How Thailand's largest miliatray Hospital connected live infection data, streamlined infection workflows and began using predictive risk to support more targeted surveillance.
Ashleigh Myall
At a Glance
Phramongkutklao Hospital worked with NEX to replace fragmented, retrospective infection tracking with a connected hospital-wide surveillance system.
Within three weeks, the Infection Prevention and Control team had a centralised view of microbiology records, active cases and infection risks across the hospital.
A subsequent six-month evaluation also demonstrated the potential for earlier, more targeted surveillance of carbapenem-resistant Enterobacterales, with an AUROC of 0.89 and 87% sensitivity over a seven-day prediction window.
The Challenge
Phramongkutklao Hospital is a 1,200-bed Royal Thai Army teaching hospital in Bangkok, providing specialist and tertiary care to military personnel and the wider public.
Its Infection Prevention and Control team manages a complex hospital environment with a high burden of healthcare-associated infection and antimicrobial resistance.
Before NEX, much of the hospital’s surveillance work was manual. Teams reviewed microbiology results retrospectively, maintained infection records across spreadsheets and brought together patient and ward information from different systems.
This made it difficult to maintain a current hospital-wide view of infection risk. It also placed a considerable administrative burden on the IPC team and limited the time available for investigation and prevention.
The hospital wanted to move from fragmented tracking towards a more timely and connected approach.
Connecting Infection Data Across the Hospital
In January 2025, Phramongkutklao Hospital partnered with NEX to implement Infection Intelligence across the hospital.
The initial deployment connected live microbiology results with relevant patient and hospital data. NEX automatically processed and organised new records for IPC review, including:
Multidrug-resistant organisms.
Bloodstream infections.
Device-associated infections.
Patient locations and ward movements.
Infection and colonisation status.
Screening and follow-up activity.
This gave the IPC team one place to review active cases, track changes in patient status and understand how infection risks were distributed across wards.
The platform was configured around the hospital’s existing workflows and local definitions rather than introducing a separate surveillance process.
What Changed for the Infection Control Team
A Centralised View Within Three Weeks. Within three weeks, the hospital had a searchable digital record of priority infections organised by patient, organism, location and time. Instead of moving between spreadsheets and separate data sources, the team could review relevant information through a shared hospital-wide view.
Earlier Situational Awareness. Live feeds showed active cases and infection status by ward. Locally defined thresholds helped the team recognise where infection signals were increasing and closer investigation might be needed.
Less Manual Administration. Automated processing reduced repetitive data entry and the need to maintain parallel infection-tracking spreadsheets. This allowed IPC staff to spend more time reviewing risks, following up cases and coordinating prevention measures.
Adoption by the Hospital Team. The platform was introduced directly into the team’s routine workflows, with users onboarded across infection surveillance and management activities. Feedback helped NEX adapt the interface, terminology and workflows to the hospital’s local requirements.
"We finally have real-time visibility across the entire hospital. It has dramatically reduced time-consuming tasks and is already helping us prioritise more effectively and prepare faster."
Dr. Vasin Vasikasin
|
Assistant Professor and Consultant in Infectious Diseases
Honorary Research Fellow at Imperial College London
|
Phramongkutklao Hospital
Moving From Surveillance to Earlier Prediction
The next stage focused on carbapenem-resistant Enterobacterales.
CRE can cause difficult-to-treat infections and spread between patients and healthcare facilities. Patients may carry CRE without symptoms, allowing colonisation to remain undetected and contribute to further transmission.
Screening can identify colonised patients so that appropriate precautions and other infection-control measures can be considered. However, screening every patient is expensive and may not be practical.
NEX explored whether hospital data could help the team identify which patients were most likely to acquire or subsequently test positive for CRE.
Early Predictive Performance
The model was evaluated over six months of prospectively generated hospital data, including both clinical and screening cultures.
Early results showed:
0.89 AUROC
The model demonstrated strong discrimination between patients who did and did not subsequently test positive for CRE.
Up to Seven Days Earlier
Risk was assessed over a seven-day prediction window, creating an opportunity for review before a positive microbiology result was reported.
87% Sensitivity
At the selected operating threshold, the model identified 87% of patients who subsequently tested positive.
These findings support the feasibility of predictive CRE surveillance within a live hospital environment. They do not, by themselves, demonstrate a reduction in infections. Continued prospective evaluation is needed to assess how predictions affect screening decisions, staff workload, transmission and patient outcomes.
The approach builds on NEX’s earlier peer-reviewed research in The Lancet Digital Health, which demonstrated that dynamic patient-contact networks and routinely collected hospital data could predict hospital-onset infection risk.
From More Screening to Smarter Screening
The purpose of predictive surveillance is not to generate more alerts or automatically place more patients under precautions.
It is to help the IPC team make better-informed choices about where limited screening and prevention resources should be directed.
By combining patient-level risk with hospital-wide situational awareness, NEX can support teams to:
Identify high-risk patients earlier.
Prioritise screening more precisely.
Investigate shared exposures faster.
Reduce low-value manual review.
Focus contact precautions where the potential benefit is greatest.
Track whether required screening and follow-up actions have taken place.
This creates a practical route from passive surveillance towards earlier, more targeted prevention.
What We Learned
The deployment demonstrated that Infection Intelligence could be integrated into a large, complex hospital and adopted within routine IPC workflows.
It also reinforced several lessons:
Implementation must begin with the workflow. Connecting data only creates value when the resulting information supports a clear IPC action.
Local configuration matters. Organisms, definitions, thresholds and surveillance priorities vary between hospitals.
Prediction must remain clinically governed. Risk estimates should support professional review, not replace it.
Operational impact matters alongside model performance. Time saved, screening efficiency, compliance and earlier intervention are as important as statistical accuracy.
Next Steps
NEX and Phramongkutklao Hospital are continuing to evaluate how Infection Intelligence affects day-to-day practice.
The next phase will examine:
The number of surveillance tests required.
How early high-risk acquisitions can be identified.
Time saved during case review and investigation.
Screening and infection-control compliance.
Possible changes in ward exposure and secondary transmission.
Performance across additional multidrug-resistant organisms.
The aim is to establish not only whether infection risk can be predicted, but whether earlier intelligence helps the hospital prevent avoidable infections.
