advancement of electronic medical record (EMR) grading and health information interoperability assessments, the issue of medical record interoperability has been resolved in secondary hospitals and above. Within medical consortia, primary care institutions’ Hospital Information Systems (HIS), EMRs, and other systems can now be connected to the information systems of higher-level hospitals. It can be said that the promotion of policies related to healthcare informatization has gradually resolved the most significant bottleneck facing the healthcare IT industry.
Building upon standardization, intelligence aims to enhance the efficiency and accuracy of physicians’ electronic medical record (EMR) documentation. Some health IT vendors have integrated “intelligent” capabilities into EMRs by using specific rules to select appropriate templates for physicians and associate entered data in real time. In contrast, next-generation health IT companies such as Dr. Mayson have improved clinical decision support systems by constructing comprehensive knowledge databases and leveraging natural language processing (NLP) as the underlying technology. This approach enables human-like logical reasoning for semantic quality control of medical records, replacing the point-to-point linear logic characteristic of traditional medical record quality control systems.
The difference between the two lies in the accuracy and richness of the recommended information. Due to the broad and complex nature of medical knowledge, a single symptom can often suggest multiple possible causes; therefore, rule-based recommendations frequently fail to cover all possibilities.
In contrast, NLP-based CDSS can process complex medical logic and provide physicians with probability distributions of potential diseases based on patient conditions, while being easily integrated into electronic health records (EHRs). By leveraging AI, this system enhances both the efficiency and quality of clinical diagnosis and treatment, while reducing the reliance on physicians’ individual knowledge base.
The most critical aspect is effective quality control. Typically, reviewing a single medical record consumes approximately 30–40 minutes of a physician’s time in the medical records department. In comparison with international data, coders in the United States and Australia encode 4–5 medical records per day, whereas in China, taking Jilin Province as an example, statistical figures indicate 15 records per day; in practice, however, coders sometimes need to process up to 50 records daily.
Furthermore, the ability to identify errors within medical records featuring a wide variety of error types places high professional demands on coders and quality control personnel. In China, vocational education for coders and quality control staff has historically been dominated by associate degree programs, with undergraduate majors only being introduced in recent years. However, hospitals require coders and quality control personnel who have majored in clinical medicine. In reality, however, professionals trained in clinical medicine are unlikely to be willing to work solely as coders or quality control personnel.
Therefore, the concept of intelligent medical record quality control has long been established. However, rule-based quality control can typically only address simple entry errors, such as missing fields in medical records or male patients being recorded with gynecological diagnoses. Only NLP-based intelligent medical record quality control systems can achieve intelligent real-time and retrospective quality control, perform more complex connotative quality assessments, identify logical errors within medical records, and thereby make large-scale medical record quality control a reality.
As standardization issues gradually fade, Dr. Mayson is targeting two key areas—effective quality control and intelligent medical record management—in an effort to leverage AI to address current challenges in hospital medical records.
Regarding effective quality control, Dr. Mayson has developed an AI healthcare solution—the Medical Record Front Page/Electronic Medical Record Quality Control System. Specifically, this system focuses on three key challenges: management of medical record front pages, connotation-based quality control of electronic medical records, and comprehensive quality control management for all medical records.
In medical record front page management, Dr. Mayson can construct patient profiles through semantic understanding of the entire medical record and supplementation of other information, thereby deducing reasonable principal diagnoses and providing necessary alerts to physicians and coders. Meanwhile, it can also identify and flag issues such as missing comorbidities and omitted surgeries or procedures based on patient profiles, enhancing the connotative quality and data quality of the medical record front page.
Regarding quality control of the substantive content of medical records, Dr. Mayson has shifted the quality control checkpoint upstream to cover all textual content of hospital medical records. This approach transforms pure formal outcome-based quality control into a combination of process-oriented substantive quality control and formal quality control. By implementing real-time intervention for defective content and prompting senior physicians to promptly review, reject, or request modifications, it improves the accuracy of medical record content, clarity of diagnoses, and standardization of treatments from the source, thereby achieving comprehensive quality control and evaluation of clinical medical records.
