Tri-Service General Hospital Leverages NetPro Trinity ETL Solution to Accelerate Medical AI and Big Data Applications
Source: iThome
Written by: Vendor Press Release | Published: 2024-08-12
Written by: Vendor Press Release | Published: 2024-08-12
Figure 1. Dr. Chen Wei-Liang(left), Director of the Medical Informatics Department at TSGH, and Dr. Hsu Chia-Jung, Deputy Director of Medical Informatics and Chief Data Officer at TSGH.
Founded in 1946, Tri-Service General Hospital (hereafter referred to as “TSGH”) has journeyed through 78 long years, steadfastly upholding the principle of “Wholehearted TSGH, Holistic Care,” and dedicated to safeguarding the health of military personnel and civilians. Recently, in a global hospital ranking conducted by the internationally renowned media outlet Newsweek in collaboration with global data company Statista, TSGH was awarded four stars, and achieved five-star recognition in subcategories such as “IT and Medical Technology” and “Medical Care Services”—a first among Taiwanese hospitals, clearly showcasing the innovative service capabilities of the military hospital.
Dr. Chen Wei-Liang, Director of the Medical Informatics Department at TSGH, stated that TSGH has always been relentless in improving medical care, patient experience, and safety. Therefore, the hospital is heavily committed to developing smart healthcare and is eager to establish a foundation for big data applications in healthcare. To this end, TSGH officially established its Data Center in April 2022, with its primary mission being the construction of a big data marketplace.
He further explained that the data marketplace not only widely supports medical research projects and presents clinical statistical data, but also combines data visualization to meet the requirements of medical center accreditation. From this we can see that behind each data marketplace lies the different needs of various departments, corresponding to different project goals and data collection formats. If traditional methods are used, they rely on labor-intensive manual coding, time-consuming data queries, and lengthy data migration. This not only causes long waits for requesting departments but may also result in missing the golden opportunity for innovative medical service development projects.
Therefore, TSGH decided to introduce an ETL (Extract, Transform, Load) solution, aiming to significantly shorten the time required to build each data marketplace from scratch. After a rigorous vendor selection process, NetPro’s Trinity ETL solution was chosen and launched concurrently with the Data Center’s inauguration in April 2022.
Dr. Chen Wei-Liang, Director of the Medical Informatics Department at TSGH, stated that TSGH has always been relentless in improving medical care, patient experience, and safety. Therefore, the hospital is heavily committed to developing smart healthcare and is eager to establish a foundation for big data applications in healthcare. To this end, TSGH officially established its Data Center in April 2022, with its primary mission being the construction of a big data marketplace.
He further explained that the data marketplace not only widely supports medical research projects and presents clinical statistical data, but also combines data visualization to meet the requirements of medical center accreditation. From this we can see that behind each data marketplace lies the different needs of various departments, corresponding to different project goals and data collection formats. If traditional methods are used, they rely on labor-intensive manual coding, time-consuming data queries, and lengthy data migration. This not only causes long waits for requesting departments but may also result in missing the golden opportunity for innovative medical service development projects.
Therefore, TSGH decided to introduce an ETL (Extract, Transform, Load) solution, aiming to significantly shorten the time required to build each data marketplace from scratch. After a rigorous vendor selection process, NetPro’s Trinity ETL solution was chosen and launched concurrently with the Data Center’s inauguration in April 2022.
Replacing Manual Coding with Automation to Accelerate Healthcare Big Data Development
Dr. Hsu Chia-Jung, Deputy Director of Medical Informatics and Chief Data Officer at TSGH, noted that there were many driving factors behind the decision to introduce ETL. The clearest one originated from a hospital evaluation that required the use of data visualization software to present statistical results. Supporting data visualization depends on a functioning data marketplace.
Another driving factor was that TSGH has periodically undergone smart hospital evaluations or participated in smart healthcare competitions that involved AI and big data applications—requiring support from the data marketplace. However, TSGH had long relied on manual coding for one-off data access and subsequent migration to user-designated locations, which was excessively time-consuming and ineffective for constructing data marketplaces at scale.
“In the past, when a department raised a request, the Informatics Department would assign someone to develop the required program, which took 2–3 weeks,” said Hsu. Furthermore, if a project involved migrating 100,000 records, it would take an additional 1–2 days. After switching to ETL, as long as the syntax is configured, the scheduler can extract, transform, and load data into the designated database—completely obviating the need for manual programming. The configuration, scheduling, and testing phase now takes up to one day, and the actual migration of 100,000 records completes in about 10 minutes—a dramatic efficiency gain. This explains why TSGH opted to replace manual coding with ETL as the foundational method for building data marketplaces.
Another driving factor was that TSGH has periodically undergone smart hospital evaluations or participated in smart healthcare competitions that involved AI and big data applications—requiring support from the data marketplace. However, TSGH had long relied on manual coding for one-off data access and subsequent migration to user-designated locations, which was excessively time-consuming and ineffective for constructing data marketplaces at scale.
