Epidemic ILI forecasting in Prishtina using surveillance data for public health policy and response

Background

Influenza-like illness (ILI), is the cornerstone of seasonal respiratory disease surveillance. We used routine ILI data to describe epidemic dynamics and generate short-term projections in Prishtina, to support evidence-based policymaking.

Methods

We used the EpiNow2 Bayesian framework, to estimate the epidemic trends and produced short-term projections of ILI incidence using surveillance data of Prishtina region. We estimated the time-varying reproduction number by reconstructing unobserved infections and linking them to reported cases via delay distributions (1.4±0.6 days) and reporting delay (2.1±1.5 days). Then, we performed a joinpoint analysis to identify turning points in weekly ILI trends and determine the epidemiological weeks at which significant changes in case patterns occurred. Also, we also estimated basic reproduction number (R₀) that was used as a key indicator to guide public health interventions and policymaking decisions.

Results

Over 30 epidemiological weeks, 13,343 ILI cases were reported (mean 642.6/week;median 469.5;CV=0.6). ILI activity increased rapidly from week 48/2025, peaking at 1,410 cases in week 51/2025, corresponding to an early R₀=1.6–1.8. Joinpoint analysis identified a significant turning point at week 15/2026, when cases declined to 201, representing an 85.7% reduction from peak levels. Prishtinë municipality consistently contributed the largest proportion of cases. A temporary zero count in week 10/2026 likely reflected reporting interruption rather than absence of transmission. Projections indicated a continued downward trend in the weeks following the last observation.

Conclusions

EpiNow2 can be used pragmatically with weekly ILI surveillance data to support short-term situational awareness and service preparedness. This framework was still useful for identifying epidemic peak timing, monitoring decline, and informing public health response, even with aggregated data.

Key messages

• Even though Influenza has a typical seasonal epidemic pattern, early detection and rapid response during the growth phase remain essential.

• Advanced tools for projections, such as EpiNow2, enables actionable real-time surveillance and timely actions.