For comprehensive medical record quality control management, the AI-based medical record front page/medical record quality control system, integrated into the hospital information system, can analyze all medical records and achieve full-process, dynamic, and orderly quality control management of various aspects of medical behavior through an intelligent rule engine. This effectively addresses the difficulties in real-time quality monitoring prevalent in the manual era. By statistically analyzing real-time data, it enables timely understanding of quality control status for medical processes—particularly for key disease categories—and the quality of medical record documentation, thereby providing data support for medical record quality improvement and performance evaluation.
“With AI support, the time required to review a medical record can be reduced to less than 30% of the original duration. Leveraging an AI-powered semantic quality control engine, rapid semantic quality audits can be performed on specialized medical records across various disciplines, such as oncology chemotherapy, obstetrics, and common cardiovascular and cerebrovascular diseases,” Jiang Songyi, Chief Medical Quality Officer at Dr. Mayson, told VCBeat. “Furthermore, the medical records department can analyze global data to identify patterns of quality control defects at the hospital, departmental, and even individual physician levels, thereby significantly enhancing the precision of quality improvement measures implemented by management.”
If medical record quality control raises the “baseline” of healthcare quality management, then Huimei Technology’s integration of medical record quality control with CDSS elevates its “ceiling.”
“Without the digitalization of information to handle the legwork, our staff in the Medical Affairs Department would be overwhelmed and unable to effectively support the development of both clinical departments and the hospital,” said Ying Jiaoqian. By applying a Clinical Decision Support System (CDSS) for medical record quality control, the hospital has achieved automated, real-time, and efficient medical record management, while also controlling the quality of clinical diagnosis and treatment at the data source.
On one hand, Clinical Decision Support Systems (CDSS) can restructure medical record management processes, enabling seamless real-time interaction between clinicians and medical record staff. On the other hand, CDSS facilitates high-efficiency “quality management” by strengthening quality control through integration with all hospital business systems, data processing, and clinical alerts, thereby enhancing clinical trust and adoption. CDSS executes quality control pathways based on predefined data quality control rules. These rules are established in accordance with clinical guidelines and standards, national quality assessment requirements, and the actual clinical needs of hospitals.
Taking the discharge diagnosis on the medical record face sheet as an example, the determination of the principal diagnosis follows the “three mosts” principle, namely, the disease diagnosis that poses the greatest threat to the patient’s health, consumes the most medical resources, and results in the longest length of stay during hospitalization. For instance, in obstetrics, the principal diagnosis refers to the major obstetric complications or associated conditions; accordingly, the system has designed a quality control rule stating that “the principal obstetric diagnosis should be selected from obstetric complications or associated conditions.”
After three months of applying a Clinical Decision Support System (CDSS) for medical record quality control, the overall adoption rate of system alerts by hospital clinicians showed an upward trend, while the total number of defects in medical records and the average number of defects per record in clinical departments gradually decreased. “The benefits are twofold: first, clinicians’ practices become more standardized; second, medical record management becomes more refined and process-oriented,” stated Ying Jiaoqian. The hospital-wide unified quality control standards not only promote homogenization of internal quality control efforts, but also enable administrators to monitor the status of medical record management in real time through multi-dimensional quality control data aggregation provided by the system.
In addition to improving efficiency and quality for clinicians and medical records departments, the promotion of the DRG payment system has endowed medical records with new value. In the practice of DRG, many hospitals have experienced undercoding due to missing content in medical records, resulting in sustained financial losses. From this perspective, DRG provides a driving force for hospitals to reconstruct medical record quality control.
Based on current NLP implementations, the technology can automatically map ICD codes at the time of patient grouping; automatically identify the principal diagnosis in cases of multiple diagnoses; intelligently merge codes and optimize cost rationality; and perform post-grouping data audits and coding accuracy verification. It also assists hospitals in internal performance evaluation, insurance payment and settlement, and regulatory audits for insurance compliance violations.
For Dr. Mayson, its expertise in NLP enables a rapid transition from medical record quality control to hospital DRG management.
Beyond Dr. Mayson, numerous CDSS companies have now begun to enter the DRG market through medical record quality control. Transformation is often accompanied by challenges and opportunities; what strategy Dr. Mayson will adopt in this market remains to be seen.