“In the past, when a department raised a request, the Informatics Department would assign someone to develop the required program, which took 2–3 weeks,” said Hsu. Furthermore, if a project involved migrating 100,000 records, it would take an additional 1–2 days. After switching to ETL, as long as the syntax is configured, the scheduler can extract, transform, and load data into the designated database—completely obviating the need for manual programming. The configuration, scheduling, and testing phase now takes up to one day, and the actual migration of 100,000 records completes in about 10 minutes—a dramatic efficiency gain. This explains why TSGH opted to replace manual coding with ETL as the foundational method for building data marketplaces.
Figure 2. Dr. Hsu Chia-Jung, Deputy Director of Medical Informatics and Chief Data Officer at TSGH
Establishing a Data Center to Support AIoT Research Projects
On the topic of smart healthcare, Hsu added that several years ago TSGH founded the AI and IoT Development Center (AIoT), which frequently initiates research projects necessitating specific data for AI/ML analysis.
Before the Data Center was established, there was no dedicated support unit for the AIoT Center, and the burden fell on the Informatics Department. Due to limited manpower, they even had to hire external part-time developers. With data integration from varied and numerous sources, complexity was high, and developing a single program took an average of 2–3 months.
Realizing the wait was too long, the hospital decided to establish a Data Center and planned to use automated tools for data collection, transformation, and loading—reducing implementation timelines from 2–3 months to 2–3 weeks. They also aimed to cultivate medical data analysis talent, planning to shift future IT roles from programmers to data analysts.
At the same time, the IT director attended a smart healthcare symposium at Taipei Veterans General Hospital and learned about NetPro’s Trinity ETL in use there. Recognizing the benefits, the director tasked the Informatics Department with quickly evaluating Trinity, believing it could greatly support TSGH’s medical big data development.
The team actively evaluated multiple ETL solutions—both domestic and international. After comprehensive consideration across several criteria, Trinity emerged as the top choice.
Before the Data Center was established, there was no dedicated support unit for the AIoT Center, and the burden fell on the Informatics Department. Due to limited manpower, they even had to hire external part-time developers. With data integration from varied and numerous sources, complexity was high, and developing a single program took an average of 2–3 months.
Realizing the wait was too long, the hospital decided to establish a Data Center and planned to use automated tools for data collection, transformation, and loading—reducing implementation timelines from 2–3 months to 2–3 weeks. They also aimed to cultivate medical data analysis talent, planning to shift future IT roles from programmers to data analysts.
At the same time, the IT director attended a smart healthcare symposium at Taipei Veterans General Hospital and learned about NetPro’s Trinity ETL in use there. Recognizing the benefits, the director tasked the Informatics Department with quickly evaluating Trinity, believing it could greatly support TSGH’s medical big data development.
The team actively evaluated multiple ETL solutions—both domestic and international. After comprehensive consideration across several criteria, Trinity emerged as the top choice.
Original Vendor Rapidly on Site to Resolve Issues
Hsu explained that during the selection process, TSGH considered several factors to determine the suitability of each ETL solution. First, as a public and medical institution, TSGH places a high priority on sensitive data security and is cautious about cloud-based SaaS solutions. Budgeting also favored perpetual licensing over subscriptions; Trinity’s buyout licensing model aligned well with TSGH’s procurement policies.
Furthermore, functionality evaluations focused particularly on a scheduler that is user-friendly, intuitive, and aligned with Taiwanese engineers’ logic—where Trinity stood out among competitors.
The most critical factor was service responsiveness. NetPro’s office, located in Neihu near TSGH, meant their staff could respond quickly on-site. By contrast, for foreign vendors, the hospital would have to open a support ticket and wait 1–2 days for replies—time the hospital could not afford. Operating on a 24/7 basis, even a minor delay in solving issues could delay critical patient care. As soon as a problem arises, a quick fix is essential.
NetPro Trinity, meeting all these criteria, received unanimous support from the Informatics team and became the final choice—a decision vindicated over time. For example, during a network outage in July, many ETL jobs remained pending despite restarting servers. NetPro engineers arrived promptly and resolved the issue, minimizing impact—demonstrating the critical value of on-site support.
Furthermore, functionality evaluations focused particularly on a scheduler that is user-friendly, intuitive, and aligned with Taiwanese engineers’ logic—where Trinity stood out among competitors.
The most critical factor was service responsiveness. NetPro’s office, located in Neihu near TSGH, meant their staff could respond quickly on-site. By contrast, for foreign vendors, the hospital would have to open a support ticket and wait 1–2 days for replies—time the hospital could not afford. Operating on a 24/7 basis, even a minor delay in solving issues could delay critical patient care. As soon as a problem arises, a quick fix is essential.
NetPro Trinity, meeting all these criteria, received unanimous support from the Informatics team and became the final choice—a decision vindicated over time. For example, during a network outage in July, many ETL jobs remained pending despite restarting servers. NetPro engineers arrived promptly and resolved the issue, minimizing impact—demonstrating the critical value of on-site support.
Planning to Purchase Stream Data Processing Module to Meet Research Needs
Additionally, because TSGH hosts many heterogeneous databases with inconsistent encoding and formats, team members initially struggled with transformation settings. After NetPro’s careful guidance, they successfully mastered best practices—enabling seamless extraction and transformation in one step, regardless of data volume.
