Early Detection of Epidemics Using Generalized Additive Models and Hidden Markov Models: A Neutrosophic Statistical Approach with Real-Time Health Data

Abstract

Detection of epidemics during their early stages allows both the reduction of public health 
emergencies and the enhancement of resource management. The proposed research methodology 
combines GAMs and HMMs as a neutrosophic statistical framework which detects epidemics in 
real-time. GAMs analyze nonlinear epidemic patterns which result from environmental along with 
mobility conditions while HMMs use probabilistic state transitions to perform classification tasks. 
Bayesian hierarchical models along with spatio-temporal neutrosophic statistical techniques 
increase the framework's capability to respond regionally and deliver geospatial forecasts. Real
world health surveillance data goes through our framework assessment where findings from GAMs 
and HMMs are compared to results obtained from RNNs and Transformer-based AI models. The 
combination of neutrosophic statistical methods with AI techniques leads to better outbreak 
prediction accuracy and generates results which can help interpret and take action in disease 
surveillance.  

DOI: 10.5281/zenodo.16884266