Strengthening Lebanon's pandemic surveillance through AI
EMERGENT Lebanon harnesses artificial-intelligence tools linked directly to laboratory data to detect early trends in high-risk communicable diseases and to automate the interpretation of potential outbreaks.
Why a laboratory-based early warning system?
Lebanon's economic crisis, which began in 2019 and was aggravated by the COVID-19 pandemic, has weakened its public-health infrastructure. National communicable-disease surveillance relies largely on case notification by health providers and on conventional epidemiological analysis.
This approach can suffer from low sensitivity, limited coverage, data-validity issues and delays between the first cases and their detection — precious time lost when an outbreak is starting.
Hospital laboratories, by contrast, generate objective, coded and timely evidence of infection every day. Linking AI tools directly to these data offers a faster and more complete picture of what is circulating.
Project at a glance
- Full title: Strengthening pandemic surveillance through artificial-intelligence automation of laboratory data for Lebanon
- Network: Lebanese hub of AI4PEP — the Global South AI for Pandemic & Epidemic Preparedness and Response Network
- Phase 1: develop and validate a prototype with a university hospital laboratory, following Ministry of Public Health (MoPH) standards
- Phase 2: roadmap for national scale-up with the MoPH
- Started: June 2023
What we set out to do
Systematise detection
Detect pathogens and conditions under national or regional surveillance by extracting data directly from hospital and laboratory information systems.
Identify trends
Identify significant data patterns using automated rules and alerts validated by subject-matter experts.
Centralise information
Consolidate surveillance information into one real-time dashboard for decision support.
Pilot & scale
Pilot the prototype in a university hospital against the existing process, then build a national roadmap with the MoPH.
Specific objectives
- Identify the key pathogens and conditions to monitor, together with an MoPH expert.
- Define surveillance metrics and interpretation rules with an expert panel.
- Build automated data collection based on interoperability standards (HL7 FHIR, LOINC).
- Develop rules-based and machine-learning algorithms that support expert-defined interpretation rules.
- Deliver a central, real-time dashboard with automated alerts.
- Validate performance using sensitivity, specificity, PPV, NPV, AUROC and AUC-PR.
- Compare pilot data and alerts against the routine MoPH surveillance process.
- Develop a roadmap for national scale-up with the Ministry of Public Health.
- Disseminate results through scientific publications, conferences and capacity building.
Priority diseases and pathogens
EMERGENT focuses on high-risk communicable diseases under national surveillance. The platform currently monitors 18 pathogens using weekly laboratory-confirmed case counts. Priority groups include:
Methods
| Component | How |
|---|---|
| Data | Automated extraction from the hospital laboratory information system; standard coding (HL7 FHIR, LOINC) |
| Analytics | Ensemble of seven anomaly detectors (five statistical, two machine-learning) on weekly counts, with a user-adjustable alert threshold |
| Human oversight | Clinicians label and override flags, enabling continuous performance monitoring |
| Output | Web dashboard and email alerts |
| Evaluation | Sensitivity, specificity, PPV, NPV, AUROC, AUC-PR; comparison with routine MoPH reporting |
Supported by a global network
EMERGENT Lebanon is part of AI4PEP, a global initiative on AI for Global Health funded by Canada's International Development Research Centre (IDRC) and UK International Development.