After deploying Trinity, TSGH also discovered unexpected value-added benefits. NetPro proactively updates clients with new features. They also listen attentively to customer needs; common requests are implemented in later releases, benefiting all users—including TSGH.
TSGH’s Trinity implementation has incrementally expanded. Starting with the standalone Trinity ETL edition, they upgraded to the enterprise edition with HA architecture. They have since deployed modules such as JCS for job scheduling, UDM for web crawling and document parsing (PDF, Word, Excel), and Metaman for enterprise data governance (metadata management, data catalog, data lineage, tagging, and other key features).
Looking ahead, TSGH plans to procure the SDM Stream Data Processing Platform. As implied, this module handles streaming data—unlike ETL’s batch processes. Their scenario: when the main system is busy, periodic data extracts every 10 minutes were used for analysis. As analysis demands grew, users found the 10-minute lag unacceptable. Deploying Trinity SDM allows event-triggered data capture and immediate delivery to designated destinations—avoiding system interference and eliminating the previous latency.
After deploying Trinity, TSGH also discovered unexpected value-added benefits. NetPro proactively updates clients with new features. They also listen attentively to customer needs; common requests are implemented in later releases, benefiting all users—including TSGH.
TSGH’s Trinity implementation has incrementally expanded. Starting with the standalone Trinity ETL edition, they upgraded to the enterprise edition with HA architecture. They have since deployed modules such as JCS for job scheduling, UDM for web crawling and document parsing (PDF, Word, Excel), and Metaman for enterprise data governance (metadata management, data catalog, data lineage, tagging, and other key features).
Looking ahead, TSGH plans to procure the SDM Stream Data Processing Platform. As implied, this module handles streaming data—unlike ETL’s batch processes. Their scenario: when the main system is busy, periodic data extracts every 10 minutes were used for analysis. As analysis demands grew, users found the 10-minute lag unacceptable. Deploying Trinity SDM allows event-triggered data capture and immediate delivery to designated destinations—avoiding system interference and eliminating the previous latency.
Continuing Reliance on ETL for HIS Upgrade and Other Major Projects
TSGH also used Trinity to support pandemic contact tracing. If a hospitalized patient tested positive, Trinity could instantly extract all contacts in the past two months—including nurses, caretakers, examiners, consulting doctors, and adjacent patients—for rapid identification and tracking.
They also used Power BI to consolidate data on COVID-19 hospitalizations—tracking numbers, age, gender, admission/discharge times. Trinity accelerated data aggregation, helping management identify trends and formulate response strategies.
Now that the pandemic has eased, Trinity returned to medical big data project use. Over the next 1–2 years, TSGH plans to leverage it for their HIS core system transformation—from Client-Server to Client-Service-Server architecture with Web Service layers—to enhance scalability and front-end access. The outpatient system is already online this year, with the inpatient system expected next year.
Once the new HIS stabilizes, TSGH plans to consolidate outpatient and inpatient systems into a single version for rollout to 13 military hospitals. To support this, the Medical Affairs Bureau plans to establish northern and southern data centers where HIS data from each hospital will be backed up and synchronized.
These data centers will function as large data warehouses, convertible into modular data marts to accelerate smart healthcare across military hospitals. TSGH, responsible for maintaining the northern data center, will assist the 13 military hospitals in cleaning their data marts, again relying on ETL—and extending Trinity’s reach beyond their own campus.
Trinity’s role doesn’t end there. With telemedicine becoming the new norm and video data expanding medical records exponentially, storage and retrieval burdens grow—necessitating ETL to manage cold/hot data separation. Additionally, the hospital’s biobank will continue to expand, storing samples alongside related data and reports. ETL will play a central role in gathering these data, further enhancing Trinity’s scope of application.
They also used Power BI to consolidate data on COVID-19 hospitalizations—tracking numbers, age, gender, admission/discharge times. Trinity accelerated data aggregation, helping management identify trends and formulate response strategies.
Now that the pandemic has eased, Trinity returned to medical big data project use. Over the next 1–2 years, TSGH plans to leverage it for their HIS core system transformation—from Client-Server to Client-Service-Server architecture with Web Service layers—to enhance scalability and front-end access. The outpatient system is already online this year, with the inpatient system expected next year.
Once the new HIS stabilizes, TSGH plans to consolidate outpatient and inpatient systems into a single version for rollout to 13 military hospitals. To support this, the Medical Affairs Bureau plans to establish northern and southern data centers where HIS data from each hospital will be backed up and synchronized.
These data centers will function as large data warehouses, convertible into modular data marts to accelerate smart healthcare across military hospitals. TSGH, responsible for maintaining the northern data center, will assist the 13 military hospitals in cleaning their data marts, again relying on ETL—and extending Trinity’s reach beyond their own campus.
Trinity’s role doesn’t end there. With telemedicine becoming the new norm and video data expanding medical records exponentially, storage and retrieval burdens grow—necessitating ETL to manage cold/hot data separation. Additionally, the hospital’s biobank will continue to expand, storing samples alongside related data and reports. ETL will play a central role in gathering these data, further enhancing Trinity’s scope of application.
